Merge branch 'drm-next-4.20' of git://people.freedesktop.org/~agd5f/linux into drm-next

More new features and fixes for 4.20:
- Add dynamic powergating support for VCN on picasso
- Scheduler cleanup
- Vega20 support for KFD
- DC cleanups and bug fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180927184348.2696-1-alexander.deucher@amd.com

118 files changed, 1856 insertions, 717 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 6cb35e3dab30..c05b39438663 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -146,6 +146,7 @@ extern int amdgpu_cik_support;
 #define AMDGPU_DEFAULT_GTT_SIZE_MB		3072ULL /* 3GB by default */
 #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
 #define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
+#define AMDGPU_FENCE_JIFFIES_TIMEOUT		(HZ / 2)
 /* AMDGPU_IB_POOL_SIZE must be a power of 2 */
 #define AMDGPU_IB_POOL_SIZE			16
 #define AMDGPU_DEBUGFS_MAX_COMPONENTS		32
@@ -408,16 +409,25 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
 	AMDGPU_DOORBELL64_GFX_RING0               = 0x8b,
 
 	/*
-	 * Other graphics doorbells can be allocated here: from 0x8c to 0xef
+	 * Other graphics doorbells can be allocated here: from 0x8c to 0xdf
 	 * Graphics voltage island aperture 1
-	 * default non-graphics QWORD index is 0xF0 - 0xFF inclusive
+	 * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
 	 */
 
-	/* sDMA engines */
-	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xF0,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
-	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xF2,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
+	/* sDMA engines  reserved from 0xe0 -oxef  */
+	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xE0,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xE1,
+	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xE8,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xE9,
+
+	/* For vega10 sriov, the sdma doorbell must be fixed as follow
+	 * to keep the same setting with host driver, or it will
+	 * happen conflicts
+	 */
+	AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0            = 0xF0,
+	AMDGPU_VEGA10_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
+	AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1            = 0xF2,
+	AMDGPU_VEGA10_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
 
 	/* Interrupt handler */
 	AMDGPU_DOORBELL64_IH                      = 0xF4,  /* For legacy interrupt ring buffer */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 0f9947edb12a..c31a8849e9f8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -76,6 +76,7 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
 		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
 		break;
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		kfd2kgd = amdgpu_amdkfd_gfx_9_0_get_functions();
 		break;
@@ -123,7 +124,7 @@ static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
 
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 {
-	int i;
+	int i, n;
 	int last_valid_bit;
 	if (adev->kfd) {
 		struct kgd2kfd_shared_resources gpu_resources = {
@@ -162,7 +163,15 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 				&gpu_resources.doorbell_physical_address,
 				&gpu_resources.doorbell_aperture_size,
 				&gpu_resources.doorbell_start_offset);
-		if (adev->asic_type >= CHIP_VEGA10) {
+
+		if (adev->asic_type < CHIP_VEGA10) {
+			kgd2kfd->device_init(adev->kfd, &gpu_resources);
+			return;
+		}
+
+		n = (adev->asic_type < CHIP_VEGA20) ? 2 : 8;
+
+		for (i = 0; i < n; i += 2) {
 			/* On SOC15 the BIF is involved in routing
 			 * doorbells using the low 12 bits of the
 			 * address. Communicate the assignments to
@@ -170,20 +179,31 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 			 * process in case of 64-bit doorbells so we
 			 * can use each doorbell assignment twice.
 			 */
-			gpu_resources.sdma_doorbell[0][0] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE0;
-			gpu_resources.sdma_doorbell[0][1] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200;
-			gpu_resources.sdma_doorbell[1][0] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE1;
-			gpu_resources.sdma_doorbell[1][1] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200;
-			/* Doorbells 0x0f0-0ff and 0x2f0-2ff are reserved for
-			 * SDMA, IH and VCN. So don't use them for the CP.
-			 */
-			gpu_resources.reserved_doorbell_mask = 0x1f0;
-			gpu_resources.reserved_doorbell_val  = 0x0f0;
+			if (adev->asic_type == CHIP_VEGA10) {
+				gpu_resources.sdma_doorbell[0][i] =
+					AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 + (i >> 1);
+				gpu_resources.sdma_doorbell[0][i+1] =
+					AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 + 0x200 + (i >> 1);
+				gpu_resources.sdma_doorbell[1][i] =
+					AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 + (i >> 1);
+				gpu_resources.sdma_doorbell[1][i+1] =
+					AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 + 0x200 + (i >> 1);
+			} else {
+				gpu_resources.sdma_doorbell[0][i] =
+					AMDGPU_DOORBELL64_sDMA_ENGINE0 + (i >> 1);
+				gpu_resources.sdma_doorbell[0][i+1] =
+					AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200 + (i >> 1);
+				gpu_resources.sdma_doorbell[1][i] =
+					AMDGPU_DOORBELL64_sDMA_ENGINE1 + (i >> 1);
+				gpu_resources.sdma_doorbell[1][i+1] =
+					AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200 + (i >> 1);
+			}
 		}
+		/* Doorbells 0x0e0-0ff and 0x2e0-2ff are reserved for
+		 * SDMA, IH and VCN. So don't use them for the CP.
+		 */
+		gpu_resources.reserved_doorbell_mask = 0x1e0;
+		gpu_resources.reserved_doorbell_val  = 0x0e0;
 
 		kgd2kfd->device_init(adev->kfd, &gpu_resources);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 056fc6ef6c63..8e0d4f7196b4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -174,7 +174,7 @@ void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
 				struct amdgpu_vm *vm);
 void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm);
 void amdgpu_amdkfd_gpuvm_release_process_vm(struct kgd_dev *kgd, void *vm);
-uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm);
+uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm);
 int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 		struct kgd_dev *kgd, uint64_t va, uint64_t size,
 		void *vm, struct kgd_mem **mem,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index b2e45c8e2e0d..244d9834a381 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -142,7 +142,7 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 static void set_scratch_backing_va(struct kgd_dev *kgd,
 					uint64_t va, uint32_t vmid);
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t page_table_base);
+		uint64_t page_table_base);
 static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
 static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
 static uint32_t read_vmid_from_vmfault_reg(struct kgd_dev *kgd);
@@ -874,7 +874,7 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 }
 
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-			uint32_t page_table_base)
+			uint64_t page_table_base)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 
@@ -882,7 +882,8 @@ static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
 		pr_err("trying to set page table base for wrong VMID\n");
 		return;
 	}
-	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8,
+		lower_32_bits(page_table_base));
 }
 
 static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index ea7c18ce7754..9f149914ad6c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -45,8 +45,6 @@ enum hqd_dequeue_request_type {
 	RESET_WAVES
 };
 
-struct vi_sdma_mqd;
-
 /*
  * Register access functions
  */
@@ -100,7 +98,7 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 static void set_scratch_backing_va(struct kgd_dev *kgd,
 					uint64_t va, uint32_t vmid);
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t page_table_base);
+		uint64_t page_table_base);
 static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
 static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
 
@@ -282,7 +280,8 @@ static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
 
 	lock_srbm(kgd, mec, pipe, 0, 0);
 
-	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK);
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+			CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
 
 	unlock_srbm(kgd);
 
@@ -834,7 +833,7 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 }
 
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t page_table_base)
+		uint64_t page_table_base)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 
@@ -842,7 +841,8 @@ static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
 		pr_err("trying to set page table base for wrong VMID\n");
 		return;
 	}
-	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8,
+			lower_32_bits(page_table_base));
 }
 
 static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
index c9176537550b..42cb4c4e0929 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -138,7 +138,7 @@ static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
 static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 		uint8_t vmid);
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t page_table_base);
+		uint64_t page_table_base);
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 static void set_scratch_backing_va(struct kgd_dev *kgd,
 					uint64_t va, uint32_t vmid);
@@ -1013,11 +1013,10 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 }
 
 static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint32_t page_table_base)
+		uint64_t page_table_base)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	uint64_t base = (uint64_t)page_table_base << PAGE_SHIFT |
-		AMDGPU_PTE_VALID;
+	uint64_t base = page_table_base | AMDGPU_PTE_VALID;
 
 	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
 		pr_err("trying to set page table base for wrong VMID %u\n",
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 6ee9dc476c86..df0a059565f9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1131,11 +1131,15 @@ void amdgpu_amdkfd_gpuvm_release_process_vm(struct kgd_dev *kgd, void *vm)
 	amdgpu_vm_release_compute(adev, avm);
 }
 
-uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm)
+uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm)
 {
 	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+	struct amdgpu_bo *pd = avm->root.base.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
 
-	return avm->pd_phys_addr >> AMDGPU_GPU_PAGE_SHIFT;
+	if (adev->asic_type < CHIP_VEGA10)
+		return avm->pd_phys_addr >> AMDGPU_GPU_PAGE_SHIFT;
+	return avm->pd_phys_addr;
 }
 
 int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 176f28777f5e..5448cf27654e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -196,6 +196,19 @@ int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s)
 }
 
 /**
+ * amdgpu_fence_schedule_fallback - schedule fallback check
+ *
+ * @ring: pointer to struct amdgpu_ring
+ *
+ * Start a timer as fallback to our interrupts.
+ */
+static void amdgpu_fence_schedule_fallback(struct amdgpu_ring *ring)
+{
+	mod_timer(&ring->fence_drv.fallback_timer,
+		  jiffies + AMDGPU_FENCE_JIFFIES_TIMEOUT);
+}
+
+/**
  * amdgpu_fence_process - check for fence activity
  *
  * @ring: pointer to struct amdgpu_ring
@@ -203,8 +216,10 @@ int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s)
  * Checks the current fence value and calculates the last
  * signalled fence value. Wakes the fence queue if the
  * sequence number has increased.
+ *
+ * Returns true if fence was processed
  */
-void amdgpu_fence_process(struct amdgpu_ring *ring)
+bool amdgpu_fence_process(struct amdgpu_ring *ring)
 {
 	struct amdgpu_fence_driver *drv = &ring->fence_drv;
 	uint32_t seq, last_seq;
@@ -216,8 +231,12 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 
 	} while (atomic_cmpxchg(&drv->last_seq, last_seq, seq) != last_seq);
 
+	if (del_timer(&ring->fence_drv.fallback_timer) &&
+	    seq != ring->fence_drv.sync_seq)
+		amdgpu_fence_schedule_fallback(ring);
+
 	if (unlikely(seq == last_seq))
-		return;
+		return false;
 
 	last_seq &= drv->num_fences_mask;
 	seq &= drv->num_fences_mask;
@@ -244,6 +263,24 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 
 		dma_fence_put(fence);
 	} while (last_seq != seq);
+
+	return true;
+}
+
+/**
+ * amdgpu_fence_fallback - fallback for hardware interrupts
+ *
+ * @work: delayed work item
+ *
+ * Checks for fence activity.
+ */
+static void amdgpu_fence_fallback(struct timer_list *t)
+{
+	struct amdgpu_ring *ring = from_timer(ring, t,
+					      fence_drv.fallback_timer);
+
+	if (amdgpu_fence_process(ring))
+		DRM_WARN("Fence fallback timer expired on ring %s\n", ring->name);
 }
 
 /**
@@ -393,6 +430,8 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
 	atomic_set(&ring->fence_drv.last_seq, 0);
 	ring->fence_drv.initialized = false;
 
+	timer_setup(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback, 0);
+
 	ring->fence_drv.num_fences_mask = num_hw_submission * 2 - 1;
 	spin_lock_init(&ring->fence_drv.lock);
 	ring->fence_drv.fences = kcalloc(num_hw_submission * 2, sizeof(void *),
@@ -468,6 +507,7 @@ void amdgpu_fence_driver_fini(struct amdgpu_device *adev)
 		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
 			       ring->fence_drv.irq_type);
 		drm_sched_fini(&ring->sched);
+		del_timer_sync(&ring->fence_drv.fallback_timer);
 		for (j = 0; j <= ring->fence_drv.num_fences_mask; ++j)
 			dma_fence_put(ring->fence_drv.fences[j]);
 		kfree(ring->fence_drv.fences);
@@ -561,6 +601,27 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
 }
 
 /**
+ * amdgpu_fence_enable_signaling - enable signalling on fence
+ * @fence: fence
+ *
+ * This function is called with fence_queue lock held, and adds a callback
+ * to fence_queue that checks if this fence is signaled, and if so it
+ * signals the fence and removes itself.
+ */
+static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	struct amdgpu_ring *ring = fence->ring;
+
+	if (!timer_pending(&ring->fence_drv.fallback_timer))
+		amdgpu_fence_schedule_fallback(ring);
+
+	DMA_FENCE_TRACE(&fence->base, "armed on ring %i!\n", ring->idx);
+
+	return true;
+}
+
+/**
  * amdgpu_fence_free - free up the fence memory
  *
  * @rcu: RCU callback head
@@ -590,6 +651,7 @@ static void amdgpu_fence_release(struct dma_fence *f)
 static const struct dma_fence_ops amdgpu_fence_ops = {
 	.get_driver_name = amdgpu_fence_get_driver_name,
 	.get_timeline_name = amdgpu_fence_get_timeline_name,
+	.enable_signaling = amdgpu_fence_enable_signaling,
 	.release = amdgpu_fence_release,
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
index 4ed86218cef3..8af67f649660 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
@@ -24,46 +24,21 @@
 #include <drm/drmP.h>
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
-#include "amdgpu_amdkfd.h"
-
-/**
- * amdgpu_ih_ring_alloc - allocate memory for the IH ring
- *
- * @adev: amdgpu_device pointer
- *
- * Allocate a ring buffer for the interrupt controller.
- * Returns 0 for success, errors for failure.
- */
-static int amdgpu_ih_ring_alloc(struct amdgpu_device *adev)
-{
-	int r;
-
-	/* Allocate ring buffer */
-	if (adev->irq.ih.ring_obj == NULL) {
-		r = amdgpu_bo_create_kernel(adev, adev->irq.ih.ring_size,
-					    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-					    &adev->irq.ih.ring_obj,
-					    &adev->irq.ih.gpu_addr,
-					    (void **)&adev->irq.ih.ring);
-		if (r) {
-			DRM_ERROR("amdgpu: failed to create ih ring buffer (%d).\n", r);
-			return r;
-		}
-	}
-	return 0;
-}
 
 /**
  * amdgpu_ih_ring_init - initialize the IH state
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to initialize
+ * @ring_size: ring size to allocate
+ * @use_bus_addr: true when we can use dma_alloc_coherent
  *
  * Initializes the IH state and allocates a buffer
  * for the IH ring buffer.
  * Returns 0 for success, errors for failure.
  */
-int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
-			bool use_bus_addr)
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			unsigned ring_size, bool use_bus_addr)
 {
 	u32 rb_bufsz;
 	int r;
@@ -71,70 +46,76 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
 	/* Align ring size */
 	rb_bufsz = order_base_2(ring_size / 4);
 	ring_size = (1 << rb_bufsz) * 4;
-	adev->irq.ih.ring_size = ring_size;
-	adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
-	adev->irq.ih.rptr = 0;
-	adev->irq.ih.use_bus_addr = use_bus_addr;
-
-	if (adev->irq.ih.use_bus_addr) {
-		if (!adev->irq.ih.ring) {
-			/* add 8 bytes for the rptr/wptr shadows and
-			 * add them to the end of the ring allocation.
-			 */
-			adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
-								 adev->irq.ih.ring_size + 8,
-								 &adev->irq.ih.rb_dma_addr);
-			if (adev->irq.ih.ring == NULL)
-				return -ENOMEM;
-			memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
-			adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
-			adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
-		}
-		return 0;
+	ih->ring_size = ring_size;
+	ih->ptr_mask = ih->ring_size - 1;
+	ih->rptr = 0;
+	ih->use_bus_addr = use_bus_addr;
+
+	if (use_bus_addr) {
+		if (ih->ring)
+			return 0;
+
+		/* add 8 bytes for the rptr/wptr shadows and
+		 * add them to the end of the ring allocation.
+		 */
+		ih->ring = dma_alloc_coherent(adev->dev, ih->ring_size + 8,
+					      &ih->rb_dma_addr, GFP_KERNEL);
+		if (ih->ring == NULL)
+			return -ENOMEM;
+
+		memset((void *)ih->ring, 0, ih->ring_size + 8);
+		ih->wptr_offs = (ih->ring_size / 4) + 0;
+		ih->rptr_offs = (ih->ring_size / 4) + 1;
 	} else {
-		r = amdgpu_device_wb_get(adev, &adev->irq.ih.wptr_offs);
+		r = amdgpu_device_wb_get(adev, &ih->wptr_offs);
+		if (r)
+			return r;
+
+		r = amdgpu_device_wb_get(adev, &ih->rptr_offs);
 		if (r) {
-			dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
+			amdgpu_device_wb_free(adev, ih->wptr_offs);
 			return r;
 		}
 
-		r = amdgpu_device_wb_get(adev, &adev->irq.ih.rptr_offs);
+		r = amdgpu_bo_create_kernel(adev, ih->ring_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &ih->ring_obj, &ih->gpu_addr,
+					    (void **)&ih->ring);
 		if (r) {
-			amdgpu_device_wb_free(adev, adev->irq.ih.wptr_offs);
-			dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
+			amdgpu_device_wb_free(adev, ih->rptr_offs);
+			amdgpu_device_wb_free(adev, ih->wptr_offs);
 			return r;
 		}
-
-		return amdgpu_ih_ring_alloc(adev);
 	}
+	return 0;
 }
 
 /**
  * amdgpu_ih_ring_fini - tear down the IH state
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to tear down
  *
  * Tears down the IH state and frees buffer
  * used for the IH ring buffer.
  */
-void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
 {
-	if (adev->irq.ih.use_bus_addr) {
-		if (adev->irq.ih.ring) {
-			/* add 8 bytes for the rptr/wptr shadows and
-			 * add them to the end of the ring allocation.
-			 */
-			pci_free_consistent(adev->pdev, adev->irq.ih.ring_size + 8,
-					    (void *)adev->irq.ih.ring,
-					    adev->irq.ih.rb_dma_addr);
-			adev->irq.ih.ring = NULL;
-		}
+	if (ih->use_bus_addr) {
+		if (!ih->ring)
+			return;
+
+		/* add 8 bytes for the rptr/wptr shadows and
+		 * add them to the end of the ring allocation.
+		 */
+		dma_free_coherent(adev->dev, ih->ring_size + 8,
+				  (void *)ih->ring, ih->rb_dma_addr);
+		ih->ring = NULL;
 	} else {
-		amdgpu_bo_free_kernel(&adev->irq.ih.ring_obj,
-				      &adev->irq.ih.gpu_addr,
-				      (void **)&adev->irq.ih.ring);
-		amdgpu_device_wb_free(adev, adev->irq.ih.wptr_offs);
-		amdgpu_device_wb_free(adev, adev->irq.ih.rptr_offs);
+		amdgpu_bo_free_kernel(&ih->ring_obj, &ih->gpu_addr,
+				      (void **)&ih->ring);
+		amdgpu_device_wb_free(adev, ih->wptr_offs);
+		amdgpu_device_wb_free(adev, ih->rptr_offs);
 	}
 }
 
@@ -142,56 +123,43 @@ void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
  * amdgpu_ih_process - interrupt handler
  *
  * @adev: amdgpu_device pointer
+ * @ih: ih ring to process
  *
  * Interrupt hander (VI), walk the IH ring.
  * Returns irq process return code.
  */
-int amdgpu_ih_process(struct amdgpu_device *adev)
+int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+		      void (*callback)(struct amdgpu_device *adev,
+				       struct amdgpu_ih_ring *ih))
 {
-	struct amdgpu_iv_entry entry;
 	u32 wptr;
 
-	if (!adev->irq.ih.enabled || adev->shutdown)
+	if (!ih->enabled || adev->shutdown)
 		return IRQ_NONE;
 
 	wptr = amdgpu_ih_get_wptr(adev);
 
 restart_ih:
 	/* is somebody else already processing irqs? */
-	if (atomic_xchg(&adev->irq.ih.lock, 1))
+	if (atomic_xchg(&ih->lock, 1))
 		return IRQ_NONE;
 
-	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, adev->irq.ih.rptr, wptr);
+	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr);
 
 	/* Order reading of wptr vs. reading of IH ring data */
 	rmb();
 
-	while (adev->irq.ih.rptr != wptr) {
-		u32 ring_index = adev->irq.ih.rptr >> 2;
-
-		/* Prescreening of high-frequency interrupts */
-		if (!amdgpu_ih_prescreen_iv(adev)) {
-			adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
-			continue;
-		}
-
-		/* Before dispatching irq to IP blocks, send it to amdkfd */
-		amdgpu_amdkfd_interrupt(adev,
-				(const void *) &adev->irq.ih.ring[ring_index]);
-
-		entry.iv_entry = (const uint32_t *)
-			&adev->irq.ih.ring[ring_index];
-		amdgpu_ih_decode_iv(adev, &entry);
-		adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
-
-		amdgpu_irq_dispatch(adev, &entry);
+	while (ih->rptr != wptr) {
+		callback(adev, ih);
+		ih->rptr &= ih->ptr_mask;
 	}
+
 	amdgpu_ih_set_rptr(adev);
-	atomic_set(&adev->irq.ih.lock, 0);
+	atomic_set(&ih->lock, 0);
 
 	/* make sure wptr hasn't changed while processing */
 	wptr = amdgpu_ih_get_wptr(adev);
-	if (wptr != adev->irq.ih.rptr)
+	if (wptr != ih->rptr)
 		goto restart_ih;
 
 	return IRQ_HANDLED;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
index 0d5b3f5201d2..9ce8c93ec19b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
@@ -24,12 +24,8 @@
 #ifndef __AMDGPU_IH_H__
 #define __AMDGPU_IH_H__
 
-#include "soc15_ih_clientid.h"
-
 struct amdgpu_device;
-
-#define AMDGPU_IH_CLIENTID_LEGACY 0
-#define AMDGPU_IH_CLIENTID_MAX SOC15_IH_CLIENTID_MAX
+struct amdgpu_iv_entry;
 
 /*
  * R6xx+ IH ring
@@ -51,22 +47,6 @@ struct amdgpu_ih_ring {
 	dma_addr_t		rb_dma_addr; /* only used when use_bus_addr = true */
 };
 
