Merge tag 'amd-drm-next-6.4-2023-04-14' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.4-2023-04-14: amdgpu: - S4 fixes for APUs - GFX11 fixes - Misc code cleanups - DCN 3.2 fixes - DCN 3.1.4 fixes - FPO/FAMS work to improve display power savings - DP fixes - UMC 8.10 code cleanup - SDMA v4 fix - GPU clock counter fixes - SMU 13 fixes - Sdma v6 invalidation fix for preemption - RAS fixes - S0ix fix - GC 9.4.3 updates amdkfd: - Fix user pointers with IOMMU - Fix coherency flag handling Signed-off-by: Dave Airlie <airlied@redhat.com> From: Alex Deucher <alexander.deucher@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230414204609.7942-1-alexander.deucher@amd.com
author: Dave Airlie <airlied@redhat.com> 2023-04-17 10:54:58 +1000
committer: Dave Airlie <airlied@redhat.com> 2023-04-17 10:54:59 +1000
commit: e82c98f2ca439356d5595ba8c9cd782f993f6f8c (patch)
tree: e07718a52c12f78e247c83f2d6ba14a0058fd819 /drivers/gpu
parent: afa351a15d80993f8ba6ae28652cc23127237e37 (diff)
parent: 541372bb62f289f4402cf55be51fb9cec7373627 (diff)
download: lwn-e82c98f2ca439356d5595ba8c9cd782f993f6f8c.tar.gz
lwn-e82c98f2ca439356d5595ba8c9cd782f993f6f8c.zip
107 files changed, 2118 insertions, 730 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index dd8bc53411bd..415a7fa395c4 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -136,6 +136,7 @@ amdgpu-y += \
 	gfx_v9_0.o \
 	gfx_v9_4.o \
 	gfx_v9_4_2.o \
+	gfx_v9_4_3.o \
 	gfx_v10_0.o \
 	imu_v11_0.o \
 	gfx_v11_0.o \
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 8cf2cc50b3de..02b827785e39 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -185,7 +185,6 @@ extern char *amdgpu_disable_cu;
 extern char *amdgpu_virtual_display;
 extern uint amdgpu_pp_feature_mask;
 extern uint amdgpu_force_long_training;
-extern int amdgpu_job_hang_limit;
 extern int amdgpu_lbpw;
 extern int amdgpu_compute_multipipe;
 extern int amdgpu_gpu_recovery;
@@ -471,7 +470,7 @@ int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv);
 /*
  * Writeback
  */
-#define AMDGPU_MAX_WB 256	/* Reserve at most 256 WB slots for amdgpu-owned rings. */
+#define AMDGPU_MAX_WB 1024	/* Reserve at most 1024 WB slots for amdgpu-owned rings. */
 
 struct amdgpu_wb {
 	struct amdgpu_bo	*wb_obj;
@@ -1222,7 +1221,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
 	((adev)->asic_funcs->flush_hdp ? (adev)->asic_funcs->flush_hdp((adev), (r)) : (adev)->hdp.funcs->flush_hdp((adev), (r)))
 #define amdgpu_asic_invalidate_hdp(adev, r) \
 	((adev)->asic_funcs->invalidate_hdp ? (adev)->asic_funcs->invalidate_hdp((adev), (r)) : \
-	 ((adev)->hdp.funcs->invalidate_hdp ? (adev)->hdp.funcs->invalidate_hdp((adev), (r)) : 0))
+	 ((adev)->hdp.funcs->invalidate_hdp ? (adev)->hdp.funcs->invalidate_hdp((adev), (r)) : (void)0))
 #define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
 #define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
 #define amdgpu_asic_get_pcie_usage(adev, cnt0, cnt1) ((adev)->asic_funcs->get_pcie_usage((adev), (cnt0), (cnt1)))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index 60b1857f469e..aeeec211861c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -981,7 +981,12 @@ static bool amdgpu_atcs_pci_probe_handle(struct pci_dev *pdev)
  */
 bool amdgpu_acpi_should_gpu_reset(struct amdgpu_device *adev)
 {
-	if (adev->flags & AMD_IS_APU)
+	if ((adev->flags & AMD_IS_APU) &&
+	    adev->gfx.imu.funcs) /* Not need to do mode2 reset for IMU enabled APUs */
+		return false;
+
+	if ((adev->flags & AMD_IS_APU) &&
+	    amdgpu_acpi_is_s3_active(adev))
 		return false;
 