-#define AMDGPU_IH_SRC_DATA_MAX_SIZE_DW 4
-
-struct amdgpu_iv_entry {
-	unsigned client_id;
-	unsigned src_id;
-	unsigned ring_id;
-	unsigned vmid;
-	unsigned vmid_src;
-	uint64_t timestamp;
-	unsigned timestamp_src;
-	unsigned pasid;
-	unsigned pasid_src;
-	unsigned src_data[AMDGPU_IH_SRC_DATA_MAX_SIZE_DW];
-	const uint32_t *iv_entry;
-};
-
 /* provided by the ih block */
 struct amdgpu_ih_funcs {
 	/* ring read/write ptr handling, called from interrupt context */
@@ -82,9 +62,11 @@ struct amdgpu_ih_funcs {
 #define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
 #define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
 
-int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
-			bool use_bus_addr);
-void amdgpu_ih_ring_fini(struct amdgpu_device *adev);
-int amdgpu_ih_process(struct amdgpu_device *adev);
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+			unsigned ring_size, bool use_bus_addr);
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
+int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
+		      void (*callback)(struct amdgpu_device *adev,
+				       struct amdgpu_ih_ring *ih));
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index b927e8798534..52c17f6219a7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -51,6 +51,7 @@
 #include "atom.h"
 #include "amdgpu_connectors.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
 
 #include <linux/pm_runtime.h>
 
@@ -123,7 +124,7 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 	int r;
 
 	spin_lock_irqsave(&adev->irq.lock, irqflags);
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
@@ -147,6 +148,34 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 }
 
 /**
+ * amdgpu_irq_callback - callback from the IH ring
+ *
+ * @adev: amdgpu device pointer
+ * @ih: amdgpu ih ring
+ *
+ * Callback from IH ring processing to handle the entry at the current position
+ * and advance the read pointer.
+ */
+static void amdgpu_irq_callback(struct amdgpu_device *adev,
+				struct amdgpu_ih_ring *ih)
+{
+	u32 ring_index = ih->rptr >> 2;
+	struct amdgpu_iv_entry entry;
+
+	/* Prescreening of high-frequency interrupts */
+	if (!amdgpu_ih_prescreen_iv(adev))
+		return;
+
+	/* Before dispatching irq to IP blocks, send it to amdkfd */
+	amdgpu_amdkfd_interrupt(adev, (const void *) &ih->ring[ring_index]);
+
+	entry.iv_entry = (const uint32_t *)&ih->ring[ring_index];
+	amdgpu_ih_decode_iv(adev, &entry);
+
+	amdgpu_irq_dispatch(adev, &entry);
+}
+
+/**
  * amdgpu_irq_handler - IRQ handler
  *
  * @irq: IRQ number (unused)
@@ -163,7 +192,7 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg)
 	struct amdgpu_device *adev = dev->dev_private;
 	irqreturn_t ret;
 
-	ret = amdgpu_ih_process(adev);
+	ret = amdgpu_ih_process(adev, &adev->irq.ih, amdgpu_irq_callback);
 	if (ret == IRQ_HANDLED)
 		pm_runtime_mark_last_busy(dev->dev);
 	return ret;
@@ -273,7 +302,7 @@ void amdgpu_irq_fini(struct amdgpu_device *adev)
 		cancel_work_sync(&adev->reset_work);
 	}
 
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
@@ -313,7 +342,7 @@ int amdgpu_irq_add_id(struct amdgpu_device *adev,
 		      unsigned client_id, unsigned src_id,
 		      struct amdgpu_irq_src *source)
 {
-	if (client_id >= AMDGPU_IH_CLIENTID_MAX)
+	if (client_id >= AMDGPU_IRQ_CLIENTID_MAX)
 		return -EINVAL;
 
 	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID)
@@ -367,7 +396,7 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
 
 	trace_amdgpu_iv(entry);
 
-	if (client_id >= AMDGPU_IH_CLIENTID_MAX) {
+	if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) {
 		DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);
 		return;
 	}
@@ -440,7 +469,7 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev)
 {
 	int i, j, k;
 
-	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+	for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) {
 		if (!adev->irq.client[i].sources)
 			continue;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index 3375ad778edc..f6ce171cb8aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -25,19 +25,38 @@
 #define __AMDGPU_IRQ_H__
 
 #include <linux/irqdomain.h>
+#include "soc15_ih_clientid.h"
 #include "amdgpu_ih.h"
 
-#define AMDGPU_MAX_IRQ_SRC_ID	0x100
+#define AMDGPU_MAX_IRQ_SRC_ID		0x100
 #define AMDGPU_MAX_IRQ_CLIENT_ID	0x100
 
+#define AMDGPU_IRQ_CLIENTID_LEGACY	0
+#define AMDGPU_IRQ_CLIENTID_MAX		SOC15_IH_CLIENTID_MAX
+
+#define AMDGPU_IRQ_SRC_DATA_MAX_SIZE_DW	4
+
 struct amdgpu_device;
-struct amdgpu_iv_entry;
 
 enum amdgpu_interrupt_state {
 	AMDGPU_IRQ_STATE_DISABLE,
 	AMDGPU_IRQ_STATE_ENABLE,
 };
 
+struct amdgpu_iv_entry {
+	unsigned client_id;
+	unsigned src_id;
+	unsigned ring_id;
+	unsigned vmid;
+	unsigned vmid_src;
+	uint64_t timestamp;
+	unsigned timestamp_src;
+	unsigned pasid;
+	unsigned pasid_src;
+	unsigned src_data[AMDGPU_IRQ_SRC_DATA_MAX_SIZE_DW];
+	const uint32_t *iv_entry;
+};
+
 struct amdgpu_irq_src {
 	unsigned				num_types;
 	atomic_t				*enabled_types;
@@ -63,7 +82,7 @@ struct amdgpu_irq {
 	bool				installed;
 	spinlock_t			lock;
 	/* interrupt sources */
-	struct amdgpu_irq_client	client[AMDGPU_IH_CLIENTID_MAX];
+	struct amdgpu_irq_client	client[AMDGPU_IRQ_CLIENTID_MAX];
 
 	/* status, etc. */
 	bool				msi_enabled; /* msi enabled */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
index 8c334fc808c2..18d989e0e362 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -1976,6 +1976,7 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
 static int amdgpu_debugfs_pm_info_pp(struct seq_file *m, struct amdgpu_device *adev)
 {
 	uint32_t value;
+	uint64_t value64;
 	uint32_t query = 0;
 	int size;
 
@@ -2014,6 +2015,10 @@ static int amdgpu_debugfs_pm_info_pp(struct seq_file *m, struct amdgpu_device *a
 		seq_printf(m, "GPU Load: %u %%\n", value);
 	seq_printf(m, "\n");
 
+	/* SMC feature mask */
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK, (void *)&value64, &size))
+		seq_printf(m, "SMC Feature Mask: 0x%016llx\n", value64);
+
 	/* UVD clocks */
 	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_POWER, (void *)&value, &size)) {
 		if (!value) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index 44fc665e4577..4caa301ce454 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -77,6 +77,7 @@ struct amdgpu_fence_driver {
 	bool				initialized;
 	struct amdgpu_irq_src		*irq_src;
 	unsigned			irq_type;
+	struct timer_list		fallback_timer;
 	unsigned			num_fences_mask;
 	spinlock_t			lock;
 	struct dma_fence		**fences;
@@ -96,7 +97,7 @@ void amdgpu_fence_driver_resume(struct amdgpu_device *adev);
 int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence,
 		      unsigned flags);
 int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s);
-void amdgpu_fence_process(struct amdgpu_ring *ring);
+bool amdgpu_fence_process(struct amdgpu_ring *ring);
 int amdgpu_fence_wait_empty(struct amdgpu_ring *ring);
 signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
 				      uint32_t wait_seq,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 2e87414422f9..e9bf70e2ac51 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -103,7 +103,7 @@ TRACE_EVENT(amdgpu_iv,
 			   __entry->src_data[2] = iv->src_data[2];
 			   __entry->src_data[3] = iv->src_data[3];
 			   ),
-	    TP_printk("client_id:%u src_id:%u ring:%u vmid:%u timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x\n",
+	    TP_printk("client_id:%u src_id:%u ring:%u vmid:%u timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x",
 		      __entry->client_id, __entry->src_id,
 		      __entry->ring_id, __entry->vmid,
 		      __entry->timestamp, __entry->pasid,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index a73674f9a0f5..2a2eb0143f48 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -36,6 +36,7 @@
 #include "soc15_common.h"
 
 #include "vcn/vcn_1_0_offset.h"
+#include "vcn/vcn_1_0_sh_mask.h"
 
 /* 1 second timeout */
 #define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
@@ -212,18 +213,158 @@ int amdgpu_vcn_resume(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int amdgpu_vcn_pause_dpg_mode(struct amdgpu_device *adev,
+				     struct dpg_pause_state *new_state)
+{
+	int ret_code;
+	uint32_t reg_data = 0;
+	uint32_t reg_data2 = 0;
+	struct amdgpu_ring *ring;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK))
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+
+			if (!ret_code) {
+				/* pause DPG non-jpeg */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+						   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+						   UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK, ret_code);
+
+				/* Restore */
+				ring = &adev->vcn.ring_enc[0];
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+
+				ring = &adev->vcn.ring_enc[1];
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+
+				ring = &adev->vcn.ring_dec;
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+					     lower_32_bits(ring->wptr) | 0x80000000);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+			}
+		} else {
+			/* unpause dpg non-jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.pause_state.fw_based = new_state->fw_based;
+	}
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.pause_state.jpeg != new_state->jpeg) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+			(~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->jpeg == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK))
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+
+			if (!ret_code) {
+				/* Make sure JPRG Snoop is disabled before sending the pause */
+				reg_data2 = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+				reg_data2 |= UVD_POWER_STATUS__JRBC_SNOOP_DIS_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, reg_data2);
+
+				/* pause DPG jpeg */
+				reg_data |= UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK, ret_code);
+
+				/* Restore */
+				ring = &adev->vcn.ring_jpeg;
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000001L | 0x00000002L);
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+					     lower_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+					     upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+
+				ring = &adev->vcn.ring_dec;
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+					     lower_32_bits(ring->wptr) | 0x80000000);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						   UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+						   UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+			}
+		} else {
+			/* unpause dpg jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.pause_state.jpeg = new_state->jpeg;
+	}
+
+	return 0;
+}
+
 static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
 		container_of(work, struct amdgpu_device, vcn.idle_work.work);
-	unsigned fences = amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
-	unsigned i;
+	unsigned int fences = 0;
+	unsigned int i;
 
 	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
 		fences += amdgpu_fence_count_emitted(&adev->vcn.ring_enc[i]);
 	}
 
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)	{
+		struct dpg_pause_state new_state;
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(&adev->vcn.ring_jpeg))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		amdgpu_vcn_pause_dpg_mode(adev, &new_state);
+	}
+
 	fences += amdgpu_fence_count_emitted(&adev->vcn.ring_jpeg);
+	fences += amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
 
 	if (fences == 0) {
 		amdgpu_gfx_off_ctrl(adev, true);
@@ -250,6 +391,22 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
 							       AMD_PG_STATE_UNGATE);
 	}
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)	{
+		struct dpg_pause_state new_state;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = adev->vcn.pause_state.fw_based;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = adev->vcn.pause_state.jpeg;
+
+		amdgpu_vcn_pause_dpg_mode(adev, &new_state);
+	}
 }
 
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)
@@ -264,7 +421,7 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
 	unsigned i;
 	int r;
 
-	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0xCAFEDEAD);
+	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
 	if (r) {
 		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
@@ -272,11 +429,11 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
 		return r;
 	}
 	amdgpu_ring_write(ring,
-		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID));
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
 		if (tmp == 0xDEADBEEF)
 			break;
 		DRM_UDELAY(1);
@@ -616,7 +773,7 @@ int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring)
 	unsigned i;
 	int r;
 
-	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0xCAFEDEAD);
+	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD);
 	r = amdgpu_ring_alloc(ring, 3);
 
 	if (r) {
@@ -626,12 +783,12 @@ int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring)
 	}
 
 	amdgpu_ring_write(ring,
-		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0, 0, 0));
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0, 0, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID));
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
 		if (tmp == 0xDEADBEEF)
 			break;
 		DRM_UDELAY(1);
@@ -665,7 +822,7 @@ static int amdgpu_vcn_jpeg_set_reg(struct amdgpu_ring *ring, uint32_t handle,
 
 	ib = &job->ibs[0];
 
-	ib->ptr[0] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH), 0, 0, PACKETJ_TYPE0);
+	ib->ptr[0] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0, 0, PACKETJ_TYPE0);
 	ib->ptr[1] = 0xDEADBEEF;
 	for (i = 2; i < 16; i += 2) {
 		ib->ptr[i] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
@@ -714,7 +871,7 @@ int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 		r = 0;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH));
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
 		if (tmp == 0xDEADBEEF)
 			break;
 		DRM_UDELAY(1);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
index 0b0b8638d73f..0b88a4672da5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
@@ -56,6 +56,16 @@ enum engine_status_constants {
 	UVD_STATUS__RBC_BUSY = 0x1,
 };
 
+enum internal_dpg_state {
+	VCN_DPG_STATE__UNPAUSE = 0,
+	VCN_DPG_STATE__PAUSE,
+};
+
+struct dpg_pause_state {
+	enum internal_dpg_state fw_based;
+	enum internal_dpg_state jpeg;
+};
+
 struct amdgpu_vcn {
 	struct amdgpu_bo	*vcpu_bo;
 	void			*cpu_addr;
@@ -69,6 +79,8 @@ struct amdgpu_vcn {
 	struct amdgpu_ring	ring_jpeg;
 	struct amdgpu_irq_src	irq;
 	unsigned		num_enc_rings;
+	enum amd_powergating_state cur_state;
+	struct dpg_pause_state pause_state;
 };
 
 int amdgpu_vcn_sw_init(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
index d2469453dca2..79220a91abe3 100644
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
@@ -6277,12 +6277,12 @@ static int ci_dpm_sw_init(void *handle)
 	int ret;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230,
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 230,
 				&adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231,
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 231,
 				&adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
index 44d10c2172f6..b5775c6a857b 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
@@ -276,7 +276,7 @@ static void cik_ih_decode_iv(struct amdgpu_device *adev,
 	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
 	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
@@ -318,7 +318,7 @@ static int cik_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -332,7 +332,7 @@ static int cik_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
@@ -468,8 +468,7 @@ static const struct amdgpu_ih_funcs cik_ih_funcs = {
 
 static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &cik_ih_funcs;
+	adev->irq.ih_funcs = &cik_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version cik_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index 154b1499b07e..b918c8886b75 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -970,19 +970,19 @@ static int cik_sdma_sw_init(void *handle)
 	}
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 224,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 247,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 247,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
@@ -1370,10 +1370,8 @@ static const struct amdgpu_buffer_funcs cik_sdma_buffer_funcs = {
 
 static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs cik_sdma_vm_pte_funcs = {
@@ -1389,15 +1387,13 @@ static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
 	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
-			adev->vm_manager.vm_pte_rqs[i] =
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
-		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_rqs[i] =
+			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
+	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version cik_sdma_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
index 960c29e17da6..df5ac4d85a00 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
@@ -255,7 +255,7 @@ static void cz_ih_decode_iv(struct amdgpu_device *adev,
 	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
 	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
@@ -297,7 +297,7 @@ static int cz_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -311,7 +311,7 @@ static int cz_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
@@ -449,8 +449,7 @@ static const struct amdgpu_ih_funcs cz_ih_funcs = {
 
 static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &cz_ih_funcs;
+	adev->irq.ih_funcs = &cz_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version cz_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 3916aa6cc4ec..4cfecdce29a3 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -2746,19 +2746,19 @@ static int dce_v10_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
 	if (r)
 		return r;
 
@@ -3570,8 +3570,7 @@ static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
 
 static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v10_0_display_funcs;
+	adev->mode_info.funcs = &dce_v10_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v10_0_crtc_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index 4ffb612a4e53..7c868916d90f 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -2867,19 +2867,19 @@ static int dce_v11_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
 	if (r)
 		return r;
 
@@ -3702,8 +3702,7 @@ static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
 
 static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v11_0_display_funcs;
+	adev->mode_info.funcs = &dce_v11_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v11_0_crtc_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index 480c5348a14f..17eaaba36017 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -2616,19 +2616,19 @@ static int dce_v6_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 42, &adev->hpd_irq);
 	if (r)
 		return r;
 
@@ -3376,8 +3376,7 @@ static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
 
 static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v6_0_display_funcs;
+	adev->mode_info.funcs = &dce_v6_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_crtc_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index 797196476c94..8c0576978d36 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -2643,19 +2643,19 @@ static int dce_v8_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
 		if (r)
 			return r;
 	}
 
 	for (i = 8; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
 		if (r)
 			return r;
 	}
 
 	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 42, &adev->hpd_irq);
 	if (r)
 		return r;
 
@@ -3458,8 +3458,7 @@ static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
 
 static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_v8_0_display_funcs;
+	adev->mode_info.funcs = &dce_v8_0_display_funcs;
 }
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_crtc_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
index 15257634a53a..fdace004544d 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
@@ -372,7 +372,7 @@ static int dce_virtual_sw_init(void *handle)
 	int r, i;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SMU_DISP_TIMER2_TRIGGER, &adev->crtc_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SMU_DISP_TIMER2_TRIGGER, &adev->crtc_irq);
 	if (r)
 		return r;
 
@@ -649,8 +649,7 @@ static const struct amdgpu_display_funcs dce_virtual_display_funcs = {
 
 static void dce_virtual_set_display_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.funcs == NULL)
-		adev->mode_info.funcs = &dce_virtual_display_funcs;
+	adev->mode_info.funcs = &dce_virtual_display_funcs;
 }
 
 static int dce_virtual_pageflip(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
index de184a886057..d76eb27945dc 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
@@ -1552,7 +1552,7 @@ static void gfx_v6_0_config_init(struct amdgpu_device *adev)
 	adev->gfx.config.double_offchip_lds_buf = 0;
 }
 
-static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v6_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config = 0;
 	u32 mc_shared_chmap, mc_arb_ramcfg;
@@ -3094,15 +3094,15 @@ static int gfx_v6_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	int i, r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184, &adev->gfx.priv_reg_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 184, &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185, &adev->gfx.priv_inst_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 185, &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
@@ -3175,7 +3175,7 @@ static int gfx_v6_0_hw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	gfx_v6_0_gpu_init(adev);
+	gfx_v6_0_constants_init(adev);
 
 	r = gfx_v6_0_rlc_resume(adev);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index fc39ebbc9d9f..0e72bc09939a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -1886,14 +1886,14 @@ static void gfx_v7_0_config_init(struct amdgpu_device *adev)
 }
 
 /**
- * gfx_v7_0_gpu_init - setup the 3D engine
+ * gfx_v7_0_constants_init - setup the 3D engine
  *
  * @adev: amdgpu_device pointer
  *
- * Configures the 3D engine and tiling configuration
- * registers so that the 3D engine is usable.
+ * init the gfx constants such as the 3D engine, tiling configuration
+ * registers, maximum number of quad pipes, render backends...
  */
-static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v7_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 sh_mem_cfg, sh_static_mem_cfg, sh_mem_base;
 	u32 tmp;
@@ -4516,18 +4516,18 @@ static int gfx_v7_0_sw_init(void *handle)
 	adev->gfx.mec.num_queue_per_pipe = 8;
 
 	/* EOP Event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 184,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 185,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
@@ -4624,7 +4624,7 @@ static int gfx_v7_0_hw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	gfx_v7_0_gpu_init(adev);
+	gfx_v7_0_constants_init(adev);
 
 	/* init rlc */
 	r = gfx_v7_0_rlc_resume(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 470dc80f4fe7..2aeef2bb93a4 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -2049,35 +2049,35 @@ static int gfx_v8_0_sw_init(void *handle)
 	adev->gfx.mec.num_queue_per_pipe = 8;
 
 	/* KIQ event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_INT_IB2, &adev->gfx.kiq.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_INT_IB2, &adev->gfx.kiq.irq);
 	if (r)
 		return r;
 
 	/* EOP Event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_END_OF_PIPE, &adev->gfx.eop_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_END_OF_PIPE, &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_REG_FAULT,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_REG_FAULT,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_INSTR_FAULT,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_INSTR_FAULT,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
 
 	/* Add CP EDC/ECC irq  */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_ECC_ERROR,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_ECC_ERROR,
 			      &adev->gfx.cp_ecc_error_irq);
 	if (r)
 		return r;
 
 	/* SQ interrupts. */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SQ_INTERRUPT_MSG,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SQ_INTERRUPT_MSG,
 			      &adev->gfx.sq_irq);
 	if (r) {
 		DRM_ERROR("amdgpu_irq_add() for SQ failed: %d\n", r);
@@ -3835,7 +3835,7 @@ static void gfx_v8_0_config_init(struct amdgpu_device *adev)
 	}
 }
 
-static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v8_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 tmp, sh_static_mem_cfg;
 	int i;
@@ -4208,31 +4208,11 @@ static int gfx_v8_0_rlc_load_microcode(struct amdgpu_device *adev)
 static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev)
 {
 	int r;
-	u32 tmp;
 
 	gfx_v8_0_rlc_stop(adev);
-
-	/* disable CG */
-	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL);
-	tmp &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK |
-		 RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
-	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
-	if (adev->asic_type == CHIP_POLARIS11 ||
-	    adev->asic_type == CHIP_POLARIS10 ||
-	    adev->asic_type == CHIP_POLARIS12 ||
-	    adev->asic_type == CHIP_VEGAM) {
-		tmp = RREG32(mmRLC_CGCG_CGLS_CTRL_3D);
-		tmp &= ~0x3;
-		WREG32(mmRLC_CGCG_CGLS_CTRL_3D, tmp);
-	}
-
-	/* disable PG */
-	WREG32(mmRLC_PG_CNTL, 0);
-
 	gfx_v8_0_rlc_reset(adev);
 	gfx_v8_0_init_pg(adev);
 
-
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
 		/* legacy rlc firmware loading */
 		r = gfx_v8_0_rlc_load_microcode(adev);
@@ -5039,7 +5019,7 @@ static int gfx_v8_0_hw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	gfx_v8_0_init_golden_registers(adev);
-	gfx_v8_0_gpu_init(adev);
+	gfx_v8_0_constants_init(adev);
 
 	r = gfx_v8_0_rlc_resume(adev);
 	if (r)
@@ -5080,6 +5060,55 @@ static int gfx_v8_0_kcq_disable(struct amdgpu_device *adev)
 	return r;
 }
 
+static bool gfx_v8_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE)
+		|| RREG32(mmGRBM_STATUS2) != 0x8)
+		return false;
+	else
+		return true;
+}
+
+static bool gfx_v8_0_rlc_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (RREG32(mmGRBM_STATUS2) != 0x8)
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v8_0_wait_for_rlc_idle(void *handle)
+{
+	unsigned int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v8_0_rlc_is_idle(handle))
+			return 0;
+
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v8_0_wait_for_idle(void *handle)
+{
+	unsigned int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v8_0_is_idle(handle))
+			return 0;
+
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
 static int gfx_v8_0_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -5098,9 +5127,16 @@ static int gfx_v8_0_hw_fini(void *handle)
 		pr_debug("For SRIOV client, shouldn't do anything.\n");
 		return 0;
 	}
-	gfx_v8_0_cp_enable(adev, false);
-	gfx_v8_0_rlc_stop(adev);
-
+	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+	if (!gfx_v8_0_wait_for_idle(adev))
+		gfx_v8_0_cp_enable(adev, false);
+	else
+		pr_err("cp is busy, skip halt cp\n");
+	if (!gfx_v8_0_wait_for_rlc_idle(adev))
+		gfx_v8_0_rlc_stop(adev);
+	else
+		pr_err("rlc is busy, skip halt rlc\n");
+	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 	return 0;
 }
 
@@ -5121,30 +5157,6 @@ static int gfx_v8_0_resume(void *handle)
 	return r;
 }
 
-static bool gfx_v8_0_is_idle(void *handle)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE))
-		return false;
-	else
-		return true;
-}
-
-static int gfx_v8_0_wait_for_idle(void *handle)
-{
-	unsigned i;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v8_0_is_idle(handle))
-			return 0;
-
-		udelay(1);
-	}
-	return -ETIMEDOUT;
-}
-
 static bool gfx_v8_0_check_soft_reset(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index f369d9603435..261bb051b14d 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -1847,7 +1847,7 @@ static void gfx_v9_0_init_compute_vmid(struct amdgpu_device *adev)
 	mutex_unlock(&adev->srbm_mutex);
 }
 
-static void gfx_v9_0_gpu_init(struct amdgpu_device *adev)
+static void gfx_v9_0_constants_init(struct amdgpu_device *adev)
 {
 	u32 tmp;
 	int i;
@@ -3235,7 +3235,7 @@ static int gfx_v9_0_hw_init(void *handle)
 
 	gfx_v9_0_init_golden_registers(adev);
 
-	gfx_v9_0_gpu_init(adev);
+	gfx_v9_0_constants_init(adev);
 
 	r = gfx_v9_0_csb_vram_pin(adev);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
index 65f58ebcf835..ceb7847b504f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
@@ -82,7 +82,8 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 		 * to get rid of the VM fault and hardware hang.
 		 */
 		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-			     (max(adev->gmc.vram_end, adev->gmc.agp_end) >> 18) + 0x1);
+			     max((adev->gmc.vram_end >> 18) + 0x1,
+				 adev->gmc.agp_end >> 18));
 	else
 		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
 			     max(adev->gmc.vram_end, adev->gmc.agp_end) >> 18);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 4411463ca719..e1c2b4e9c7b2 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -859,11 +859,11 @@ static int gmc_v6_0_sw_init(void *handle)
 		adev->gmc.vram_type = gmc_v6_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 146, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 147, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
@@ -1180,8 +1180,7 @@ static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
 
 static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gmc.gmc_funcs == NULL)
-		adev->gmc.gmc_funcs = &gmc_v6_0_gmc_funcs;
+	adev->gmc.gmc_funcs = &gmc_v6_0_gmc_funcs;
 }
 
 static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index ae776ce9a415..910c4ce19cb3 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -991,11 +991,11 @@ static int gmc_v7_0_sw_init(void *handle)
 		adev->gmc.vram_type = gmc_v7_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
@@ -1388,8 +1388,7 @@ static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
 
 static void gmc_v7_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gmc.gmc_funcs == NULL)
-		adev->gmc.gmc_funcs = &gmc_v7_0_gmc_funcs;
+	adev->gmc.gmc_funcs = &gmc_v7_0_gmc_funcs;
 }
 
 static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 53ae49b8bde8..1d3265c97b70 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -1095,11 +1095,11 @@ static int gmc_v8_0_sw_init(void *handle)
 		adev->gmc.vram_type = gmc_v8_0_convert_vram_type(tmp);
 	}
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
 	if (r)
 		return r;
 
@@ -1733,8 +1733,7 @@ static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
 
 static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gmc.gmc_funcs == NULL)
-		adev->gmc.gmc_funcs = &gmc_v8_0_gmc_funcs;
+	adev->gmc.gmc_funcs = &gmc_v8_0_gmc_funcs;
 }
 
 static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index aad3c7c5fb3a..f35d7a554ad5 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -593,8 +593,7 @@ static const struct amdgpu_gmc_funcs gmc_v9_0_gmc_funcs = {
 
 static void gmc_v9_0_set_gmc_funcs(struct amdgpu_device *adev)
 {
-	if (adev->gmc.gmc_funcs == NULL)
-		adev->gmc.gmc_funcs = &gmc_v9_0_gmc_funcs;
+	adev->gmc.gmc_funcs = &gmc_v9_0_gmc_funcs;
 }
 
 static int gmc_v9_0_early_init(void *handle)
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
index 842c4b677b4d..cf0fc61aebe6 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
@@ -255,7 +255,7 @@ static void iceland_ih_decode_iv(struct amdgpu_device *adev,
 	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
 	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
@@ -297,7 +297,7 @@ static int iceland_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -311,7 +311,7 @@ static int iceland_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
@@ -447,8 +447,7 @@ static const struct amdgpu_ih_funcs iceland_ih_funcs = {
 
 static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &iceland_ih_funcs;
+	adev->irq.ih_funcs = &iceland_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version iceland_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
index cb79a93c2eb7..d0e478f43443 100644
--- a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
@@ -2995,12 +2995,12 @@ static int kv_dpm_sw_init(void *handle)
 	int ret;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230,
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 230,
 				&adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231,
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 231,
 				&adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
index 80698b5ffa4a..14649f8475f3 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
@@ -100,7 +100,8 @@ static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 		 * to get rid of the VM fault and hardware hang.
 		 */
 		WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-			     (max(adev->gmc.vram_end, adev->gmc.agp_end) >> 18) + 0x1);
+			     max((adev->gmc.vram_end >> 18) + 0x1,
+				 adev->gmc.agp_end >> 18));
 	else
 		WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
 			     max(adev->gmc.vram_end, adev->gmc.agp_end) >> 18);
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
index 842567b53df5..64e875d528dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
@@ -580,11 +580,11 @@ int xgpu_vi_mailbox_add_irq_id(struct amdgpu_device *adev)
 {
 	int r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 135, &adev->virt.rcv_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 135, &adev->virt.rcv_irq);
 	if (r)
 		return r;
 
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 138, &adev->virt.ack_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 138, &adev->virt.ack_irq);
 	if (r) {
 		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
 		return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index c403bdf8ad70..cd781abc4953 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -898,19 +898,19 @@ static int sdma_v2_4_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
@@ -1296,10 +1296,8 @@ static const struct amdgpu_buffer_funcs sdma_v2_4_buffer_funcs = {
 
 static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v2_4_vm_pte_funcs = {
@@ -1315,15 +1313,13 @@ static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
 	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
-			adev->vm_manager.vm_pte_rqs[i] =
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
-		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_rqs[i] =
+			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
+	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v2_4_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index 2677d6a1bf42..6d5c8ac64874 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -1177,19 +1177,19 @@ static int sdma_v3_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* SDMA trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
 			      &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 241,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
 
 	/* SDMA Privileged inst */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
 			      &adev->sdma.illegal_inst_irq);
 	if (r)
 		return r;
@@ -1736,10 +1736,8 @@ static const struct amdgpu_buffer_funcs sdma_v3_0_buffer_funcs = {
 
 static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v3_0_vm_pte_funcs = {
@@ -1755,15 +1753,13 @@ static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
-			adev->vm_manager.vm_pte_rqs[i] =
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
-		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_rqs[i] =
+			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
+	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v3_0_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index 2ea1f0d8f5be..a3e2ed15fff2 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -1320,9 +1320,15 @@ static int sdma_v4_0_sw_init(void *handle)
 		DRM_INFO("use_doorbell being set to: [%s]\n",
 				ring->use_doorbell?"true":"false");
 
-		ring->doorbell_index = (i == 0) ?
-			(AMDGPU_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
-			: (AMDGPU_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
+		if (adev->asic_type == CHIP_VEGA10)
+			ring->doorbell_index = (i == 0) ?
+				(AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
+				: (AMDGPU_VEGA10_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
+		else
+			ring->doorbell_index = (i == 0) ?
+				(AMDGPU_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
+				: (AMDGPU_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
+
 
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024,
@@ -1801,10 +1807,8 @@ static const struct amdgpu_buffer_funcs sdma_v4_0_buffer_funcs = {
 
 static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs sdma_v4_0_vm_pte_funcs = {
@@ -1820,15 +1824,13 @@ static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
-			adev->vm_manager.vm_pte_rqs[i] =
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
-		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_rqs[i] =
+			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
+	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v4_0_ip_block = {
diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c
index fafaf259b17b..d4ceaf440f26 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_dma.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c
@@ -502,12 +502,12 @@ static int si_dma_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* DMA0 trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224, &adev->sdma.trap_irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 224, &adev->sdma.trap_irq);
 	if (r)
 		return r;
 
 	/* DMA1 trap event */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 244, &adev->sdma.trap_irq_1);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 244, &adev->sdma.trap_irq_1);
 	if (r)
 		return r;
 
@@ -863,10 +863,8 @@ static const struct amdgpu_buffer_funcs si_dma_buffer_funcs = {
 
 static void si_dma_set_buffer_funcs(struct amdgpu_device *adev)
 {
-	if (adev->mman.buffer_funcs == NULL) {
-		adev->mman.buffer_funcs = &si_dma_buffer_funcs;
-		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
-	}
+	adev->mman.buffer_funcs = &si_dma_buffer_funcs;
+	adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
 }
 
 static const struct amdgpu_vm_pte_funcs si_dma_vm_pte_funcs = {
@@ -882,15 +880,13 @@ static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev)
 	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
-	if (adev->vm_manager.vm_pte_funcs == NULL) {
-		adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
-		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
-			adev->vm_manager.vm_pte_rqs[i] =
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
-		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_rqs[i] =
+			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
+	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version si_dma_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/si_dpm.c b/drivers/gpu/drm/amd/amdgpu/si_dpm.c
index 1de96995e690..da58040fdbdc 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_dpm.c
@@ -7687,11 +7687,11 @@ static int si_dpm_sw_init(void *handle)
 	int ret;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230, &adev->pm.dpm.thermal.irq);
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 230, &adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
 
-	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231, &adev->pm.dpm.thermal.irq);
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 231, &adev->pm.dpm.thermal.irq);
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c
index 60dad63098a2..b3d7d9f83202 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c
@@ -142,7 +142,7 @@ static void si_ih_decode_iv(struct amdgpu_device *adev,
 	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
 	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
@@ -170,7 +170,7 @@ static int si_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, false);
 	if (r)
 		return r;
 
@@ -182,7 +182,7 @@ static int si_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 
 	return 0;
 }
@@ -308,8 +308,7 @@ static const struct amdgpu_ih_funcs si_ih_funcs = {
 
 static void si_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &si_ih_funcs;
+	adev->irq.ih_funcs = &si_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version si_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 138c4810a3de..fb26039beeba 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -739,7 +739,8 @@ static int soc15_common_early_init(void *handle)
 
 			adev->pg_flags = AMD_PG_SUPPORT_SDMA |
 				AMD_PG_SUPPORT_MMHUB |
-				AMD_PG_SUPPORT_VCN;
+				AMD_PG_SUPPORT_VCN |
+				AMD_PG_SUPPORT_VCN_DPG;
 		} else {
 			adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
 				AMD_CG_SUPPORT_GFX_MGLS |
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h b/drivers/gpu/drm/amd/amdgpu/soc15_common.h
index f5d602540673..958b10a57073 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15_common.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15_common.h
@@ -57,13 +57,33 @@
 			loop--;					\
 			if (!loop) {				\
 				DRM_ERROR("Register(%d) [%s] failed to reach value 0x%08x != 0x%08x\n", \
-							inst, #reg, expected_value, (tmp_ & (mask))); \
+					  inst, #reg, (unsigned)expected_value, (unsigned)(tmp_ & (mask))); \
 				ret = -ETIMEDOUT;		\
 				break;				\
 			}					\
 		}						\
 	} while (0)
 
+#define RREG32_SOC15_DPG_MODE(ip, inst, reg, mask, sram_sel) 	\
+		({ WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_MASK, mask); \
+			WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_CTL,	\
+				UVD_DPG_LMA_CTL__MASK_EN_MASK |				\
+				((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) \
+				<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) | \
+				(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT));	\
+			RREG32_SOC15(ip, inst, mmUVD_DPG_LMA_DATA); })
+
+#define WREG32_SOC15_DPG_MODE(ip, inst, reg, value, mask, sram_sel)	\
+	do {							\
+		WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_DATA, value);	\
+		WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_MASK, mask);		\
+		WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_CTL,	\
+			UVD_DPG_LMA_CTL__READ_WRITE_MASK |	\
+			((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) \
+			<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) |	\
+			(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); \
+	} while (0)
+
 #endif
 
 
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
index 52853d8a8fdd..3abffd06b5c7 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
@@ -266,7 +266,7 @@ static void tonga_ih_decode_iv(struct amdgpu_device *adev,
 	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
 	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
 
-	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 	entry->src_id = dw[0] & 0xff;
 	entry->src_data[0] = dw[1] & 0xfffffff;
 	entry->ring_id = dw[2] & 0xff;
@@ -317,7 +317,7 @@ static int tonga_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 64 * 1024, true);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 64 * 1024, true);
 	if (r)
 		return r;
 
@@ -334,7 +334,7 @@ static int tonga_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 	amdgpu_irq_remove_domain(adev);
 
 	return 0;
@@ -513,8 +513,7 @@ static const struct amdgpu_ih_funcs tonga_ih_funcs = {
 
 static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &tonga_ih_funcs;
+	adev->irq.ih_funcs = &tonga_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version tonga_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
index 8a926d1df939..1fc17bf39fed 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -108,7 +108,7 @@ static int uvd_v4_2_sw_init(void *handle)
 	int r;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
index 50248059412e..fde6ad5ac9ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -105,7 +105,7 @@ static int uvd_v5_0_sw_init(void *handle)
 	int r;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
index 6ae82cc2e55e..8ef4a5392112 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -393,14 +393,14 @@ static int uvd_v6_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* UVD TRAP */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
 	if (r)
 		return r;
 
 	/* UVD ENC TRAP */
 	if (uvd_v6_0_enc_support(adev)) {
 		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
-			r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + VISLANDS30_IV_SRCID_UVD_ENC_GEN_PURP, &adev->uvd.inst->irq);
+			r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + VISLANDS30_IV_SRCID_UVD_ENC_GEN_PURP, &adev->uvd.inst->irq);
 			if (r)
 				return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
index 7eaa54ba016b..ea28828360d3 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
@@ -417,7 +417,7 @@ static int vce_v2_0_sw_init(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* VCE */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 167, &adev->vce.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 167, &adev->vce.irq);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
index c8390f9adfd6..6dbd39730070 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -423,7 +423,7 @@ static int vce_v3_0_sw_init(void *handle)
 	int r, i;
 
 	/* VCE */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_VCE_TRAP, &adev->vce.irq);
+	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_VCE_TRAP, &adev->vce.irq);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
index 2664bb2c47c3..63d7f97e81b7 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
@@ -198,7 +198,8 @@ static int vcn_v1_0_hw_init(void *handle)
 
 done:
 	if (!r)
-		DRM_INFO("VCN decode and encode initialized successfully.\n");
+		DRM_INFO("VCN decode and encode initialized successfully(under %s).\n",
+			(adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)?"DPG Mode":"SPG Mode");
 
 	return r;
 }
@@ -266,13 +267,13 @@ static int vcn_v1_0_resume(void *handle)
 }
 
 /**
- * vcn_v1_0_mc_resume - memory controller programming
+ * vcn_v1_0_mc_resume_spg_mode - memory controller programming
  *
  * @adev: amdgpu_device pointer
  *
  * Let the VCN memory controller know it's offsets
  */
-static void vcn_v1_0_mc_resume(struct amdgpu_device *adev)
+static void vcn_v1_0_mc_resume_spg_mode(struct amdgpu_device *adev)
 {
 	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
 	uint32_t offset;
@@ -319,6 +320,65 @@ static void vcn_v1_0_mc_resume(struct amdgpu_device *adev)
 			adev->gfx.config.gb_addr_config);
 }
 
+static void vcn_v1_0_mc_resume_dpg_mode(struct amdgpu_device *adev)
+{
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
+	uint32_t offset;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo),
+			     0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi),
+			     0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, 0,
+			     0xFFFFFFFF, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.gpu_addr), 0xFFFFFFFF, 0);
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.gpu_addr), 0xFFFFFFFF, 0);
+		offset = size;
+		WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0xFFFFFFFF, 0);
+	}
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size, 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.gpu_addr + offset), 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.gpu_addr + offset), 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, 0,
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_HEAP_SIZE,
+			     0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.gpu_addr + offset + AMDGPU_VCN_HEAP_SIZE),
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.gpu_addr + offset + AMDGPU_VCN_HEAP_SIZE),
+			     0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CACHE_SIZE2,
+			AMDGPU_VCN_STACK_SIZE + (AMDGPU_VCN_SESSION_SIZE * 40),
+			     0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_UDEC_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_JPEG_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_JPEG_UV_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config, 0xFFFFFFFF, 0);
+}
+
 /**
  * vcn_v1_0_disable_clock_gating - disable VCN clock gating
  *
@@ -519,6 +579,62 @@ static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev)
 	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
 }
 
+static void vcn_v1_0_clock_gating_dpg_mode(struct amdgpu_device *adev, uint8_t sram_sel)
+{
+	uint32_t reg_data = 0;
+
+	/* disable JPEG CGC */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmJPEG_CGC_CTRL, reg_data, 0xFFFFFFFF, sram_sel);
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmJPEG_CGC_GATE, 0, 0xFFFFFFFF, sram_sel);
+
+	/* enable sw clock gating control */
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		reg_data = 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		reg_data = 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	reg_data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	reg_data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_CGC_CTRL, reg_data, 0xFFFFFFFF, sram_sel);
+
+	reg_data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+		 UVD_CGC_CTRL__SYS_MODE_MASK |
+		 UVD_CGC_CTRL__UDEC_MODE_MASK |
+		 UVD_CGC_CTRL__MPEG2_MODE_MASK |
+		 UVD_CGC_CTRL__REGS_MODE_MASK |
+		 UVD_CGC_CTRL__RBC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+		 UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+		 UVD_CGC_CTRL__IDCT_MODE_MASK |
+		 UVD_CGC_CTRL__MPRD_MODE_MASK |
+		 UVD_CGC_CTRL__MPC_MODE_MASK |
+		 UVD_CGC_CTRL__LBSI_MODE_MASK |
+		 UVD_CGC_CTRL__LRBBM_MODE_MASK |
+		 UVD_CGC_CTRL__WCB_MODE_MASK |
+		 UVD_CGC_CTRL__VCPU_MODE_MASK |
+		 UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_CGC_CTRL, reg_data, 0xFFFFFFFF, sram_sel);
+
+	/* turn off clock gating */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_CGC_GATE, 0, 0xFFFFFFFF, sram_sel);
+
+	/* turn on SUVD clock gating */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SUVD_CGC_GATE, 1, 0xFFFFFFFF, sram_sel);
+
+	/* turn on sw mode in UVD_SUVD_CGC_CTRL */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SUVD_CGC_CTRL, 0, 0xFFFFFFFF, sram_sel);
+}
+
 static void vcn_1_0_disable_static_power_gating(struct amdgpu_device *adev)
 {
 	uint32_t data = 0;
@@ -614,7 +730,7 @@ static void vcn_1_0_enable_static_power_gating(struct amdgpu_device *adev)
  *
  * Setup and start the VCN block
  */
-static int vcn_v1_0_start(struct amdgpu_device *adev)
+static int vcn_v1_0_start_spg_mode(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
 	uint32_t rb_bufsz, tmp;
@@ -628,7 +744,7 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	/* disable clock gating */
 	vcn_v1_0_disable_clock_gating(adev);
 
-	vcn_v1_0_mc_resume(adev);
+	vcn_v1_0_mc_resume_spg_mode(adev);
 