 	if (amdgpu_sriov_vf(adev))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index f99d4873bf22..0385f7f69278 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -96,7 +96,7 @@ static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
 					 size_t *start_offset)
 {
 	/*
-	 * The first num_doorbells are used by amdgpu.
+	 * The first num_kernel_doorbells are used by amdgpu.
 	 * amdkfd takes whatever's left in the aperture.
 	 */
 	if (adev->enable_mes) {
@@ -109,11 +109,11 @@ static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
 		*aperture_base = adev->doorbell.base;
 		*aperture_size = 0;
 		*start_offset = 0;
-	} else if (adev->doorbell.size > adev->doorbell.num_doorbells *
+	} else if (adev->doorbell.size > adev->doorbell.num_kernel_doorbells *
 						sizeof(u32)) {
 		*aperture_base = adev->doorbell.base;
 		*aperture_size = adev->doorbell.size;
-		*start_offset = adev->doorbell.num_doorbells * sizeof(u32);
+		*start_offset = adev->doorbell.num_kernel_doorbells * sizeof(u32);
 	} else {
 		*aperture_base = 0;
 		*aperture_size = 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index c87515210c4f..83a83ced2439 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -82,6 +82,25 @@ static bool kfd_mem_is_attached(struct amdgpu_vm *avm,
 	return false;
 }
 
+/**
+ * reuse_dmamap() - Check whether adev can share the original
+ * userptr BO
+ *
+ * If both adev and bo_adev are in direct mapping or
+ * in the same iommu group, they can share the original BO.
+ *
+ * @adev: Device to which can or cannot share the original BO
+ * @bo_adev: Device to which allocated BO belongs to
+ *
+ * Return: returns true if adev can share original userptr BO,
+ * false otherwise.
+ */
+static bool reuse_dmamap(struct amdgpu_device *adev, struct amdgpu_device *bo_adev)
+{
+	return (adev->ram_is_direct_mapped && bo_adev->ram_is_direct_mapped) ||
+			(adev->dev->iommu_group == bo_adev->dev->iommu_group);
+}
+
 /* Set memory usage limits. Current, limits are
  *  System (TTM + userptr) memory - 15/16th System RAM
  *  TTM memory - 3/8th System RAM
@@ -253,15 +272,19 @@ create_dmamap_sg_bo(struct amdgpu_device *adev,
 		 struct kgd_mem *mem, struct amdgpu_bo **bo_out)
 {
 	struct drm_gem_object *gem_obj;
-	int ret, align;
+	int ret;
+	uint64_t flags = 0;
 
 	ret = amdgpu_bo_reserve(mem->bo, false);
 	if (ret)
 		return ret;
 
-	align = 1;
-	ret = amdgpu_gem_object_create(adev, mem->bo->tbo.base.size, align,
-			AMDGPU_GEM_DOMAIN_CPU, AMDGPU_GEM_CREATE_PREEMPTIBLE,
+	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR)
+		flags |= mem->bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+					AMDGPU_GEM_CREATE_UNCACHED);
+
+	ret = amdgpu_gem_object_create(adev, mem->bo->tbo.base.size, 1,
+			AMDGPU_GEM_DOMAIN_CPU, AMDGPU_GEM_CREATE_PREEMPTIBLE | flags,
 			ttm_bo_type_sg, mem->bo->tbo.base.resv, &gem_obj);
 
 	amdgpu_bo_unreserve(mem->bo);
@@ -481,9 +504,6 @@ kfd_mem_dmamap_userptr(struct kgd_mem *mem,
 	if (unlikely(ret))
 		goto release_sg;
 
-	drm_prime_sg_to_dma_addr_array(ttm->sg, ttm->dma_address,
-				       ttm->num_pages);
-
 	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
 	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 	if (ret)
@@ -805,11 +825,11 @@ static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem,
 			 va + bo_size, vm);
 
 		if ((adev == bo_adev && !(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) ||
-		    (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm) && adev->ram_is_direct_mapped) ||
-		    same_hive) {
+		    (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm) && reuse_dmamap(adev, bo_adev)) ||
+			same_hive) {
 			/* Mappings on the local GPU, or VRAM mappings in the
-			 * local hive, or userptr mapping IOMMU direct map mode
-			 * share the original BO
+			 * local hive, or userptr mapping can reuse dma map
+			 * address space share the original BO
 			 */
 			attachment[i]->type = KFD_MEM_ATT_SHARED;
 			bo[i] = mem->bo;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index fac9312b1695..a2292acf06d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -602,7 +602,7 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
 	if (amdgpu_device_skip_hw_access(adev))
 		return 0;
 