 	/* disable interupt */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0,
@@ -799,6 +915,170 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int vcn_v1_0_start_dpg_mode(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+	uint32_t rb_bufsz, tmp, reg_data;
+	uint32_t lmi_swap_cntl;
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+
+	vcn_1_0_enable_static_power_gating(adev);
+
+	/* enable dynamic power gating mode */
+	reg_data = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+	reg_data |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+	reg_data |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+	WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, reg_data);
+
+	/* enable clock gating */
+	vcn_v1_0_clock_gating_dpg_mode(adev, 0);
+
+	/* enable VCPU clock */
+	reg_data = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+	reg_data |= UVD_VCPU_CNTL__CLK_EN_MASK;
+	reg_data |= UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP_MASK;
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CNTL, reg_data, 0xFFFFFFFF, 0);
+
+	/* disable interupt */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MASTINT_EN,
+			0, UVD_MASTINT_EN__VCPU_EN_MASK, 0);
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_CTRL2,
+			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SOFT_RESET,
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK,
+		0xFFFFFFFF, 0);
+
+	/* initialize VCN memory controller */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_CTRL,
+		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		0x00100000L, 0xFFFFFFFF, 0);
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+#endif
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl, 0xFFFFFFFF, 0);
+
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_MUXA0, 0x40c2040, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_MUXA1, 0x0, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_MUXB0, 0x40c2040, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_MUXB1, 0x0, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_ALU, 0, 0xFFFFFFFF, 0);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MPC_SET_MUX, 0x88, 0xFFFFFFFF, 0);
+
+	vcn_v1_0_mc_resume_dpg_mode(adev);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SOFT_RESET,
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK, 0xFFFFFFFF, 0);
+
+	/* enable VCPU clock */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_VCPU_CNTL,
+			UVD_VCPU_CNTL__CLK_EN_MASK, 0xFFFFFFFF, 0);
+
+	/* enable UMC */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_CTRL2,
+			0, UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
+
+	/* boot up the VCPU */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SOFT_RESET, 0, 0xFFFFFFFF, 0);
+
+	/* enable master interrupt */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_MASTINT_EN,
+			(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+			(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK), 0);
+
+	vcn_v1_0_clock_gating_dpg_mode(adev, 1);
+	/* setup mmUVD_LMI_CTRL */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_LMI_CTRL,
+			(UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+				UVD_LMI_CTRL__CRC_RESET_MASK |
+				UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+				UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+				UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+				(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+				0x00100000L), 0xFFFFFFFF, 1);
+
+	tmp = adev->gfx.config.gb_addr_config;
+	/* setup VCN global tiling registers */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_JPEG_ADDR_CONFIG, tmp, 0xFFFFFFFF, 1);
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_JPEG_UV_ADDR_CONFIG, tmp, 0xFFFFFFFF, 1);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_SOC15_DPG_MODE(UVD, 0, mmUVD_SYS_INT_EN,
+									UVD_SYS_INT_EN__UVD_JRBC_EN_MASK, 0xFFFFFFFF, 1);
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
+								(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+								lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+								upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+								lower_32_bits(ring->wptr));
+
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0,
+			~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	/* initialize wptr */
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
+
+	/* copy patch commands to the jpeg ring */
+	vcn_v1_0_jpeg_ring_set_patch_ring(ring,
+		(ring->wptr + ring->max_dw * amdgpu_sched_hw_submission));
+
+	return 0;
+}
+
+static int vcn_v1_0_start(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		r = vcn_v1_0_start_dpg_mode(adev);
+	else
+		r = vcn_v1_0_start_spg_mode(adev);
+	return r;
+}
+
 /**
  * vcn_v1_0_stop - stop VCN block
  *
@@ -806,7 +1086,7 @@ static int vcn_v1_0_start(struct amdgpu_device *adev)
  *
  * stop the VCN block
  */
-static int vcn_v1_0_stop(struct amdgpu_device *adev)
+static int vcn_v1_0_stop_spg_mode(struct amdgpu_device *adev)
 {
 	/* force RBC into idle state */
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, 0x11010101);
@@ -836,6 +1116,33 @@ static int vcn_v1_0_stop(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int vcn_v1_0_stop_dpg_mode(struct amdgpu_device *adev)
+{
+	int ret_code;
+
+	/* Wait for power status to be 1 */
+	SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 0x1,
+			UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+
+	/* disable dynamic power gating mode */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0,
+			~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	return 0;
+}
+
+static int vcn_v1_0_stop(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+		r = vcn_v1_0_stop_dpg_mode(adev);
+	else
+		r = vcn_v1_0_stop_spg_mode(adev);
+
+	return r;
+}
+
 static bool vcn_v1_0_is_idle(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1633,12 +1940,20 @@ static int vcn_v1_0_set_powergating_state(void *handle,
 	 * revisit this when there is a cleaner line between
 	 * the smc and the hw blocks
 	 */
+	int ret;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
+	if(state == adev->vcn.cur_state)
+		return 0;
+
 	if (state == AMD_PG_STATE_GATE)
-		return vcn_v1_0_stop(adev);
+		ret = vcn_v1_0_stop(adev);
 	else
-		return vcn_v1_0_start(adev);
+		ret = vcn_v1_0_start(adev);
+
+	if(!ret)
+		adev->vcn.cur_state = state;
+	return ret;
 }
 
 static const struct amd_ip_funcs vcn_v1_0_ip_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
index acbe5a770207..a99f71797aa3 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
@@ -380,7 +380,7 @@ static int vega10_ih_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	r = amdgpu_ih_ring_init(adev, 256 * 1024, true);
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, true);
 	if (r)
 		return r;
 
@@ -397,7 +397,7 @@ static int vega10_ih_sw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_irq_fini(adev);
-	amdgpu_ih_ring_fini(adev);
+	amdgpu_ih_ring_fini(adev, &adev->irq.ih);
 
 	return 0;
 }
@@ -494,8 +494,7 @@ static const struct amdgpu_ih_funcs vega10_ih_funcs = {
 
 static void vega10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
 {
-	if (adev->irq.ih_funcs == NULL)
-		adev->irq.ih_funcs = &vega10_ih_funcs;
+	adev->irq.ih_funcs = &vega10_ih_funcs;
 }
 
 const struct amdgpu_ip_block_version vega10_ih_ip_block =
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 758398bdb39b..14d5b5fa822d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -447,6 +447,24 @@ static int kfd_ioctl_set_cu_mask(struct file *filp, struct kfd_process *p,
 	return retval;
 }
 
+static int kfd_ioctl_get_queue_wave_state(struct file *filep,
+					  struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_queue_wave_state_args *args = data;
+	int r;
+
+	mutex_lock(&p->mutex);
+
+	r = pqm_get_wave_state(&p->pqm, args->queue_id,
+			       (void __user *)args->ctl_stack_address,
+			       &args->ctl_stack_used_size,
+			       &args->save_area_used_size);
+
+	mutex_unlock(&p->mutex);
+
+	return r;
+}
+
 static int kfd_ioctl_set_memory_policy(struct file *filep,
 					struct kfd_process *p, void *data)
 {
@@ -1615,6 +1633,9 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_CU_MASK,
 			kfd_ioctl_set_cu_mask, 0),
 
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_QUEUE_WAVE_STATE,
+			kfd_ioctl_get_queue_wave_state, 0)
+
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
index d4560f1869bd..56412b0e7e1c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@@ -647,6 +647,7 @@ static int kfd_fill_gpu_cache_info(struct kfd_dev *kdev,
 		num_of_cache_types = ARRAY_SIZE(polaris11_cache_info);
 		break;
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 		pcache_info = vega10_cache_info;
 		num_of_cache_types = ARRAY_SIZE(vega10_cache_info);
 		break;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 9b4e6ad4a7df..a9f18ea7e354 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -53,6 +53,7 @@ static const struct kfd_device_info kaveri_device_info = {
 	.needs_iommu_device = true,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info carrizo_device_info = {
@@ -69,6 +70,7 @@ static const struct kfd_device_info carrizo_device_info = {
 	.needs_iommu_device = true,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info raven_device_info = {
@@ -84,6 +86,7 @@ static const struct kfd_device_info raven_device_info = {
 	.needs_iommu_device = true,
 	.needs_pci_atomics = true,
 	.num_sdma_engines = 1,
+	.num_sdma_queues_per_engine = 2,
 };
 #endif
 
@@ -101,6 +104,7 @@ static const struct kfd_device_info hawaii_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info tonga_device_info = {
@@ -116,21 +120,7 @@ static const struct kfd_device_info tonga_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = true,
 	.num_sdma_engines = 2,
-};
-
-static const struct kfd_device_info tonga_vf_device_info = {
-	.asic_family = CHIP_TONGA,
-	.max_pasid_bits = 16,
-	.max_no_of_hqd  = 24,
-	.doorbell_size  = 4,
-	.ih_ring_entry_size = 4 * sizeof(uint32_t),
-	.event_interrupt_class = &event_interrupt_class_cik,
-	.num_of_watch_points = 4,
-	.mqd_size_aligned = MQD_SIZE_ALIGNED,
-	.supports_cwsr = false,
-	.needs_iommu_device = false,
-	.needs_pci_atomics = false,
-	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info fiji_device_info = {
@@ -146,6 +136,7 @@ static const struct kfd_device_info fiji_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = true,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info fiji_vf_device_info = {
@@ -161,6 +152,7 @@ static const struct kfd_device_info fiji_vf_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 
@@ -177,6 +169,7 @@ static const struct kfd_device_info polaris10_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = true,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info polaris10_vf_device_info = {
@@ -192,6 +185,7 @@ static const struct kfd_device_info polaris10_vf_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info polaris11_device_info = {
@@ -207,6 +201,7 @@ static const struct kfd_device_info polaris11_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = true,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info vega10_device_info = {
@@ -222,6 +217,7 @@ static const struct kfd_device_info vega10_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
 static const struct kfd_device_info vega10_vf_device_info = {
@@ -237,8 +233,24 @@ static const struct kfd_device_info vega10_vf_device_info = {
 	.needs_iommu_device = false,
 	.needs_pci_atomics = false,
 	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 2,
 };
 
+static const struct kfd_device_info vega20_device_info = {
+	.asic_family = CHIP_VEGA20,
+	.max_pasid_bits = 16,
+	.max_no_of_hqd	= 24,
+	.doorbell_size	= 8,
+	.ih_ring_entry_size = 8 * sizeof(uint32_t),
+	.event_interrupt_class = &event_interrupt_class_v9,
+	.num_of_watch_points = 4,
+	.mqd_size_aligned = MQD_SIZE_ALIGNED,
+	.supports_cwsr = true,
+	.needs_iommu_device = false,
+	.needs_pci_atomics = false,
+	.num_sdma_engines = 2,
+	.num_sdma_queues_per_engine = 8,
+};
 
 struct kfd_deviceid {
 	unsigned short did;
@@ -293,7 +305,6 @@ static const struct kfd_deviceid supported_devices[] = {
 	{ 0x6928, &tonga_device_info },		/* Tonga */
 	{ 0x6929, &tonga_device_info },		/* Tonga */
 	{ 0x692B, &tonga_device_info },		/* Tonga */
-	{ 0x692F, &tonga_vf_device_info },	/* Tonga vf */
 	{ 0x6938, &tonga_device_info },		/* Tonga */
 	{ 0x6939, &tonga_device_info },		/* Tonga */
 	{ 0x7300, &fiji_device_info },		/* Fiji */
@@ -328,6 +339,12 @@ static const struct kfd_deviceid supported_devices[] = {
 	{ 0x6868, &vega10_device_info },	/* Vega10 */
 	{ 0x686C, &vega10_vf_device_info },	/* Vega10  vf*/
 	{ 0x687F, &vega10_device_info },	/* Vega10 */
+	{ 0x66a0, &vega20_device_info },	/* Vega20 */
+	{ 0x66a1, &vega20_device_info },	/* Vega20 */
+	{ 0x66a2, &vega20_device_info },	/* Vega20 */
+	{ 0x66a3, &vega20_device_info },	/* Vega20 */
+	{ 0x66a7, &vega20_device_info },	/* Vega20 */
+	{ 0x66af, &vega20_device_info }		/* Vega20 */
 };
 
 static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
@@ -366,6 +383,10 @@ struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
 		return NULL;
 	}
 
+	kfd = kzalloc(sizeof(*kfd), GFP_KERNEL);
+	if (!kfd)
+		return NULL;
+
 	/* Allow BIF to recode atomics to PCIe 3.0 AtomicOps.
 	 * 32 and 64-bit requests are possible and must be
 	 * supported.
@@ -377,12 +398,10 @@ struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
 		dev_info(kfd_device,
 			 "skipped device %x:%x, PCI rejects atomics\n",
 			 pdev->vendor, pdev->device);
+		kfree(kfd);
 		return NULL;
-	}
-
-	kfd = kzalloc(sizeof(*kfd), GFP_KERNEL);
-	if (!kfd)
-		return NULL;
+	} else if (!ret)
+		kfd->pci_atomic_requested = true;
 
 	kfd->kgd = kgd;
 	kfd->device_info = device_info;
@@ -419,6 +438,10 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 {
 	unsigned int size;
 
+	kfd->mec_fw_version = kfd->kfd2kgd->get_fw_version(kfd->kgd,
+			KGD_ENGINE_MEC1);
+	kfd->sdma_fw_version = kfd->kfd2kgd->get_fw_version(kfd->kgd,
+			KGD_ENGINE_SDMA1);
 	kfd->shared_resources = *gpu_resources;
 
 	kfd->vm_info.first_vmid_kfd = ffs(gpu_resources->compute_vmid_bitmap)-1;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index ec0d62a16e53..d6af31ccf0ee 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -109,7 +109,7 @@ static unsigned int get_num_sdma_engines(struct device_queue_manager *dqm)
 unsigned int get_num_sdma_queues(struct device_queue_manager *dqm)
 {
 	return dqm->dev->device_info->num_sdma_engines
-			* KFD_SDMA_QUEUES_PER_ENGINE;
+			* dqm->dev->device_info->num_sdma_queues_per_engine;
 }
 
 void program_sh_mem_settings(struct device_queue_manager *dqm,
@@ -656,7 +656,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 	struct queue *q;
 	struct mqd_manager *mqd_mgr;
 	struct kfd_process_device *pdd;
-	uint32_t pd_base;
+	uint64_t pd_base;
 	int retval = 0;
 
 	pdd = qpd_to_pdd(qpd);
@@ -676,7 +676,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 
 	/* Update PD Base in QPD */
 	qpd->page_table_base = pd_base;
-	pr_debug("Updated PD address to 0x%08x\n", pd_base);
+	pr_debug("Updated PD address to 0x%llx\n", pd_base);
 
 	if (!list_empty(&qpd->queues_list)) {
 		dqm->dev->kfd2kgd->set_vm_context_page_table_base(
@@ -717,7 +717,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
 {
 	struct queue *q;
 	struct kfd_process_device *pdd;
-	uint32_t pd_base;
+	uint64_t pd_base;
 	int retval = 0;
 
 	pdd = qpd_to_pdd(qpd);
@@ -737,7 +737,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
 
 	/* Update PD Base in QPD */
 	qpd->page_table_base = pd_base;
-	pr_debug("Updated PD address to 0x%08x\n", pd_base);
+	pr_debug("Updated PD address to 0x%llx\n", pd_base);
 
 	/* activate all active queues on the qpd */
 	list_for_each_entry(q, &qpd->queues_list, list) {
@@ -761,7 +761,7 @@ static int register_process(struct device_queue_manager *dqm,
 {
 	struct device_process_node *n;
 	struct kfd_process_device *pdd;
-	uint32_t pd_base;
+	uint64_t pd_base;
 	int retval;
 
 	n = kzalloc(sizeof(*n), GFP_KERNEL);
@@ -779,6 +779,7 @@ static int register_process(struct device_queue_manager *dqm,
 
 	/* Update PD Base in QPD */
 	qpd->page_table_base = pd_base;
+	pr_debug("Updated PD address to 0x%llx\n", pd_base);
 
 	retval = dqm->asic_ops.update_qpd(dqm, qpd);
 
@@ -1528,6 +1529,41 @@ static int process_termination_nocpsch(struct device_queue_manager *dqm,
 	return retval;
 }
 
+static int get_wave_state(struct device_queue_manager *dqm,
+			  struct queue *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct mqd_manager *mqd;
+	int r;
+
+	dqm_lock(dqm);
+
+	if (q->properties.type != KFD_QUEUE_TYPE_COMPUTE ||
+	    q->properties.is_active || !q->device->cwsr_enabled) {
+		r = -EINVAL;
+		goto dqm_unlock;
+	}
+
+	mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_COMPUTE);
+	if (!mqd) {
+		r = -ENOMEM;
+		goto dqm_unlock;
+	}
+
+	if (!mqd->get_wave_state) {
+		r = -EINVAL;
+		goto dqm_unlock;
+	}
+
+	r = mqd->get_wave_state(mqd, q->mqd, ctl_stack, ctl_stack_used_size,
+				save_area_used_size);
+
+dqm_unlock:
+	dqm_unlock(dqm);
+	return r;
+}
 
 static int process_termination_cpsch(struct device_queue_manager *dqm,
 		struct qcm_process_device *qpd)
@@ -1649,6 +1685,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.process_termination = process_termination_cpsch;
 		dqm->ops.evict_process_queues = evict_process_queues_cpsch;
 		dqm->ops.restore_process_queues = restore_process_queues_cpsch;
+		dqm->ops.get_wave_state = get_wave_state;
 		break;
 	case KFD_SCHED_POLICY_NO_HWS:
 		/* initialize dqm for no cp scheduling */
@@ -1668,6 +1705,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.evict_process_queues = evict_process_queues_nocpsch;
 		dqm->ops.restore_process_queues =
 			restore_process_queues_nocpsch;
+		dqm->ops.get_wave_state = get_wave_state;
 		break;
 	default:
 		pr_err("Invalid scheduling policy %d\n", dqm->sched_policy);
@@ -1695,6 +1733,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		break;
 
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		device_queue_manager_init_v9(&dqm->asic_ops);
 		break;
@@ -1806,7 +1845,9 @@ int dqm_debugfs_hqds(struct seq_file *m, void *data)
 	}
 
 	for (pipe = 0; pipe < get_num_sdma_engines(dqm); pipe++) {
-		for (queue = 0; queue < KFD_SDMA_QUEUES_PER_ENGINE; queue++) {
+		for (queue = 0;
+		     queue < dqm->dev->device_info->num_sdma_queues_per_engine;
+		     queue++) {
 			r = dqm->dev->kfd2kgd->hqd_sdma_dump(
 				dqm->dev->kgd, pipe, queue, &dump, &n_regs);
 			if (r)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
index 00da3169a004..70e38a2e23b9 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
@@ -33,7 +33,6 @@
 
 #define KFD_UNMAP_LATENCY_MS			(4000)
 #define QUEUE_PREEMPT_DEFAULT_TIMEOUT_MS (2 * KFD_UNMAP_LATENCY_MS + 1000)
-#define KFD_SDMA_QUEUES_PER_ENGINE		(2)
 
 struct device_process_node {
 	struct qcm_process_device *qpd;
@@ -82,6 +81,8 @@ struct device_process_node {
  *
  * @restore_process_queues: Restore all evicted queues queues of a process
  *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
  */
 
 struct device_queue_manager_ops {
@@ -137,6 +138,12 @@ struct device_queue_manager_ops {
 				    struct qcm_process_device *qpd);
 	int (*restore_process_queues)(struct device_queue_manager *dqm,
 				      struct qcm_process_device *qpd);
+
+	int	(*get_wave_state)(struct device_queue_manager *dqm,
+				  struct queue *q,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
 };
 
 struct device_queue_manager_asic_ops {
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
index 97d5423c5673..3d66cec414af 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
@@ -400,6 +400,7 @@ int kfd_init_apertures(struct kfd_process *process)
 				kfd_init_apertures_vi(pdd, id);
 				break;
 			case CHIP_VEGA10:
+			case CHIP_VEGA20:
 			case CHIP_RAVEN:
 				kfd_init_apertures_v9(pdd, id);
 				break;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
index 9f84b4d9fb88..6c31f7370193 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
@@ -322,6 +322,7 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
 		break;
 
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		kernel_queue_init_v9(&kq->ops_asic_specific);
 		break;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_v9.c
index 684a3bf07efd..33830b1a5a54 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue_v9.c
@@ -71,8 +71,7 @@ static int pm_map_process_v9(struct packet_manager *pm,
 		uint32_t *buffer, struct qcm_process_device *qpd)
 {
 	struct pm4_mes_map_process *packet;
-	uint64_t vm_page_table_base_addr =
-		(uint64_t)(qpd->page_table_base) << 12;
+	uint64_t vm_page_table_base_addr = qpd->page_table_base;
 
 	packet = (struct pm4_mes_map_process *)buffer;
 	memset(buffer, 0, sizeof(struct pm4_mes_map_process));
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
index 3bc25ab84f34..e33019a7a883 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
@@ -39,6 +39,7 @@ struct mqd_manager *mqd_manager_init(enum KFD_MQD_TYPE type,
 	case CHIP_POLARIS11:
 		return mqd_manager_init_vi_tonga(type, dev);
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		return mqd_manager_init_v9(type, dev);
 	default:
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
index 4e84052d4e21..f8261313ae7b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
@@ -43,6 +43,9 @@
  *
  * @is_occupied: Checks if the relevant HQD slot is occupied.
  *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
+ *
  * @mqd_mutex: Mqd manager mutex.
  *
  * @dev: The kfd device structure coupled with this module.
@@ -85,6 +88,11 @@ struct mqd_manager {
 				uint64_t queue_address,	uint32_t pipe_id,
 				uint32_t queue_id);
 
+	int	(*get_wave_state)(struct mqd_manager *mm, void *mqd,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
+
 #if defined(CONFIG_DEBUG_FS)
 	int	(*debugfs_show_mqd)(struct seq_file *m, void *data);
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
index 0cedb37cf513..f381c1cb27bd 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
@@ -266,6 +266,28 @@ static bool is_occupied(struct mqd_manager *mm, void *mqd,
 		pipe_id, queue_id);
 }
 
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v9_mqd *m;
+
+	/* Control stack is located one page after MQD. */
+	void *mqd_ctl_stack = (void *)((uintptr_t)mqd + PAGE_SIZE);
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset;
+
+	if (copy_to_user(ctl_stack, mqd_ctl_stack, m->cp_hqd_cntl_stack_size))
+		return -EFAULT;
+
+	return 0;
+}
+
 static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
 			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
@@ -435,6 +457,7 @@ struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
 		mqd->update_mqd = update_mqd;
 		mqd->destroy_mqd = destroy_mqd;
 		mqd->is_occupied = is_occupied;
+		mqd->get_wave_state = get_wave_state;
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
index b81fda3754da..6469b3456f00 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
@@ -269,6 +269,28 @@ static bool is_occupied(struct mqd_manager *mm, void *mqd,
 		pipe_id, queue_id);
 }
 
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct vi_mqd *m;
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv8 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+
+	return 0;
+}
+
 static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
 			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
@@ -436,6 +458,7 @@ struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
 		mqd->update_mqd = update_mqd;
 		mqd->destroy_mqd = destroy_mqd;
 		mqd->is_occupied = is_occupied;
+		mqd->get_wave_state = get_wave_state;
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
index 1092631765cb..c6080ed3b6a7 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
@@ -229,6 +229,7 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)
 		pm->pmf = &kfd_vi_pm_funcs;
 		break;
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		pm->pmf = &kfd_v9_pm_funcs;
 		break;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index b0064b08aa11..53ff86d45d91 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -176,6 +176,7 @@ struct kfd_device_info {
 	bool needs_iommu_device;
 	bool needs_pci_atomics;
 	unsigned int num_sdma_engines;
+	unsigned int num_sdma_queues_per_engine;
 };
 
 struct kfd_mem_obj {
@@ -247,6 +248,10 @@ struct kfd_dev {
 	/* Debug manager */
 	struct kfd_dbgmgr           *dbgmgr;
 
+	/* Firmware versions */
+	uint16_t mec_fw_version;
+	uint16_t sdma_fw_version;
+
 	/* Maximum process number mapped to HW scheduler */
 	unsigned int max_proc_per_quantum;
 
@@ -257,6 +262,8 @@ struct kfd_dev {
 
 	/* xGMI */
 	uint64_t hive_id;
+
+	bool pci_atomic_requested;
 };
 
 /* KGD2KFD callbacks */
@@ -500,11 +507,11 @@ struct qcm_process_device {
 	 * All the memory management data should be here too
 	 */
 	uint64_t gds_context_area;
+	uint64_t page_table_base;
 	uint32_t sh_mem_config;
 	uint32_t sh_mem_bases;
 	uint32_t sh_mem_ape1_base;
 	uint32_t sh_mem_ape1_limit;
-	uint32_t page_table_base;
 	uint32_t gds_size;
 	uint32_t num_gws;
 	uint32_t num_oac;
@@ -856,6 +863,11 @@ int pqm_set_cu_mask(struct process_queue_manager *pqm, unsigned int qid,
 			struct queue_properties *p);
 struct kernel_queue *pqm_get_kernel_queue(struct process_queue_manager *pqm,
 						unsigned int qid);
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size);
 
 int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
 				unsigned int fence_value,
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
index c8cad9c078ae..fcaaf93681ac 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
@@ -408,6 +408,28 @@ struct kernel_queue *pqm_get_kernel_queue(
 	return NULL;
 }
 
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("amdkfd: No queue %d exists for operation\n",
+			 qid);
+		return -EFAULT;
+	}
+
+	return pqn->q->device->dqm->ops.get_wave_state(pqn->q->device->dqm,
+						       pqn->q,
+						       ctl_stack,
+						       ctl_stack_used_size,
+						       save_area_used_size);
+}
+
 #if defined(CONFIG_DEBUG_FS)
 
 int pqm_debugfs_mqds(struct seq_file *m, void *data)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index 0dff66be8d7a..e3843c5929ed 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -482,11 +482,11 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 				(unsigned long long int) 0);
 
 		sysfs_show_32bit_prop(buffer, "fw_version",
-			dev->gpu->kfd2kgd->get_fw_version(
-						dev->gpu->kgd,
-						KGD_ENGINE_MEC1));
+				dev->gpu->mec_fw_version);
 		sysfs_show_32bit_prop(buffer, "capability",
 				dev->node_props.capability);
+		sysfs_show_32bit_prop(buffer, "sdma_fw_version",
+				dev->gpu->sdma_fw_version);
 	}
 
 	return sysfs_show_32bit_prop(buffer, "max_engine_clk_ccompute",
@@ -1127,17 +1127,40 @@ static void kfd_fill_mem_clk_max_info(struct kfd_topology_device *dev)
 
 static void kfd_fill_iolink_non_crat_info(struct kfd_topology_device *dev)
 {
-	struct kfd_iolink_properties *link;
+	struct kfd_iolink_properties *link, *cpu_link;
+	struct kfd_topology_device *cpu_dev;
+	uint32_t cap;
+	uint32_t cpu_flag = CRAT_IOLINK_FLAGS_ENABLED;
+	uint32_t flag = CRAT_IOLINK_FLAGS_ENABLED;
 
 	if (!dev || !dev->gpu)
 		return;
 
-	/* GPU only creates direck links so apply flags setting to all */
-	if (dev->gpu->device_info->asic_family == CHIP_HAWAII)
-		list_for_each_entry(link, &dev->io_link_props, list)
-			link->flags = CRAT_IOLINK_FLAGS_ENABLED |
-				CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT |
-				CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT;
+	pcie_capability_read_dword(dev->gpu->pdev,
+			PCI_EXP_DEVCAP2, &cap);
+
+	if (!(cap & (PCI_EXP_DEVCAP2_ATOMIC_COMP32 |
+		     PCI_EXP_DEVCAP2_ATOMIC_COMP64)))
+		cpu_flag |= CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT |
+			CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT;
+
+	if (!dev->gpu->pci_atomic_requested ||
+	    dev->gpu->device_info->asic_family == CHIP_HAWAII)
+		flag |= CRAT_IOLINK_FLAGS_NO_ATOMICS_32_BIT |
+			CRAT_IOLINK_FLAGS_NO_ATOMICS_64_BIT;
+
+	/* GPU only creates direct links so apply flags setting to all */
+	list_for_each_entry(link, &dev->io_link_props, list) {
+		link->flags = flag;
+		cpu_dev = kfd_topology_device_by_proximity_domain(
+				link->node_to);
+		if (cpu_dev) {
+			list_for_each_entry(cpu_link,
+					    &cpu_dev->io_link_props, list)
+				if (cpu_link->node_to == link->node_from)
+					cpu_link->flags = cpu_flag;
+		}
+	}
 }
 
 int kfd_topology_add_device(struct kfd_dev *gpu)
@@ -1255,6 +1278,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 			HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK);
 		break;
 	case CHIP_VEGA10:
+	case CHIP_VEGA20:
 	case CHIP_RAVEN:
 		dev->node_props.capability |= ((HSA_CAP_DOORBELL_TYPE_2_0 <<
 			HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT) &
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 6c849e055038..715422bb30db 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -338,14 +338,6 @@ static int dm_set_powergating_state(void *handle,
 /* Prototypes of private functions */
 static int dm_early_init(void* handle);
 
-static void hotplug_notify_work_func(struct work_struct *work)
-{
-	struct amdgpu_display_manager *dm = container_of(work, struct amdgpu_display_manager, mst_hotplug_work);
-	struct drm_device *dev = dm->ddev;
-
-	drm_kms_helper_hotplug_event(dev);
-}
-
 /* Allocate memory for FBC compressed data  */
 static void amdgpu_dm_fbc_init(struct drm_connector *connector)
 {
@@ -447,8 +439,6 @@ static int amdgpu_dm_init(struct amdgpu_device *adev)
 		goto error;
 	}
 
-	INIT_WORK(&adev->dm.mst_hotplug_work, hotplug_notify_work_func);
-
 	adev->dm.freesync_module = mod_freesync_create(adev->dm.dc);
 	if (!adev->dm.freesync_module) {
 		DRM_ERROR(
@@ -1214,7 +1204,7 @@ static int dce110_register_irq_handlers(struct amdgpu_device *adev)
 	struct dc_interrupt_params int_params = {0};
 	int r;
 	int i;
-	unsigned client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	unsigned client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
 
 	if (adev->asic_type == CHIP_VEGA10 ||
 	    adev->asic_type == CHIP_VEGA12 ||
@@ -4079,6 +4069,7 @@ static void amdgpu_dm_do_flip(struct drm_crtc *crtc,
 	/* TODO eliminate or rename surface_update */
 	struct dc_surface_update surface_updates[1] = { {0} };
 	struct dm_crtc_state *acrtc_state = to_dm_crtc_state(crtc->state);
+	struct dc_stream_status *stream_status;
 
 
 	/* Prepare wait for target vblank early - before the fence-waits */
@@ -4134,7 +4125,19 @@ static void amdgpu_dm_do_flip(struct drm_crtc *crtc,
 
 	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
 
-	surface_updates->surface = dc_stream_get_status(acrtc_state->stream)->plane_states[0];
+	stream_status = dc_stream_get_status(acrtc_state->stream);
+	if (!stream_status) {
+		DRM_ERROR("No stream status for CRTC: id=%d\n",
+			acrtc->crtc_id);
+		return;
+	}
+
+	surface_updates->surface = stream_status->plane_states[0];
+	if (!surface_updates->surface) {
+		DRM_ERROR("No surface for CRTC: id=%d\n",
+			acrtc->crtc_id);
+		return;
+	}
 	surface_updates->flip_addr = &addr;
 
 	dc_commit_updates_for_stream(adev->dm.dc,
@@ -4797,6 +4800,8 @@ void set_freesync_on_stream(struct amdgpu_display_manager *dm,
 	mod_freesync_build_vrr_infopacket(dm->freesync_module,
 					  new_stream,
 					  &vrr,
+					  packet_type_fs1,
+					  NULL,
 					  &vrr_infopacket);
 
 	new_crtc_state->adjust = vrr.adjust;
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
index d4f1bdf93207..978b34a5011c 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
@@ -108,8 +108,6 @@ struct amdgpu_display_manager {
 
 	const struct dc_link *backlight_link;
 
-	struct work_struct mst_hotplug_work;
-
 	struct mod_freesync *freesync_module;
 
 	/**
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
index 6d16b4a0353d..0fab64a2a915 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
@@ -105,6 +105,8 @@ bool dm_pp_apply_display_requirements(
 			adev->powerplay.pp_funcs->display_configuration_change(
 				adev->powerplay.pp_handle,
 				&adev->pm.pm_display_cfg);
+
+		amdgpu_pm_compute_clocks(adev);
 	}
 
 	return true;
diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calc_auto.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calc_auto.c
index 5e2ea12fbb73..d0fc54f8fb1c 100644
--- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calc_auto.c
+++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calc_auto.c
@@ -1625,11 +1625,11 @@ void dispclkdppclkdcfclk_deep_sleep_prefetch_parameters_watermarks_and_performan
 				else {
 					v->dsty_after_scaler = 0.0;
 				}
-				v->v_update_offset_pix =dcn_bw_ceil2(v->htotal[k] / 4.0, 1.0);
+				v->v_update_offset_pix[k] = dcn_bw_ceil2(v->htotal[k] / 4.0, 1.0);
 				v->total_repeater_delay_time = v->max_inter_dcn_tile_repeaters * (2.0 / v->dppclk + 3.0 / v->dispclk);
-				v->v_update_width_pix = (14.0 / v->dcf_clk_deep_sleep + 12.0 / v->dppclk + v->total_repeater_delay_time) * v->pixel_clock[k];
-				v->v_ready_offset_pix =dcn_bw_max2(150.0 / v->dppclk, v->total_repeater_delay_time + 20.0 / v->dcf_clk_deep_sleep + 10.0 / v->dppclk) * v->pixel_clock[k];
-				v->t_setup = (v->v_update_offset_pix + v->v_update_width_pix + v->v_ready_offset_pix) / v->pixel_clock[k];
+				v->v_update_width_pix[k] = (14.0 / v->dcf_clk_deep_sleep + 12.0 / v->dppclk + v->total_repeater_delay_time) * v->pixel_clock[k];
+				v->v_ready_offset_pix[k] = dcn_bw_max2(150.0 / v->dppclk, v->total_repeater_delay_time + 20.0 / v->dcf_clk_deep_sleep + 10.0 / v->dppclk) * v->pixel_clock[k];
+				v->t_setup = (v->v_update_offset_pix[k] + v->v_update_width_pix[k] + v->v_ready_offset_pix[k]) / v->pixel_clock[k];
 				v->v_startup[k] =dcn_bw_min2(v->v_startup_lines, v->max_vstartup_lines[k]);
 				if (v->prefetch_mode == 0.0) {
 					v->t_wait =dcn_bw_max3(v->dram_clock_change_latency + v->urgent_latency, v->sr_enter_plus_exit_time, v->urgent_latency);
diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
index 80ec09eef44f..3208188b7ed4 100644
--- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
+++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c
@@ -1096,9 +1096,9 @@ bool dcn_validate_bandwidth(
 			if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state)
 				continue;
 
-			pipe->pipe_dlg_param.vupdate_width = v->v_update_width[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
-			pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
-			pipe->pipe_dlg_param.vready_offset = v->v_ready_offset[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
+			pipe->pipe_dlg_param.vupdate_width = v->v_update_width_pix[input_idx];
+			pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset_pix[input_idx];
+			pipe->pipe_dlg_param.vready_offset = v->v_ready_offset_pix[input_idx];
 			pipe->pipe_dlg_param.vstartup_start = v->v_startup[input_idx];
 
 			pipe->pipe_dlg_param.htotal = pipe->stream->timing.h_total;
@@ -1137,9 +1137,9 @@ bool dcn_validate_bandwidth(
 					 TIMING_3D_FORMAT_SIDE_BY_SIDE))) {
 					if (hsplit_pipe && hsplit_pipe->plane_state == pipe->plane_state) {
 						/* update previously split pipe */
-						hsplit_pipe->pipe_dlg_param.vupdate_width = v->v_update_width[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
-						hsplit_pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
-						hsplit_pipe->pipe_dlg_param.vready_offset = v->v_ready_offset[input_idx][v->dpp_per_plane[input_idx] == 2 ? 1 : 0];
+						hsplit_pipe->pipe_dlg_param.vupdate_width = v->v_update_width_pix[input_idx];
+						hsplit_pipe->pipe_dlg_param.vupdate_offset = v->v_update_offset_pix[input_idx];
+						hsplit_pipe->pipe_dlg_param.vready_offset = v->v_ready_offset_pix[input_idx];
 						hsplit_pipe->pipe_dlg_param.vstartup_start = v->v_startup[input_idx];
 
 						hsplit_pipe->pipe_dlg_param.htotal = pipe->stream->timing.h_total;
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c
index 1c438eedf77a..76fe5a9af3bf 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@@ -460,9 +460,25 @@ void dc_link_set_preferred_link_settings(struct dc *dc,
 					 struct dc_link_settings *link_setting,
 					 struct dc_link *link)
 {
+	int i;
+	struct pipe_ctx *pipe;
+	struct dc_stream_state *link_stream;
 	struct dc_link_settings store_settings = *link_setting;
-	struct dc_stream_state *link_stream =
-		link->dc->current_state->res_ctx.pipe_ctx[0].stream;
+
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &dc->current_state->res_ctx.pipe_ctx[i];
+		if (pipe->stream && pipe->stream->sink
+			&& pipe->stream->sink->link) {
+			if (pipe->stream->sink->link == link)
+				break;
+		}
+	}
+
+	/* Stream not found */
+	if (i == MAX_PIPES)
+		return;
+
+	link_stream = link->dc->current_state->res_ctx.pipe_ctx[i].stream;
 
 	link->preferred_link_setting = store_settings;
 	if (link_stream)
@@ -1160,9 +1176,6 @@ static enum surface_update_type get_plane_info_update_type(const struct dc_surfa
 		 */
 		update_flags->bits.bpp_change = 1;
 
-	if (u->gamma && dce_use_lut(u->plane_info->format))
-		update_flags->bits.gamma_change = 1;
-
 	if (memcmp(&u->plane_info->tiling_info, &u->surface->tiling_info,
 			sizeof(union dc_tiling_info)) != 0) {
 		update_flags->bits.swizzle_change = 1;
@@ -1179,7 +1192,6 @@ static enum surface_update_type get_plane_info_update_type(const struct dc_surfa
 	if (update_flags->bits.rotation_change
 			|| update_flags->bits.stereo_format_change
 			|| update_flags->bits.pixel_format_change
-			|| update_flags->bits.gamma_change
 			|| update_flags->bits.bpp_change
 			|| update_flags->bits.bandwidth_change
 			|| update_flags->bits.output_tf_change)
@@ -1269,13 +1281,26 @@ static enum surface_update_type det_surface_update(const struct dc *dc,
 	if (u->coeff_reduction_factor)
 		update_flags->bits.coeff_reduction_change = 1;
 
+	if (u->gamma) {
+		enum surface_pixel_format format = SURFACE_PIXEL_FORMAT_GRPH_BEGIN;
+
+		if (u->plane_info)
+			format = u->plane_info->format;
+		else if (u->surface)
+			format = u->surface->format;
+
+		if (dce_use_lut(format))
+			update_flags->bits.gamma_change = 1;
+	}
+
 	if (update_flags->bits.in_transfer_func_change) {
 		type = UPDATE_TYPE_MED;
 		elevate_update_type(&overall_type, type);
 	}
 
 	if (update_flags->bits.input_csc_change
-			|| update_flags->bits.coeff_reduction_change) {
+			|| update_flags->bits.coeff_reduction_change
+			|| update_flags->bits.gamma_change) {
 		type = UPDATE_TYPE_FULL;
 		elevate_update_type(&overall_type, type);
 	}
@@ -1379,7 +1404,7 @@ static void notify_display_count_to_smu(
 	 * sent as part of pplib_apply_display_requirements.
 	 * So just return.
 	 */
-	if (!pp_smu->set_display_count)
+	if (!pp_smu || !pp_smu->set_display_count)
 		return;
 
 	display_count = 0;
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index bd58dbae7d3e..9f9503a9b9aa 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -2559,23 +2559,24 @@ void core_link_enable_stream(
 			pipe_ctx->stream_res.stream_enc,
 			&stream->timing);
 
-	resource_build_info_frame(pipe_ctx);
-	core_dc->hwss.update_info_frame(pipe_ctx);
-
-	/* eDP lit up by bios already, no need to enable again. */
-	if (pipe_ctx->stream->signal == SIGNAL_TYPE_EDP &&
-			pipe_ctx->stream->apply_edp_fast_boot_optimization) {
-		pipe_ctx->stream->apply_edp_fast_boot_optimization = false;
-		pipe_ctx->stream->dpms_off = false;
-		return;
-	}
+	if (!IS_FPGA_MAXIMUS_DC(core_dc->ctx->dce_environment)) {
+		resource_build_info_frame(pipe_ctx);
+		core_dc->hwss.update_info_frame(pipe_ctx);
+
+		/* eDP lit up by bios already, no need to enable again. */
+		if (pipe_ctx->stream->signal == SIGNAL_TYPE_EDP &&
+				pipe_ctx->stream->apply_edp_fast_boot_optimization) {
+			pipe_ctx->stream->apply_edp_fast_boot_optimization = false;
+			pipe_ctx->stream->dpms_off = false;
+			return;
+		}
 
-	if (pipe_ctx->stream->dpms_off)
-		return;
+		if (pipe_ctx->stream->dpms_off)
+			return;
 
-	status = enable_link(state, pipe_ctx);
+		status = enable_link(state, pipe_ctx);
 
-	if (status != DC_OK) {
+		if (status != DC_OK) {
 			DC_LOG_WARNING("enabling link %u failed: %d\n",
 			pipe_ctx->stream->sink->link->link_index,
 			status);
@@ -2590,23 +2591,26 @@ void core_link_enable_stream(
 				BREAK_TO_DEBUGGER();
 				return;
 			}
-	}
+		}
 
-	core_dc->hwss.enable_audio_stream(pipe_ctx);
+		core_dc->hwss.enable_audio_stream(pipe_ctx);
 
-	/* turn off otg test pattern if enable */
-	if (pipe_ctx->stream_res.tg->funcs->set_test_pattern)
-		pipe_ctx->stream_res.tg->funcs->set_test_pattern(pipe_ctx->stream_res.tg,
-				CONTROLLER_DP_TEST_PATTERN_VIDEOMODE,
-				COLOR_DEPTH_UNDEFINED);
+		/* turn off otg test pattern if enable */
+		if (pipe_ctx->stream_res.tg->funcs->set_test_pattern)
+			pipe_ctx->stream_res.tg->funcs->set_test_pattern(pipe_ctx->stream_res.tg,
+					CONTROLLER_DP_TEST_PATTERN_VIDEOMODE,
+					COLOR_DEPTH_UNDEFINED);
 
-	core_dc->hwss.enable_stream(pipe_ctx);
+		core_dc->hwss.enable_stream(pipe_ctx);
 
-	if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST)
-		allocate_mst_payload(pipe_ctx);
+		if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST)
+			allocate_mst_payload(pipe_ctx);
+
+		core_dc->hwss.unblank_stream(pipe_ctx,
+			&pipe_ctx->stream->sink->link->cur_link_settings);
+
+	}
 
-	core_dc->hwss.unblank_stream(pipe_ctx,
-		&pipe_ctx->stream->sink->link->cur_link_settings);
 }
 
 void core_link_disable_stream(struct pipe_ctx *pipe_ctx, int option)
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index 7691139363a9..11ea2a226952 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -38,7 +38,7 @@
 #include "inc/compressor.h"
 #include "dml/display_mode_lib.h"
 
-#define DC_VER "3.1.66"
+#define DC_VER "3.1.67"
 
 #define MAX_SURFACES 3
 #define MAX_STREAMS 6
diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c
index bf6261a1584b..d42afa081452 100644
--- a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c
+++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c
@@ -468,6 +468,9 @@ static void dce12_update_clocks(struct dccg *dccg,
 {
 	struct dm_pp_clock_for_voltage_req clock_voltage_req = {0};
 
+	/* TODO: Investigate why this is needed to fix display corruption. */
+	new_clocks->dispclk_khz = new_clocks->dispclk_khz * 115 / 100;
+
 	if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, dccg->clks.dispclk_khz)) {
 		clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAY_CLK;
 		clock_voltage_req.clocks_in_khz = new_clocks->dispclk_khz;
diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
index dc1eed5ba996..6b7cccc486d8 100644
--- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c
@@ -1377,26 +1377,13 @@ static enum dc_status apply_single_controller_ctx_to_hw(
 	/*  */
 	dc->hwss.enable_stream_timing(pipe_ctx, context, dc);
 
-	/* FPGA does not program backend */
-	if (IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment)) {
-		pipe_ctx->stream_res.opp->funcs->opp_set_dyn_expansion(
-		pipe_ctx->stream_res.opp,
-		COLOR_SPACE_YCBCR601,
-		stream->timing.display_color_depth,
-		pipe_ctx->stream->signal);
-
-		pipe_ctx->stream_res.opp->funcs->opp_program_fmt(
-			pipe_ctx->stream_res.opp,
-			&stream->bit_depth_params,
-			&stream->clamping);
-		return DC_OK;
-	}
 	/* TODO: move to stream encoder */
 	if (pipe_ctx->stream->signal != SIGNAL_TYPE_VIRTUAL)
 		if (DC_OK != bios_parser_crtc_source_select(pipe_ctx)) {
 			BREAK_TO_DEBUGGER();
 			return DC_ERROR_UNEXPECTED;
 		}
+
 	pipe_ctx->stream_res.opp->funcs->opp_set_dyn_expansion(
 			pipe_ctx->stream_res.opp,
 			COLOR_SPACE_YCBCR601,
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
index 6bd4ec39f869..a881ff5559ec 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
@@ -44,6 +44,7 @@
 #include "dcn10_hubp.h"
 #include "dcn10_hubbub.h"
 #include "dcn10_cm_common.h"
+#include "dc_link_dp.h"
 
 #define DC_LOGGER_INIT(logger)
 
diff --git a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h
index ddbb673caa08..e688eb9b975c 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h
@@ -504,10 +504,10 @@ struct dcn_bw_internal_vars {
 	float prefetch_mode;
 	float dstx_after_scaler;
 	float dsty_after_scaler;
-	float v_update_offset_pix;
+	float v_update_offset_pix[number_of_planes_minus_one + 1];
 	float total_repeater_delay_time;
-	float v_update_width_pix;
-	float v_ready_offset_pix;
+	float v_update_width_pix[number_of_planes_minus_one + 1];
+	float v_ready_offset_pix[number_of_planes_minus_one + 1];
 	float t_setup;
 	float t_wait;
 	float bandwidth_available_for_immediate_flip;
diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
index e1688902a1b0..4018c7180d00 100644
--- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
+++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
@@ -480,22 +480,11 @@ bool mod_freesync_get_v_position(struct mod_freesync *mod_freesync,
 	return false;
 }
 
-void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
-		const struct dc_stream_state *stream,
-		const struct mod_vrr_params *vrr,
-		struct dc_info_packet *infopacket)
+static void build_vrr_infopacket_header_v1(enum signal_type signal,
+		struct dc_info_packet *infopacket,
+		unsigned int *payload_size)
 {
-	/* SPD info packet for FreeSync */
-	unsigned char checksum = 0;
-	unsigned int idx, payload_size = 0;
-
-	/* Check if Freesync is supported. Return if false. If true,
-	 * set the corresponding bit in the info packet
-	 */
-	if (!vrr->supported || !vrr->send_vsif)
-		return;
-
-	if (dc_is_hdmi_signal(stream->signal)) {
+	if (dc_is_hdmi_signal(signal)) {
 
 		/* HEADER */
 
@@ -510,9 +499,9 @@ void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
 		/* HB2  = [Bits 7:5 = 0] [Bits 4:0 = Length = 0x08] */
 		infopacket->hb2 = 0x08;
 
-		payload_size = 0x08;
+		*payload_size = 0x08;
 
-	} else if (dc_is_dp_signal(stream->signal)) {
+	} else if (dc_is_dp_signal(signal)) {
 
 		/* HEADER */
 
@@ -536,9 +525,62 @@ void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
 		 */
 		infopacket->hb3 = 0x04;
 
-		payload_size = 0x1B;
+		*payload_size = 0x1B;
 	}
+}
+
+static void build_vrr_infopacket_header_v2(enum signal_type signal,
+		struct dc_info_packet *infopacket,
+		unsigned int *payload_size)
+{
+	if (dc_is_hdmi_signal(signal)) {
+
+		/* HEADER */
+
+		/* HB0  = Packet Type = 0x83 (Source Product
+		 *	  Descriptor InfoFrame)
+		 */
+		infopacket->hb0 = DC_HDMI_INFOFRAME_TYPE_SPD;
+
+		/* HB1  = Version = 0x02 */
+		infopacket->hb1 = 0x02;
+
+		/* HB2  = [Bits 7:5 = 0] [Bits 4:0 = Length = 0x09] */
+		infopacket->hb2 = 0x09;
+
+		*payload_size = 0x0A;
 
+	} else if (dc_is_dp_signal(signal)) {
+
+		/* HEADER */
+
+		/* HB0  = Secondary-data Packet ID = 0 - Only non-zero
+		 *	  when used to associate audio related info packets
+		 */
+		infopacket->hb0 = 0x00;
+
+		/* HB1  = Packet Type = 0x83 (Source Product
+		 *	  Descriptor InfoFrame)
+		 */
+		infopacket->hb1 = DC_HDMI_INFOFRAME_TYPE_SPD;
+
+		/* HB2  = [Bits 7:0 = Least significant eight bits -
+		 *	  For INFOFRAME, the value must be 1Bh]
+		 */
+		infopacket->hb2 = 0x1B;
+
+		/* HB3  = [Bits 7:2 = INFOFRAME SDP Version Number = 0x2]
+		 *	  [Bits 1:0 = Most significant two bits = 0x00]
+		 */
+		infopacket->hb3 = 0x08;
+
+		*payload_size = 0x1B;
+	}
+}
+
+static void build_vrr_infopacket_data(const struct mod_vrr_params *vrr,
+		struct dc_info_packet *infopacket)
+{
 	/* PB1 = 0x1A (24bit AMD IEEE OUI (0x00001A) - Byte 0) */
 	infopacket->sb[1] = 0x1A;
 
@@ -576,15 +618,39 @@ void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
 	 */
 	infopacket->sb[8] = (unsigned char)(vrr->max_refresh_in_uhz / 1000000);
 
-	/* PB9 - PB27  = Reserved */
 
+	//FreeSync HDR
+	infopacket->sb[9] = 0;
+	infopacket->sb[10] = 0;
+}
+
+static void build_vrr_infopacket_fs2_data(enum color_transfer_func app_tf,
+		struct dc_info_packet *infopacket)
+{
+	if (app_tf != transfer_func_unknown) {
+		infopacket->valid = true;
+
+		infopacket->sb[6] |= 0x08;  // PB6 = [Bit 3 = Native Color Active]
+
+		if (app_tf == transfer_func_gamma_22) {
+			infopacket->sb[9] |= 0x04;  // PB6 = [Bit 2 = Gamma 2.2 EOTF Active]
+		}
+	}
+}
+
+static void build_vrr_infopacket_checksum(unsigned int *payload_size,
+		struct dc_info_packet *infopacket)
+{
 	/* Calculate checksum */
+	unsigned int idx = 0;
+	unsigned char checksum = 0;
+
 	checksum += infopacket->hb0;
 	checksum += infopacket->hb1;
 	checksum += infopacket->hb2;
 	checksum += infopacket->hb3;
 
-	for (idx = 1; idx <= payload_size; idx++)
+	for (idx = 1; idx <= *payload_size; idx++)
 		checksum += infopacket->sb[idx];
 
 	/* PB0 = Checksum (one byte complement) */
@@ -593,6 +659,64 @@ void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
 	infopacket->valid = true;
 }
 
+static void build_vrr_infopacket_v1(enum signal_type signal,
+		const struct mod_vrr_params *vrr,
+		struct dc_info_packet *infopacket)
+{
+	/* SPD info packet for FreeSync */
+	unsigned int payload_size = 0;
+
+	build_vrr_infopacket_header_v1(signal, infopacket, &payload_size);
+	build_vrr_infopacket_data(vrr, infopacket);
+	build_vrr_infopacket_checksum(&payload_size, infopacket);
+
+	infopacket->valid = true;
+}
+
+static void build_vrr_infopacket_v2(enum signal_type signal,
+		const struct mod_vrr_params *vrr,
+		const enum color_transfer_func *app_tf,
+		struct dc_info_packet *infopacket)
+{
+	unsigned int payload_size = 0;
+
+	build_vrr_infopacket_header_v2(signal, infopacket, &payload_size);
+	build_vrr_infopacket_data(vrr, infopacket);
+
+	if (app_tf != NULL)
+		build_vrr_infopacket_fs2_data(*app_tf, infopacket);
+
+	build_vrr_infopacket_checksum(&payload_size, infopacket);
+
+	infopacket->valid = true;
+}
+
+void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
+		const struct dc_stream_state *stream,
+		const struct mod_vrr_params *vrr,
+		enum vrr_packet_type packet_type,
+		const enum color_transfer_func *app_tf,
+		struct dc_info_packet *infopacket)
+{
+	/* SPD info packet for FreeSync */
+
+	/* Check if Freesync is supported. Return if false. If true,
+	 * set the corresponding bit in the info packet
+	 */
+	if (!vrr->supported || !vrr->send_vsif)
+		return;
+
+	switch (packet_type) {
+	case packet_type_fs2:
+		build_vrr_infopacket_v2(stream->signal, vrr, app_tf, infopacket);
+		break;
+	case packet_type_vrr:
+	case packet_type_fs1:
+	default:
+		build_vrr_infopacket_v1(stream->signal, vrr, infopacket);
+	}
+}
+
 void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync,
 		const struct dc_stream_state *stream,
 		struct mod_freesync_config *in_config,
diff --git a/drivers/gpu/drm/amd/display/modules/inc/mod_freesync.h b/drivers/gpu/drm/amd/display/modules/inc/mod_freesync.h
index a0f32cde721c..949a8b62aa98 100644
--- a/drivers/gpu/drm/amd/display/modules/inc/mod_freesync.h
+++ b/drivers/gpu/drm/amd/display/modules/inc/mod_freesync.h
@@ -54,7 +54,7 @@
 #ifndef MOD_FREESYNC_H_
 #define MOD_FREESYNC_H_
 
-#include "dm_services.h"
+#include "mod_shared.h"
 
 // Access structures
 struct mod_freesync {
@@ -144,6 +144,8 @@ void mod_freesync_get_settings(struct mod_freesync *mod_freesync,
 void mod_freesync_build_vrr_infopacket(struct mod_freesync *mod_freesync,
 		const struct dc_stream_state *stream,
 		const struct mod_vrr_params *vrr,
+		enum vrr_packet_type packet_type,
+		const enum color_transfer_func *app_tf,
 		struct dc_info_packet *infopacket);
 
 void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync,
diff --git a/drivers/gpu/drm/amd/display/modules/inc/mod_shared.h b/drivers/gpu/drm/amd/display/modules/inc/mod_shared.h
new file mode 100644
index 000000000000..238c431ae483
--- /dev/null
+++ b/drivers/gpu/drm/amd/display/modules/inc/mod_shared.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2016 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.
+ *
+ * Authors: AMD
+ *
+ */
+
+
+#ifndef MOD_SHARED_H_
+#define MOD_SHARED_H_
+
+enum color_transfer_func {
+	transfer_func_unknown,
+	transfer_func_srgb,
+	transfer_func_bt709,
+	transfer_func_pq2084,
+	transfer_func_pq2084_interim,
+	transfer_func_linear_0_1,
+	transfer_func_linear_0_125,
+	transfer_func_dolbyvision,
+	transfer_func_gamma_22,
+	transfer_func_gamma_26
+};
+
+enum vrr_packet_type {
+	packet_type_vrr,
+	packet_type_fs1,
+	packet_type_fs2
+};
+
+#endif /* MOD_SHARED_H_ */
diff --git a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c
index 52378fc69079..ff8bfb9b43b0 100644
--- a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c
+++ b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c
@@ -48,9 +48,12 @@ static void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
 	unsigned int i;
 	unsigned int pixelEncoding = 0;
 	unsigned int colorimetryFormat = 0;
+	bool stereo3dSupport = false;
 
-	if (stream->timing.timing_3d_format != TIMING_3D_FORMAT_NONE && stream->view_format != VIEW_3D_FORMAT_NONE)
+	if (stream->timing.timing_3d_format != TIMING_3D_FORMAT_NONE && stream->view_format != VIEW_3D_FORMAT_NONE) {
 		vscPacketRevision = 1;
+		stereo3dSupport = true;
+	}
 
 	/*VSC packet set to 2 when DP revision >= 1.2*/
 	if (stream->psr_version != 0)
@@ -94,12 +97,59 @@ static void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
 		info_packet->hb2 = 0x01;	// 01h = Revision number. VSC SDP supporting 3D stereo only
 		info_packet->hb3 = 0x01;	// 01h = VSC SDP supporting 3D stereo only (HB2 = 01h).
 
-		if (stream->timing.timing_3d_format == TIMING_3D_FORMAT_INBAND_FA)
-			info_packet->sb[0] = 0x1;
-
 		info_packet->valid = true;
 	}
 
+	if (stereo3dSupport) {
+		/* ==============================================================================================================|
+		 * A. STEREO 3D
+		 * ==============================================================================================================|
+		 * VSC Payload (1 byte) From DP1.2 spec
+		 *
+		 * Bits 3:0 (Stereo Interface Method Code)  |  Bits 7:4 (Stereo Interface Method Specific Parameter)
+		 * -----------------------------------------------------------------------------------------------------
+		 * 0 = Non Stereo Video                     |  Must be set to 0x0
+		 * -----------------------------------------------------------------------------------------------------
+		 * 1 = Frame/Field Sequential               |  0x0: L + R view indication based on MISC1 bit 2:1
+		 *                                          |  0x1: Right when Stereo Signal = 1
+		 *                                          |  0x2: Left when Stereo Signal = 1
+		 *                                          |  (others reserved)
+		 * -----------------------------------------------------------------------------------------------------
+		 * 2 = Stacked Frame                        |  0x0: Left view is on top and right view on bottom
+		 *                                          |  (others reserved)
+		 * -----------------------------------------------------------------------------------------------------
+		 * 3 = Pixel Interleaved                    |  0x0: horiz interleaved, right view pixels on even lines
+		 *                                          |  0x1: horiz interleaved, right view pixels on odd lines
+		 *                                          |  0x2: checker board, start with left view pixel
+		 *                                          |  0x3: vertical interleaved, start with left view pixels
+		 *                                          |  0x4: vertical interleaved, start with right view pixels
+		 *                                          |  (others reserved)
+		 * -----------------------------------------------------------------------------------------------------
+		 * 4 = Side-by-side                         |  0x0: left half represents left eye view
+		 *                                          |  0x1: left half represents right eye view
+		 */
+		switch (stream->timing.timing_3d_format) {
+		case TIMING_3D_FORMAT_HW_FRAME_PACKING:
+		case TIMING_3D_FORMAT_SW_FRAME_PACKING:
+		case TIMING_3D_FORMAT_TOP_AND_BOTTOM:
+		case TIMING_3D_FORMAT_TB_SW_PACKED:
+			info_packet->sb[0] = 0x02; // Stacked Frame, Left view is on top and right view on bottom.
+			break;
+		case TIMING_3D_FORMAT_DP_HDMI_INBAND_FA:
+		case TIMING_3D_FORMAT_INBAND_FA:
+			info_packet->sb[0] = 0x01; // Frame/Field Sequential, L + R view indication based on MISC1 bit 2:1
+			break;
+		case TIMING_3D_FORMAT_SIDE_BY_SIDE:
+		case TIMING_3D_FORMAT_SBS_SW_PACKED:
+			info_packet->sb[0] = 0x04; // Side-by-side
+			break;
+		default:
+			info_packet->sb[0] = 0x00; // No Stereo Video, Shall be cleared to 0x0.
+			break;
+		}
+
+	}
+
 	/* 05h = VSC SDP supporting 3D stereo, PSR2, and Pixel Encoding/Colorimetry Format indication.
 	 *   Added in DP1.3, a DP Source device is allowed to indicate the pixel encoding/colorimetry
 	 *   format to the DP Sink device with VSC SDP only when the DP Sink device supports it
diff --git a/drivers/gpu/drm/amd/display/modules/stats/stats.c b/drivers/gpu/drm/amd/display/modules/stats/stats.c
index 3d4c1b1ab8c4..03121ca64fe4 100644
--- a/drivers/gpu/drm/amd/display/modules/stats/stats.c
+++ b/drivers/gpu/drm/amd/display/modules/stats/stats.c
@@ -186,12 +186,8 @@ void mod_stats_destroy(struct mod_stats *mod_stats)
 	if (mod_stats != NULL) {
 		struct core_stats *core_stats = MOD_STATS_TO_CORE(mod_stats);
 
-		if (core_stats->time != NULL)
-			kfree(core_stats->time);
-
-		if (core_stats->events != NULL)
-			kfree(core_stats->events);
-
+		kfree(core_stats->time);
+		kfree(core_stats->events);
 		kfree(core_stats);
 	}
 }
diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h
index 86b167ec9863..2083c308007c 100644
--- a/drivers/gpu/drm/amd/include/amd_shared.h
+++ b/drivers/gpu/drm/amd/include/amd_shared.h
@@ -109,6 +109,7 @@ enum amd_powergating_state {
 #define AMD_PG_SUPPORT_GFX_PIPELINE		(1 << 12)
 #define AMD_PG_SUPPORT_MMHUB			(1 << 13)
 #define AMD_PG_SUPPORT_VCN			(1 << 14)
+#define AMD_PG_SUPPORT_VCN_DPG	(1 << 15)
 
 enum PP_FEATURE_MASK {
 	PP_SCLK_DPM_MASK = 0x1,
diff --git a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_offset.h
index 216a401028de..4b7da589e14a 100644
--- a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_offset.h
+++ b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_offset.h
@@ -33,6 +33,14 @@
 #define mmUVD_POWER_STATUS_BASE_IDX                                                                    1
 #define mmCC_UVD_HARVESTING                                                                            0x00c7
 #define mmCC_UVD_HARVESTING_BASE_IDX                                                                   1
+#define mmUVD_DPG_LMA_CTL                                                                              0x00d1
+#define mmUVD_DPG_LMA_CTL_BASE_IDX                                                                     1
+#define mmUVD_DPG_LMA_DATA                                                                             0x00d2
+#define mmUVD_DPG_LMA_DATA_BASE_IDX                                                                    1
+#define mmUVD_DPG_LMA_MASK                                                                             0x00d3
+#define mmUVD_DPG_LMA_MASK_BASE_IDX                                                                    1
+#define mmUVD_DPG_PAUSE                                                                                0x00d4
+#define mmUVD_DPG_PAUSE_BASE_IDX                                                                       1
 #define mmUVD_SCRATCH1                                                                                 0x00d5
 #define mmUVD_SCRATCH1_BASE_IDX                                                                        1
 #define mmUVD_SCRATCH2                                                                                 0x00d6
diff --git a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_sh_mask.h
index 124383dac284..26382f5d5354 100644
--- a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_sh_mask.h
+++ b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_1_0_sh_mask.h
@@ -87,6 +87,26 @@
 //CC_UVD_HARVESTING
 #define CC_UVD_HARVESTING__UVD_DISABLE__SHIFT                                                                 0x1
 #define CC_UVD_HARVESTING__UVD_DISABLE_MASK                                                                   0x00000002L
+//UVD_DPG_LMA_CTL
+#define UVD_DPG_LMA_CTL__READ_WRITE__SHIFT                                                                    0x0
+#define UVD_DPG_LMA_CTL__MASK_EN__SHIFT                                                                       0x1
+#define UVD_DPG_LMA_CTL__ADDR_AUTO_INCREMENT__SHIFT                                                           0x2
+#define UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT                                                                      0x4
+#define UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT                                                               0x10
+#define UVD_DPG_LMA_CTL__READ_WRITE_MASK                                                                      0x00000001L
+#define UVD_DPG_LMA_CTL__MASK_EN_MASK                                                                         0x00000002L
+#define UVD_DPG_LMA_CTL__ADDR_AUTO_INCREMENT_MASK                                                             0x00000004L
+#define UVD_DPG_LMA_CTL__SRAM_SEL_MASK                                                                        0x00000010L
+#define UVD_DPG_LMA_CTL__READ_WRITE_ADDR_MASK                                                                 0xFFFF0000L
+//UVD_DPG_PAUSE
+#define UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ__SHIFT                                                              0x0
+#define UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK__SHIFT                                                              0x1
+#define UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ__SHIFT                                                                0x2
+#define UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK__SHIFT                                                                0x3
+#define UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK                                                                0x00000001L
+#define UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK                                                                0x00000002L
+#define UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK                                                                  0x00000004L
+#define UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK                                                                  0x00000008L
 //UVD_SCRATCH1
 #define UVD_SCRATCH1__SCRATCH1_DATA__SHIFT                                                                    0x0
 #define UVD_SCRATCH1__SCRATCH1_DATA_MASK                                                                      0xFFFFFFFFL
@@ -983,6 +1003,7 @@
 #define UVD_MASTINT_EN__SYS_EN_MASK                                                                           0x00000004L
 #define UVD_MASTINT_EN__INT_OVERRUN_MASK                                                                      0x007FFFF0L
 //UVD_SYS_INT_EN
+#define UVD_SYS_INT_EN__UVD_JRBC_EN__SHIFT                                                                    0x4
 #define UVD_SYS_INT_EN__UVD_JRBC_EN_MASK                                                                      0x00000010L
 //JPEG_CGC_CTRL
 #define JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT                                                                  0x0
@@ -1138,7 +1159,11 @@
 #define UVD_VCPU_CACHE_SIZE2__CACHE_SIZE2_MASK                                                                0x001FFFFFL
 //UVD_VCPU_CNTL
 #define UVD_VCPU_CNTL__CLK_EN__SHIFT                                                                          0x9
+#define UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP__SHIFT                                                         0x11
+#define UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT                                                                 0x14
 #define UVD_VCPU_CNTL__CLK_EN_MASK                                                                            0x00000200L
+#define UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP_MASK                                                           0x00020000L
+#define UVD_VCPU_CNTL__PRB_TIMEOUT_VAL_MASK                                                                   0x0FF00000L
 //UVD_SOFT_RESET
 #define UVD_SOFT_RESET__RBC_SOFT_RESET__SHIFT                                                                 0x0
 #define UVD_SOFT_RESET__LBSI_SOFT_RESET__SHIFT                                                                0x1
diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
index f43ed96cfa6c..64ecffd52126 100644
--- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
@@ -146,10 +146,10 @@ struct kgd2kfd_shared_resources {
 	 * is reserved: (D & reserved_doorbell_mask) == reserved_doorbell_val
 	 *
 	 * KFD currently uses 1024 (= 0x3ff) doorbells per process. If
-	 * doorbells 0x0f0-0x0f7 and 0x2f-0x2f7 are reserved, that means
-	 * mask would be set to 0x1f8 and val set to 0x0f0.
+	 * doorbells 0x0e0-0x0ff and 0x2e0-0x2ff are reserved, that means
+	 * mask would be set to 0x1e0 and val set to 0x0e0.
 	 */
-	unsigned int sdma_doorbell[2][2];
+	unsigned int sdma_doorbell[2][8];
 	unsigned int reserved_doorbell_mask;
 	unsigned int reserved_doorbell_val;
 
@@ -409,9 +409,9 @@ struct kfd2kgd_calls {
 			struct dma_fence **ef);
 	void (*destroy_process_vm)(struct kgd_dev *kgd, void *vm);
 	void (*release_process_vm)(struct kgd_dev *kgd, void *vm);
-	uint32_t (*get_process_page_dir)(void *vm);
+	uint64_t (*get_process_page_dir)(void *vm);
 	void (*set_vm_context_page_table_base)(struct kgd_dev *kgd,
-			uint32_t vmid, uint32_t page_table_base);
+			uint32_t vmid, uint64_t page_table_base);
 	int (*alloc_memory_of_gpu)(struct kgd_dev *kgd, uint64_t va,
 			uint64_t size, void *vm,
 			struct kgd_mem **mem, uint64_t *offset,
diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
index 448dee481a38..bd7404532029 100644
--- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
@@ -113,6 +113,7 @@ enum amd_pp_sensors {
 	AMDGPU_PP_SENSOR_GPU_POWER,
 	AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK,
 	AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK,
+	AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK,
 };
 
 enum amd_pp_task {
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.c b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.c
index d27c1c9df286..4588bddf8b33 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.c
@@ -488,7 +488,8 @@ int pp_atomfwctrl_get_gpio_information(struct pp_hwmgr *hwmgr,
 	return 0;
 }
 
-int pp_atomfwctrl_get_clk_information_by_clkid(struct pp_hwmgr *hwmgr, BIOS_CLKID id, uint32_t *frequency)
+int pp_atomfwctrl_get_clk_information_by_clkid(struct pp_hwmgr *hwmgr,
+					       uint8_t id, uint32_t *frequency)
 {
 	struct amdgpu_device *adev = hwmgr->adev;
 	struct atom_get_smu_clock_info_parameters_v3_1   parameters;
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.h b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.h
index 22e21668c93a..fe9e8ceef50e 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.h
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.h
@@ -236,7 +236,7 @@ int pp_atomfwctrl_get_vbios_bootup_values(struct pp_hwmgr *hwmgr,
 int pp_atomfwctrl_get_smc_dpm_information(struct pp_hwmgr *hwmgr,
 			struct pp_atomfwctrl_smc_dpm_parameters *param);
 int pp_atomfwctrl_get_clk_information_by_clkid(struct pp_hwmgr *hwmgr,
-					BIOS_CLKID id, uint32_t *frequency);
+					uint8_t id, uint32_t *frequency);
 
 #endif
 
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c
index 9808bd48b386..5d1dae25a466 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c
@@ -552,6 +552,8 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
 {
 	struct smu10_hwmgr *data = hwmgr->backend;
 	struct amdgpu_device *adev = hwmgr->adev;
+	uint32_t min_sclk = hwmgr->display_config->min_core_set_clock;
+	uint32_t min_mclk = hwmgr->display_config->min_mem_set_clock/100;
 
 	if (hwmgr->smu_version < 0x1E3700) {
 		pr_info("smu firmware version too old, can not set dpm level\n");
@@ -563,6 +565,13 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
 	    (adev->rev_id >= 8))
 		return 0;
 
+	if (min_sclk < data->gfx_min_freq_limit)
+		min_sclk = data->gfx_min_freq_limit;
+
+	min_sclk /= 100; /* transfer 10KHz to MHz */
+	if (min_mclk < data->clock_table.FClocks[0].Freq)
+		min_mclk = data->clock_table.FClocks[0].Freq;
+
 	switch (level) {
 	case AMD_DPM_FORCED_LEVEL_HIGH:
 	case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
@@ -595,18 +604,18 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
 	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinGfxClk,
-						data->gfx_min_freq_limit/100);
+						min_sclk);
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetSoftMaxGfxClk,
-						data->gfx_min_freq_limit/100);
+						min_sclk);
 		break;
 	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinFclkByFreq,
-						SMU10_UMD_PSTATE_MIN_FCLK);
+						min_mclk);
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetSoftMaxFclkByFreq,
-						SMU10_UMD_PSTATE_MIN_FCLK);
+						min_mclk);
 		break;
 	case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
 		smum_send_msg_to_smc_with_parameter(hwmgr,
@@ -638,12 +647,12 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
 	case AMD_DPM_FORCED_LEVEL_AUTO:
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinGfxClk,
-						data->gfx_min_freq_limit/100);
+						min_sclk);
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinFclkByFreq,
 						hwmgr->display_config->num_display > 3 ?
 						SMU10_UMD_PSTATE_PEAK_FCLK :
-						SMU10_UMD_PSTATE_MIN_FCLK);
+						min_mclk);
 
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinSocclkByFreq,
@@ -674,10 +683,10 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
 						data->gfx_min_freq_limit/100);
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetHardMinFclkByFreq,
-						SMU10_UMD_PSTATE_MIN_FCLK);
+						min_mclk);
 		smum_send_msg_to_smc_with_parameter(hwmgr,
 						PPSMC_MSG_SetSoftMaxFclkByFreq,
-						SMU10_UMD_PSTATE_MIN_FCLK);
+						min_mclk);
 		break;
 	case AMD_DPM_FORCED_LEVEL_MANUAL:
 	case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
index 04b7da0e39a6..6c99cbf51c08 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
@@ -4106,17 +4106,17 @@ static int smu7_register_irq_handlers(struct pp_hwmgr *hwmgr)
 	source->funcs = &smu7_irq_funcs;
 
 	amdgpu_irq_add_id((struct amdgpu_device *)(hwmgr->adev),
-			AMDGPU_IH_CLIENTID_LEGACY,
+			AMDGPU_IRQ_CLIENTID_LEGACY,
 			VISLANDS30_IV_SRCID_CG_TSS_THERMAL_LOW_TO_HIGH,
 			source);
 	amdgpu_irq_add_id((struct amdgpu_device *)(hwmgr->adev),
-			AMDGPU_IH_CLIENTID_LEGACY,
+			AMDGPU_IRQ_CLIENTID_LEGACY,
 			VISLANDS30_IV_SRCID_CG_TSS_THERMAL_HIGH_TO_LOW,
 			source);
 
 	/* Register CTF(GPIO_19) interrupt */
 	amdgpu_irq_add_id((struct amdgpu_device *)(hwmgr->adev),
-			AMDGPU_IH_CLIENTID_LEGACY,
+			AMDGPU_IRQ_CLIENTID_LEGACY,
 			VISLANDS30_IV_SRCID_GPIO_19,
 			source);
 
@@ -5035,6 +5035,18 @@ static int smu7_get_performance_level(struct pp_hwmgr *hwmgr, const struct pp_hw
 	return 0;
 }
 
+static int smu7_power_off_asic(struct pp_hwmgr *hwmgr)
+{
+	int result;
+
+	result = smu7_disable_dpm_tasks(hwmgr);
+	PP_ASSERT_WITH_CODE((0 == result),
+			"[disable_dpm_tasks] Failed to disable DPM!",
+			);
+
+	return result;
+}
+
 static const struct pp_hwmgr_func smu7_hwmgr_funcs = {
 	.backend_init = &smu7_hwmgr_backend_init,
 	.backend_fini = &smu7_hwmgr_backend_fini,
@@ -5092,6 +5104,7 @@ static const struct pp_hwmgr_func smu7_hwmgr_funcs = {
 	.get_power_profile_mode = smu7_get_power_profile_mode,
 	.set_power_profile_mode = smu7_set_power_profile_mode,
 	.get_performance_level = smu7_get_performance_level,
+	.power_off_asic = smu7_power_off_asic,
 };
 
 uint8_t smu7_get_sleep_divider_id_from_clock(uint32_t clock,
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu8_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu8_hwmgr.c
index b8637049198d..53cf787560f7 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu8_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu8_hwmgr.c
@@ -880,7 +880,7 @@ static int smu8_set_power_state_tasks(struct pp_hwmgr *hwmgr, const void *input)
 	smu8_update_low_mem_pstate(hwmgr, input);
 
 	return 0;
-};
+}
 
 
 static int smu8_setup_asic_task(struct pp_hwmgr *hwmgr)
@@ -934,14 +934,6 @@ static void smu8_reset_cc6_data(struct pp_hwmgr *hwmgr)
 	hw_data->cc6_settings.cpu_pstate_disable = false;
 }
 
-static int smu8_power_off_asic(struct pp_hwmgr *hwmgr)
-{
-	smu8_power_up_display_clock_sys_pll(hwmgr);
-	smu8_clear_nb_dpm_flag(hwmgr);
-	smu8_reset_cc6_data(hwmgr);
-	return 0;
-};
-
 static void smu8_program_voting_clients(struct pp_hwmgr *hwmgr)
 {
 	cgs_write_ind_register(hwmgr->device, CGS_IND_REG__SMC,
@@ -1011,6 +1003,17 @@ static void smu8_reset_acp_boot_level(struct pp_hwmgr *hwmgr)
 	data->acp_boot_level = 0xff;
 }
 
+static int smu8_enable_dpm_tasks(struct pp_hwmgr *hwmgr)
+{
+	smu8_program_voting_clients(hwmgr);
+	if (smu8_start_dpm(hwmgr))
+		return -EINVAL;
+	smu8_program_bootup_state(hwmgr);
+	smu8_reset_acp_boot_level(hwmgr);
+
+	return 0;
+}
+
 static int smu8_disable_dpm_tasks(struct pp_hwmgr *hwmgr)
 {
 	smu8_disable_nb_dpm(hwmgr);
@@ -1020,18 +1023,16 @@ static int smu8_disable_dpm_tasks(struct pp_hwmgr *hwmgr)
 		return -EINVAL;
 
 	return 0;
-};
+}
 
-static int smu8_enable_dpm_tasks(struct pp_hwmgr *hwmgr)
+static int smu8_power_off_asic(struct pp_hwmgr *hwmgr)
 {
-	smu8_program_voting_clients(hwmgr);
-	if (smu8_start_dpm(hwmgr))
-		return -EINVAL;
-	smu8_program_bootup_state(hwmgr);
-	smu8_reset_acp_boot_level(hwmgr);
-
+	smu8_disable_dpm_tasks(hwmgr);
+	smu8_power_up_display_clock_sys_pll(hwmgr);
+	smu8_clear_nb_dpm_flag(hwmgr);
+	smu8_reset_cc6_data(hwmgr);
 	return 0;
-};
+}
 
 static int smu8_apply_state_adjust_rules(struct pp_hwmgr *hwmgr,
 				struct pp_power_state  *prequest_ps,
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c
index 2aab1b475945..8ad4e6960efd 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu_helper.c
@@ -545,7 +545,7 @@ int phm_irq_process(struct amdgpu_device *adev,
 	uint32_t client_id = entry->client_id;
 	uint32_t src_id = entry->src_id;
 
-	if (client_id == AMDGPU_IH_CLIENTID_LEGACY) {
+	if (client_id == AMDGPU_IRQ_CLIENTID_LEGACY) {
 		if (src_id == VISLANDS30_IV_SRCID_CG_TSS_THERMAL_LOW_TO_HIGH)
 			pr_warn("GPU over temperature range detected on PCIe %d:%d.%d!\n",
 						PCI_BUS_NUM(adev->pdev->devfn),
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
index ca9be583fb62..419a1d77d661 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c
@@ -39,6 +39,7 @@
 #include "soc15_common.h"
 #include "pppcielanes.h"
 #include "vega10_hwmgr.h"
+#include "vega10_smumgr.h"
 #include "vega10_processpptables.h"
 #include "vega10_pptable.h"
 #include "vega10_thermal.h"
@@ -3713,6 +3714,11 @@ static int vega10_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 			SMUSVI0_PLANE0_CURRENTVID__CURRENT_SVI0_PLANE0_VID__SHIFT;
 		*((uint32_t *)value) = (uint32_t)convert_to_vddc((uint8_t)val_vid);
 		return 0;
+	case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK:
+		ret = vega10_get_enabled_smc_features(hwmgr, (uint64_t *)value);
+		if (!ret)
+			*size = 8;
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -4940,16 +4946,6 @@ static const struct pp_hwmgr_func vega10_hwmgr_funcs = {
 	.get_performance_level = vega10_get_performance_level,
 };
 
-int vega10_enable_smc_features(struct pp_hwmgr *hwmgr,
-		bool enable, uint32_t feature_mask)
-{
-	int msg = enable ? PPSMC_MSG_EnableSmuFeatures :
-			PPSMC_MSG_DisableSmuFeatures;
-
-	return smum_send_msg_to_smc_with_parameter(hwmgr,
-			msg, feature_mask);
-}
-
 int vega10_hwmgr_init(struct pp_hwmgr *hwmgr)
 {
 	hwmgr->hwmgr_func = &vega10_hwmgr_funcs;
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
index 339820da9e6a..89870556de1b 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.h
@@ -441,7 +441,5 @@ int vega10_update_uvd_dpm(struct pp_hwmgr *hwmgr, bool bgate);
 int vega10_update_samu_dpm(struct pp_hwmgr *hwmgr, bool bgate);
 int vega10_update_acp_dpm(struct pp_hwmgr *hwmgr, bool bgate);
 int vega10_enable_disable_vce_dpm(struct pp_hwmgr *hwmgr, bool enable);
-int vega10_enable_smc_features(struct pp_hwmgr *hwmgr,
-		bool enable, uint32_t feature_mask);
 
 #endif /* _VEGA10_HWMGR_H_ */
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c
index 22364875a943..2d88abf97e7b 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c
@@ -23,6 +23,7 @@
 
 #include "hwmgr.h"
 #include "vega10_hwmgr.h"
+#include "vega10_smumgr.h"
 #include "vega10_powertune.h"
 #include "vega10_ppsmc.h"
 #include "vega10_inc.h"
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_thermal.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_thermal.c
index aa044c1955fe..407762b36901 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_thermal.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_thermal.c
@@ -23,6 +23,7 @@
 
 #include "vega10_thermal.h"
 #include "vega10_hwmgr.h"
+#include "vega10_smumgr.h"
 #include "vega10_ppsmc.h"
 #include "vega10_inc.h"
 #include "soc15_common.h"
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c
index 0789d64246ca..9600e2f226e9 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c
@@ -745,8 +745,8 @@ static int vega12_init_smc_table(struct pp_hwmgr *hwmgr)
 
 	memcpy(pp_table, pptable_information->smc_pptable, sizeof(PPTable_t));
 
-	result = vega12_copy_table_to_smc(hwmgr,
-			(uint8_t *)pp_table, TABLE_PPTABLE);
+	result = smum_smc_table_manager(hwmgr,
+					(uint8_t *)pp_table, TABLE_PPTABLE, false);
 	PP_ASSERT_WITH_CODE(!result,
 			"Failed to upload PPtable!", return result);
 
@@ -1317,7 +1317,11 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 		break;
 	case AMDGPU_PP_SENSOR_GPU_POWER:
 		ret = vega12_get_gpu_power(hwmgr, (uint32_t *)value);
-
+		break;
+	case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK:
+		ret = vega12_get_enabled_smc_features(hwmgr, (uint64_t *)value);
+		if (!ret)
+			*size = 8;
 		break;
 	default:
 		ret = -EINVAL;
@@ -2103,8 +2107,8 @@ static int vega12_display_configuration_changed_task(struct pp_hwmgr *hwmgr)
 
 	if ((data->water_marks_bitmap & WaterMarksExist) &&
 			!(data->water_marks_bitmap & WaterMarksLoaded)) {
-		result = vega12_copy_table_to_smc(hwmgr,
-			(uint8_t *)wm_table, TABLE_WATERMARKS);
+		result = smum_smc_table_manager(hwmgr,
+						(uint8_t *)wm_table, TABLE_WATERMARKS, false);
 		PP_ASSERT_WITH_CODE(result, "Failed to update WMTABLE!", return EINVAL);
 		data->water_marks_bitmap |= WaterMarksLoaded;
 	}
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c
index d45cbfe8e184..2a554f9edcda 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c
@@ -461,7 +461,7 @@ static int vega20_get_number_of_dpm_level(struct pp_hwmgr *hwmgr,
 			"[GetNumOfDpmLevel] failed to get dpm levels!",
 			return ret);
 
-	vega20_read_arg_from_smc(hwmgr, num_of_levels);
+	*num_of_levels = smum_get_argument(hwmgr);
 	PP_ASSERT_WITH_CODE(*num_of_levels > 0,
 			"[GetNumOfDpmLevel] number of clk levels is invalid!",
 			return -EINVAL);
@@ -481,7 +481,7 @@ static int vega20_get_dpm_frequency_by_index(struct pp_hwmgr *hwmgr,
 			"[GetDpmFreqByIndex] failed to get dpm freq by index!",
 			return ret);
 
-	vega20_read_arg_from_smc(hwmgr, clk);
+	*clk = smum_get_argument(hwmgr);
 	PP_ASSERT_WITH_CODE(*clk,
 			"[GetDpmFreqByIndex] clk value is invalid!",
 			return -EINVAL);
@@ -743,8 +743,8 @@ static int vega20_init_smc_table(struct pp_hwmgr *hwmgr)
 
 	memcpy(pp_table, pptable_information->smc_pptable, sizeof(PPTable_t));
 
-	result = vega20_copy_table_to_smc(hwmgr,
-			(uint8_t *)pp_table, TABLE_PPTABLE);
+	result = smum_smc_table_manager(hwmgr,
+					(uint8_t *)pp_table, TABLE_PPTABLE, false);
 	PP_ASSERT_WITH_CODE(!result,
 			"[InitSMCTable] Failed to upload PPtable!",
 			return result);
@@ -1044,7 +1044,7 @@ static int vega20_od8_get_gfx_clock_base_voltage(
 			"[GetBaseVoltage] failed to get GFXCLK AVFS voltage from SMU!",
 			return ret);
 
-	vega20_read_arg_from_smc(hwmgr, voltage);
+	*voltage = smum_get_argument(hwmgr);
 	*voltage = *voltage / VOLTAGE_SCALE;
 
 	return 0;
@@ -1067,7 +1067,7 @@ static int vega20_od8_initialize_default_settings(
 	vega20_od8_set_feature_id(hwmgr);
 
 	/* Set default values */
-	ret = vega20_copy_table_from_smc(hwmgr, (uint8_t *)od_table, TABLE_OVERDRIVE);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)od_table, TABLE_OVERDRIVE, true);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to export over drive table!",
 			return ret);
@@ -1195,7 +1195,7 @@ static int vega20_od8_initialize_default_settings(
 		}
 	}
 
-	ret = vega20_copy_table_to_smc(hwmgr, (uint8_t *)od_table, TABLE_OVERDRIVE);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)od_table, TABLE_OVERDRIVE, false);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to import over drive table!",
 			return ret);
@@ -1214,7 +1214,7 @@ static int vega20_od8_set_settings(
 	struct vega20_od8_single_setting *od8_settings =
 			data->od8_settings.od8_settings_array;
 
-	ret = vega20_copy_table_from_smc(hwmgr, (uint8_t *)(&od_table), TABLE_OVERDRIVE);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)(&od_table), TABLE_OVERDRIVE, true);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to export over drive table!",
 			return ret);
@@ -1271,7 +1271,7 @@ static int vega20_od8_set_settings(
 		break;
 	}
 
-	ret = vega20_copy_table_to_smc(hwmgr, (uint8_t *)(&od_table), TABLE_OVERDRIVE);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)(&od_table), TABLE_OVERDRIVE, false);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to import over drive table!",
 			return ret);
@@ -1401,7 +1401,7 @@ static int vega20_get_max_sustainable_clock(struct pp_hwmgr *hwmgr,
 			(clock_select << 16))) == 0,
 			"[GetMaxSustainableClock] Failed to get max DC clock from SMC!",
 			return ret);
-	vega20_read_arg_from_smc(hwmgr, clock);
+	*clock = smum_get_argument(hwmgr);
 
 	/* if DC limit is zero, return AC limit */
 	if (*clock == 0) {
@@ -1410,7 +1410,7 @@ static int vega20_get_max_sustainable_clock(struct pp_hwmgr *hwmgr,
 			(clock_select << 16))) == 0,
 			"[GetMaxSustainableClock] failed to get max AC clock from SMC!",
 			return ret);
-		vega20_read_arg_from_smc(hwmgr, clock);
+		*clock = smum_get_argument(hwmgr);
 	}
 
 	return 0;
@@ -1770,14 +1770,14 @@ static int vega20_get_clock_ranges(struct pp_hwmgr *hwmgr,
 				PPSMC_MSG_GetMaxDpmFreq, (clock_select << 16))) == 0,
 				"[GetClockRanges] Failed to get max clock from SMC!",
 				return ret);
-		vega20_read_arg_from_smc(hwmgr, clock);
+		*clock = smum_get_argument(hwmgr);
 	} else {
 		PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc_with_parameter(hwmgr,
 				PPSMC_MSG_GetMinDpmFreq,
 				(clock_select << 16))) == 0,
 				"[GetClockRanges] Failed to get min clock from SMC!",
 				return ret);
-		vega20_read_arg_from_smc(hwmgr, clock);
+		*clock = smum_get_argument(hwmgr);
 	}
 
 	return 0;
@@ -1841,7 +1841,7 @@ static int vega20_get_gpu_power(struct pp_hwmgr *hwmgr,
 	int ret = 0;
 	SmuMetrics_t metrics_table;
 
-	ret = vega20_copy_table_from_smc(hwmgr, (uint8_t *)&metrics_table, TABLE_SMU_METRICS);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)&metrics_table, TABLE_SMU_METRICS, true);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to export SMU METRICS table!",
 			return ret);
@@ -1862,7 +1862,7 @@ static int vega20_get_current_gfx_clk_freq(struct pp_hwmgr *hwmgr, uint32_t *gfx
 			PPSMC_MSG_GetDpmClockFreq, (PPCLK_GFXCLK << 16))) == 0,
 			"[GetCurrentGfxClkFreq] Attempt to get Current GFXCLK Frequency Failed!",
 			return ret);
-	vega20_read_arg_from_smc(hwmgr, &gfx_clk);
+	gfx_clk = smum_get_argument(hwmgr);
 
 	*gfx_freq = gfx_clk * 100;
 
@@ -1880,7 +1880,7 @@ static int vega20_get_current_mclk_freq(struct pp_hwmgr *hwmgr, uint32_t *mclk_f
 			PPSMC_MSG_GetDpmClockFreq, (PPCLK_UCLK << 16))) == 0,
 			"[GetCurrentMClkFreq] Attempt to get Current MCLK Frequency Failed!",
 			return ret);
-	vega20_read_arg_from_smc(hwmgr, &mem_clk);
+	mem_clk = smum_get_argument(hwmgr);
 
 	*mclk_freq = mem_clk * 100;
 
@@ -1893,7 +1893,7 @@ static int vega20_get_current_activity_percent(struct pp_hwmgr *hwmgr,
 	int ret = 0;
 	SmuMetrics_t metrics_table;
 
-	ret = vega20_copy_table_from_smc(hwmgr, (uint8_t *)&metrics_table, TABLE_SMU_METRICS);
+	ret = smum_smc_table_manager(hwmgr, (uint8_t *)&metrics_table, TABLE_SMU_METRICS, true);
 	PP_ASSERT_WITH_CODE(!ret,
 			"Failed to export SMU METRICS table!",
 			return ret);
@@ -1941,6 +1941,11 @@ static int vega20_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 		*size = 16;
 		ret = vega20_get_gpu_power(hwmgr, (uint32_t *)value);
 		break;
+	case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK:
+		ret = vega20_get_enabled_smc_features(hwmgr, (uint64_t *)value);
+		if (!ret)
+			*size = 8;
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -2612,18 +2617,18 @@ static int vega20_odn_edit_dpm_table(struct pp_hwmgr *hwmgr,
 		data->gfxclk_overdrive = false;
 		data->memclk_overdrive = false;
 
-		ret = vega20_copy_table_from_smc(hwmgr,
-				(uint8_t *)od_table,
-				TABLE_OVERDRIVE);
+		ret = smum_smc_table_manager(hwmgr,
+					     (uint8_t *)od_table,
+					     TABLE_OVERDRIVE, true);
 		PP_ASSERT_WITH_CODE(!ret,
 				"Failed to export overdrive table!",
 				return ret);
 		break;
 
 	case PP_OD_COMMIT_DPM_TABLE:
-		ret = vega20_copy_table_to_smc(hwmgr,
-				(uint8_t *)od_table,
-				TABLE_OVERDRIVE);
+		ret = smum_smc_table_manager(hwmgr,
+					     (uint8_t *)od_table,
+					     TABLE_OVERDRIVE, false);
 		PP_ASSERT_WITH_CODE(!ret,
 				"Failed to import overdrive table!",
 				return ret);
@@ -2847,8 +2852,8 @@ static int vega20_display_configuration_changed_task(struct pp_hwmgr *hwmgr)
 
 	if ((data->water_marks_bitmap & WaterMarksExist) &&
 	    !(data->water_marks_bitmap & WaterMarksLoaded)) {
-		result = vega20_copy_table_to_smc(hwmgr,
-			(uint8_t *)wm_table, TABLE_WATERMARKS);
+		result = smum_smc_table_manager(hwmgr,
+						(uint8_t *)wm_table, TABLE_WATERMARKS, false);
 		PP_ASSERT_WITH_CODE(!result,
 				"Failed to update WMTABLE!",
 				return result);
diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_thermal.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_thermal.c
index 2984ddd5428c..1c951a5d827d 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_thermal.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_thermal.c
@@ -37,10 +37,7 @@ static int vega20_get_current_rpm(struct pp_hwmgr *hwmgr, uint32_t *current_rpm)
 				PPSMC_MSG_GetCurrentRpm)) == 0,
 			"Attempt to get current RPM from SMC Failed!",
 			return ret);
-	PP_ASSERT_WITH_CODE((ret = vega20_read_arg_from_smc(hwmgr,
-			current_rpm)) == 0,
-			"Attempt to read current RPM from SMC Failed!",
-			return ret);
+	*current_rpm = smum_get_argument(hwmgr);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c
index f7e3bc22bb93..a74c5be1ec18 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c
@@ -724,11 +724,13 @@ static int smu8_start_smu(struct pp_hwmgr *hwmgr)
 	if (hwmgr->chip_id == CHIP_STONEY)
 		fw_to_check &= ~(UCODE_ID_SDMA1_MASK | UCODE_ID_CP_MEC_JT2_MASK);
 
-	ret = smu8_request_smu_load_fw(hwmgr);
-	if (ret)
-		pr_err("SMU firmware load failed\n");
+	smu8_request_smu_load_fw(hwmgr);
 
-	smu8_check_fw_load_finish(hwmgr, fw_to_check);
+	ret = smu8_check_fw_load_finish(hwmgr, fw_to_check);
+	if (ret) {
+		pr_err("SMU firmware load failed\n");
+		return ret;
+	}
 
 	ret = smu8_load_mec_firmware(hwmgr);
 	if (ret)
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c
index 5d19115f410c..c81acc3192ad 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.c
@@ -88,8 +88,18 @@ static int vega10_copy_table_to_smc(struct pp_hwmgr *hwmgr,
 	return 0;
 }
 
-static int vega10_get_smc_features(struct pp_hwmgr *hwmgr,
-		uint32_t *features_enabled)
+int vega10_enable_smc_features(struct pp_hwmgr *hwmgr,
+			       bool enable, uint32_t feature_mask)
+{
+	int msg = enable ? PPSMC_MSG_EnableSmuFeatures :
+			PPSMC_MSG_DisableSmuFeatures;
+
+	return smum_send_msg_to_smc_with_parameter(hwmgr,
+			msg, feature_mask);
+}
+
+int vega10_get_enabled_smc_features(struct pp_hwmgr *hwmgr,
+			    uint64_t *features_enabled)
 {
 	if (features_enabled == NULL)
 		return -EINVAL;
@@ -102,9 +112,9 @@ static int vega10_get_smc_features(struct pp_hwmgr *hwmgr,
 
 static bool vega10_is_dpm_running(struct pp_hwmgr *hwmgr)
 {
-	uint32_t features_enabled = 0;
+	uint64_t features_enabled = 0;
 
-	vega10_get_smc_features(hwmgr, &features_enabled);
+	vega10_get_enabled_smc_features(hwmgr, &features_enabled);
 
 	if (features_enabled & SMC_DPM_FEATURES)
 		return true;
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.h b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.h
index 424e868bc768..bad760f22624 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.h
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega10_smumgr.h
@@ -42,6 +42,10 @@ struct vega10_smumgr {
 	struct smu_table_array            smu_tables;
 };
 
+int vega10_enable_smc_features(struct pp_hwmgr *hwmgr,
+			       bool enable, uint32_t feature_mask);
+int vega10_get_enabled_smc_features(struct pp_hwmgr *hwmgr,
+				    uint64_t *features_enabled);
 
 #endif
 
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c
index 7f0e2109f40d..ddb801517667 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c
@@ -37,8 +37,8 @@
  * @param   hwmgr    the address of the HW manager
  * @param   table_id    the driver's table ID to copy from
  */
-int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id)
+static int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr,
+				      uint8_t *table, int16_t table_id)
 {
 	struct vega12_smumgr *priv =
 			(struct vega12_smumgr *)(hwmgr->smu_backend);
@@ -75,8 +75,8 @@ int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr,
  * @param   hwmgr    the address of the HW manager
  * @param   table_id    the table to copy from
  */
-int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id)
+static int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr,
+				    uint8_t *table, int16_t table_id)
 {
 	struct vega12_smumgr *priv =
 			(struct vega12_smumgr *)(hwmgr->smu_backend);
@@ -351,6 +351,19 @@ static int vega12_start_smu(struct pp_hwmgr *hwmgr)
 	return 0;
 }
 
+static int vega12_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table,
+				    uint16_t table_id, bool rw)
+{
+	int ret;
+
+	if (rw)
+		ret = vega12_copy_table_from_smc(hwmgr, table, table_id);
+	else
+		ret = vega12_copy_table_to_smc(hwmgr, table, table_id);
+
+	return ret;
+}
+
 const struct pp_smumgr_func vega12_smu_funcs = {
 	.smu_init = &vega12_smu_init,
 	.smu_fini = &vega12_smu_fini,
@@ -362,4 +375,5 @@ const struct pp_smumgr_func vega12_smu_funcs = {
 	.upload_pptable_settings = NULL,
 	.is_dpm_running = vega12_is_dpm_running,
 	.get_argument = smu9_get_argument,
+	.smc_table_manager = vega12_smc_table_manager,
 };
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.h b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.h
index b285cbc04019..aeec965ce81f 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.h
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.h
@@ -48,10 +48,6 @@ struct vega12_smumgr {
 #define SMU_FEATURES_HIGH_MASK       0xFFFFFFFF00000000
 #define SMU_FEATURES_HIGH_SHIFT      32
 
-int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id);
-int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id);
 int vega12_enable_smc_features(struct pp_hwmgr *hwmgr,
 		bool enable, uint64_t feature_mask);
 int vega12_get_enabled_smc_features(struct pp_hwmgr *hwmgr,
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c
index fe7f71079e0e..b7ff7d4d6f44 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c
@@ -148,19 +148,11 @@ static int vega20_send_msg_to_smc_with_parameter(struct pp_hwmgr *hwmgr,
 	return (ret == PPSMC_Result_OK) ? 0 : -EIO;
 }
 
-/*
- * Retrieve an argument from SMC.
- * @param    hwmgr  the address of the powerplay hardware manager.
- * @param    arg     pointer to store the argument from SMC.
- * @return   Always return 0.
- */
-int vega20_read_arg_from_smc(struct pp_hwmgr *hwmgr, uint32_t *arg)
+static uint32_t vega20_get_argument(struct pp_hwmgr *hwmgr)
 {
 	struct amdgpu_device *adev = hwmgr->adev;
 
-	*arg = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82);
-
-	return 0;
+	return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82);
 }
 
 /*
@@ -168,8 +160,8 @@ int vega20_read_arg_from_smc(struct pp_hwmgr *hwmgr, uint32_t *arg)
  * @param   hwmgr    the address of the HW manager
  * @param   table_id    the driver's table ID to copy from
  */
-int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id)
+static int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr,
+				      uint8_t *table, int16_t table_id)
 {
 	struct vega20_smumgr *priv =
 			(struct vega20_smumgr *)(hwmgr->smu_backend);
@@ -208,8 +200,8 @@ int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr,
  * @param   hwmgr    the address of the HW manager
  * @param   table_id    the table to copy from
  */
-int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id)
+static int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr,
+				    uint8_t *table, int16_t table_id)
 {
 	struct vega20_smumgr *priv =
 			(struct vega20_smumgr *)(hwmgr->smu_backend);
@@ -345,18 +337,12 @@ int vega20_get_enabled_smc_features(struct pp_hwmgr *hwmgr,
 			PPSMC_MSG_GetEnabledSmuFeaturesLow)) == 0,
 			"[GetEnabledSMCFeatures] Attemp to get SMU features Low failed!",
 			return ret);
-	PP_ASSERT_WITH_CODE((ret = vega20_read_arg_from_smc(hwmgr,
-			&smc_features_low)) == 0,
-			"[GetEnabledSMCFeatures] Attemp to read SMU features Low argument failed!",
-			return ret);
+	smc_features_low = vega20_get_argument(hwmgr);
 	PP_ASSERT_WITH_CODE((ret = vega20_send_msg_to_smc(hwmgr,
 			PPSMC_MSG_GetEnabledSmuFeaturesHigh)) == 0,
 			"[GetEnabledSMCFeatures] Attemp to get SMU features High failed!",
 			return ret);
-	PP_ASSERT_WITH_CODE((ret = vega20_read_arg_from_smc(hwmgr,
-			&smc_features_high)) == 0,
-			"[GetEnabledSMCFeatures] Attemp to read SMU features High argument failed!",
-			return ret);
+	smc_features_high = vega20_get_argument(hwmgr);
 
 	*features_enabled = ((((uint64_t)smc_features_low << SMU_FEATURES_LOW_SHIFT) & SMU_FEATURES_LOW_MASK) |
 			(((uint64_t)smc_features_high << SMU_FEATURES_HIGH_SHIFT) & SMU_FEATURES_HIGH_MASK));
@@ -574,6 +560,19 @@ static bool vega20_is_dpm_running(struct pp_hwmgr *hwmgr)
 		return false;
 }
 
+static int vega20_smc_table_manager(struct pp_hwmgr *hwmgr, uint8_t *table,
+				    uint16_t table_id, bool rw)
+{
+	int ret;
+
+	if (rw)
+		ret = vega20_copy_table_from_smc(hwmgr, table, table_id);
+	else
+		ret = vega20_copy_table_to_smc(hwmgr, table, table_id);
+
+	return ret;
+}
+
 const struct pp_smumgr_func vega20_smu_funcs = {
 	.smu_init = &vega20_smu_init,
 	.smu_fini = &vega20_smu_fini,
@@ -584,4 +583,6 @@ const struct pp_smumgr_func vega20_smu_funcs = {
 	.download_pptable_settings = NULL,
 	.upload_pptable_settings = NULL,
 	.is_dpm_running = vega20_is_dpm_running,
+	.get_argument = vega20_get_argument,
+	.smc_table_manager = vega20_smc_table_manager,
 };
diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h
index 505eb0d82e3b..77349c3f0162 100644
--- a/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h
+++ b/drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.h
@@ -47,11 +47,6 @@ struct vega20_smumgr {
 #define SMU_FEATURES_HIGH_MASK       0xFFFFFFFF00000000
 #define SMU_FEATURES_HIGH_SHIFT      32
 
-int vega20_read_arg_from_smc(struct pp_hwmgr *hwmgr, uint32_t *arg);
-int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id);
-int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr,
-		uint8_t *table, int16_t table_id);
 int vega20_enable_smc_features(struct pp_hwmgr *hwmgr,
 		bool enable, uint64_t feature_mask);
 int vega20_get_enabled_smc_features(struct pp_hwmgr *hwmgr,
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
index 69e9b431bf1f..e7c3ed6c9a2e 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
@@ -105,7 +105,7 @@ static void etnaviv_sched_timedout_job(struct drm_sched_job *sched_job)
 	change = dma_addr - gpu->hangcheck_dma_addr;
 	if (change < 0 || change > 16) {
 		gpu->hangcheck_dma_addr = dma_addr;
-		schedule_delayed_work(&sched_job->work_tdr,
+		schedule_delayed_work(&sched_job->sched->work_tdr,
 				      sched_job->sched->timeout);
 		return;
 	}
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index 9ca741f3a0bc..4e8505d51795 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -197,19 +197,15 @@ static void drm_sched_job_finish(struct work_struct *work)
 	 * manages to find this job as the next job in the list, the fence
 	 * signaled check below will prevent the timeout to be restarted.
 	 */
-	cancel_delayed_work_sync(&s_job->work_tdr);
+	cancel_delayed_work_sync(&sched->work_tdr);
 
 	spin_lock(&sched->job_list_lock);
-	/* queue TDR for next job */
-	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
-	    !list_is_last(&s_job->node, &sched->ring_mirror_list)) {
-		struct drm_sched_job *next = list_next_entry(s_job, node);
-
-		if (!dma_fence_is_signaled(&next->s_fence->finished))
-			schedule_delayed_work(&next->work_tdr, sched->timeout);
-	}
 	/* remove job from ring_mirror_list */
 	list_del(&s_job->node);
+	/* queue TDR for next job */
+	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
+	    !list_empty(&sched->ring_mirror_list))
+		schedule_delayed_work(&sched->work_tdr, sched->timeout);
 	spin_unlock(&sched->job_list_lock);
 
 	dma_fence_put(&s_job->s_fence->finished);
@@ -236,16 +232,21 @@ static void drm_sched_job_begin(struct drm_sched_job *s_job)
 	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
 	    list_first_entry_or_null(&sched->ring_mirror_list,
 				     struct drm_sched_job, node) == s_job)
-		schedule_delayed_work(&s_job->work_tdr, sched->timeout);
+		schedule_delayed_work(&sched->work_tdr, sched->timeout);
 	spin_unlock(&sched->job_list_lock);
 }
 
 static void drm_sched_job_timedout(struct work_struct *work)
 {
-	struct drm_sched_job *job = container_of(work, struct drm_sched_job,
-						 work_tdr.work);
+	struct drm_gpu_scheduler *sched;
+	struct drm_sched_job *job;
+
+	sched = container_of(work, struct drm_gpu_scheduler, work_tdr.work);
+	job = list_first_entry_or_null(&sched->ring_mirror_list,
+				       struct drm_sched_job, node);
 
-	job->sched->ops->timedout_job(job);
+	if (job)
+		job->sched->ops->timedout_job(job);
 }
 
 /**
@@ -315,7 +316,7 @@ void drm_sched_job_recovery(struct drm_gpu_scheduler *sched)
 	s_job = list_first_entry_or_null(&sched->ring_mirror_list,
 					 struct drm_sched_job, node);
 	if (s_job && sched->timeout != MAX_SCHEDULE_TIMEOUT)
-		schedule_delayed_work(&s_job->work_tdr, sched->timeout);
+		schedule_delayed_work(&sched->work_tdr, sched->timeout);
 
 	list_for_each_entry_safe(s_job, tmp, &sched->ring_mirror_list, node) {
 		struct drm_sched_fence *s_fence = s_job->s_fence;
@@ -384,7 +385,6 @@ int drm_sched_job_init(struct drm_sched_job *job,
 
 	INIT_WORK(&job->finish_work, drm_sched_job_finish);
 	INIT_LIST_HEAD(&job->node);
-	INIT_DELAYED_WORK(&job->work_tdr, drm_sched_job_timedout);
 
 	return 0;
 }
@@ -575,6 +575,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
 	INIT_LIST_HEAD(&sched->ring_mirror_list);
 	spin_lock_init(&sched->job_list_lock);
 	atomic_set(&sched->hw_rq_count, 0);
+	INIT_DELAYED_WORK(&sched->work_tdr, drm_sched_job_timedout);
 	atomic_set(&sched->num_jobs, 0);
 	atomic64_set(&sched->job_id_count, 0);
 
diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
index a5501581d96b..9243dea6e6ad 100644
--- a/drivers/gpu/drm/v3d/v3d_sched.c
+++ b/drivers/gpu/drm/v3d/v3d_sched.c
@@ -168,7 +168,7 @@ v3d_job_timedout(struct drm_sched_job *sched_job)
 		job->timedout_ctca = ctca;
 		job->timedout_ctra = ctra;
 
-		schedule_delayed_work(&job->base.work_tdr,
+		schedule_delayed_work(&job->base.sched->work_tdr,
 				      job->base.sched->timeout);
 		return;
 	}
diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index daec50f887b3..d87b268f1781 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -175,8 +175,6 @@ struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f);
  *               finished to remove the job from the
  *               @drm_gpu_scheduler.ring_mirror_list.
  * @node: used to append this struct to the @drm_gpu_scheduler.ring_mirror_list.
- * @work_tdr: schedules a delayed call to @drm_sched_job_timedout after the timeout
- *            interval is over.
  * @id: a unique id assigned to each job scheduled on the scheduler.
  * @karma: increment on every hang caused by this job. If this exceeds the hang
  *         limit of the scheduler then the job is marked guilty and will not
@@ -195,7 +193,6 @@ struct drm_sched_job {
 	struct dma_fence_cb		finish_cb;
 	struct work_struct		finish_work;
 	struct list_head		node;
-	struct delayed_work		work_tdr;
 	uint64_t			id;
 	atomic_t			karma;
 	enum drm_sched_priority		s_priority;
@@ -259,6 +256,8 @@ struct drm_sched_backend_ops {
  *                 finished.
  * @hw_rq_count: the number of jobs currently in the hardware queue.
  * @job_id_count: used to assign unique id to the each job.
+ * @work_tdr: schedules a delayed call to @drm_sched_job_timedout after the
+ *            timeout interval is over.
  * @thread: the kthread on which the scheduler which run.
  * @ring_mirror_list: the list of jobs which are currently in the job queue.
  * @job_list_lock: lock to protect the ring_mirror_list.
@@ -278,6 +277,7 @@ struct drm_gpu_scheduler {
 	wait_queue_head_t		job_scheduled;
 	atomic_t			hw_rq_count;
 	atomic64_t			job_id_count;
+	struct delayed_work		work_tdr;
 	struct task_struct		*thread;
 	struct list_head		ring_mirror_list;
 	spinlock_t			job_list_lock;
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h
index 01674b56e14f..f5ff8a76e208 100644
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@@ -82,6 +82,14 @@ struct kfd_ioctl_set_cu_mask_args {
 	__u64 cu_mask_ptr;		/* to KFD */
 };
 
+struct kfd_ioctl_get_queue_wave_state_args {
+	uint64_t ctl_stack_address;	/* to KFD */
+	uint32_t ctl_stack_used_size;	/* from KFD */
+	uint32_t save_area_used_size;	/* from KFD */
+	uint32_t queue_id;		/* to KFD */
+	uint32_t pad;
+};
+
 /* For kfd_ioctl_set_memory_policy_args.default_policy and alternate_policy */
 #define KFD_IOC_CACHE_POLICY_COHERENT 0
 #define KFD_IOC_CACHE_POLICY_NONCOHERENT 1
@@ -475,7 +483,10 @@ struct kfd_ioctl_unmap_memory_from_gpu_args {
 #define AMDKFD_IOC_SET_CU_MASK		\
 		AMDKFD_IOW(0x1A, struct kfd_ioctl_set_cu_mask_args)
 
+#define AMDKFD_IOC_GET_QUEUE_WAVE_STATE		\
+		AMDKFD_IOWR(0x1B, struct kfd_ioctl_get_queue_wave_state_args)
+
 #define AMDKFD_COMMAND_START		0x01
-#define AMDKFD_COMMAND_END		0x1B
+#define AMDKFD_COMMAND_END		0x1C
 
 #endif