-	if (index < adev->doorbell.num_doorbells) {
+	if (index < adev->doorbell.num_kernel_doorbells) {
 		return readl(adev->doorbell.ptr + index);
 	} else {
 		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
@@ -625,7 +625,7 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
 	if (amdgpu_device_skip_hw_access(adev))
 		return;
 
-	if (index < adev->doorbell.num_doorbells) {
+	if (index < adev->doorbell.num_kernel_doorbells) {
 		writel(v, adev->doorbell.ptr + index);
 	} else {
 		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
@@ -646,7 +646,7 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
 	if (amdgpu_device_skip_hw_access(adev))
 		return 0;
 
-	if (index < adev->doorbell.num_doorbells) {
+	if (index < adev->doorbell.num_kernel_doorbells) {
 		return atomic64_read((atomic64_t *)(adev->doorbell.ptr + index));
 	} else {
 		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
@@ -669,7 +669,7 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
 	if (amdgpu_device_skip_hw_access(adev))
 		return;
 
-	if (index < adev->doorbell.num_doorbells) {
+	if (index < adev->doorbell.num_kernel_doorbells) {
 		atomic64_set((atomic64_t *)(adev->doorbell.ptr + index), v);
 	} else {
 		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
@@ -1060,7 +1060,7 @@ static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
 	if (adev->asic_type < CHIP_BONAIRE) {
 		adev->doorbell.base = 0;
 		adev->doorbell.size = 0;
-		adev->doorbell.num_doorbells = 0;
+		adev->doorbell.num_kernel_doorbells = 0;
 		adev->doorbell.ptr = NULL;
 		return 0;
 	}
@@ -1075,27 +1075,27 @@ static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
 	adev->doorbell.size = pci_resource_len(adev->pdev, 2);
 
 	if (adev->enable_mes) {
-		adev->doorbell.num_doorbells =
+		adev->doorbell.num_kernel_doorbells =
 			adev->doorbell.size / sizeof(u32);
 	} else {
-		adev->doorbell.num_doorbells =
+		adev->doorbell.num_kernel_doorbells =
 			min_t(u32, adev->doorbell.size / sizeof(u32),
 			      adev->doorbell_index.max_assignment+1);
-		if (adev->doorbell.num_doorbells == 0)
+		if (adev->doorbell.num_kernel_doorbells == 0)
 			return -EINVAL;
 
 		/* For Vega, reserve and map two pages on doorbell BAR since SDMA
 		 * paging queue doorbell use the second page. The
 		 * AMDGPU_DOORBELL64_MAX_ASSIGNMENT definition assumes all the
 		 * doorbells are in the first page. So with paging queue enabled,
-		 * the max num_doorbells should + 1 page (0x400 in dword)
+		 * the max num_kernel_doorbells should + 1 page (0x400 in dword)
 		 */
 		if (adev->asic_type >= CHIP_VEGA10)
-			adev->doorbell.num_doorbells += 0x400;
+			adev->doorbell.num_kernel_doorbells += 0x400;
 	}
 
 	adev->doorbell.ptr = ioremap(adev->doorbell.base,
-				     adev->doorbell.num_doorbells *
+				     adev->doorbell.num_kernel_doorbells *
 				     sizeof(u32));
 	if (adev->doorbell.ptr == NULL)
 		return -ENOMEM;
@@ -2184,7 +2184,6 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		adev->has_pr3 = parent ? pci_pr3_present(parent) : false;
 	}
 
-	amdgpu_amdkfd_device_probe(adev);
 
 	adev->pm.pp_feature = amdgpu_pp_feature_mask;
 	if (amdgpu_sriov_vf(adev) || sched_policy == KFD_SCHED_POLICY_NO_HWS)
@@ -2240,6 +2239,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	if (!total)
 		return -ENODEV;
 
+	amdgpu_amdkfd_device_probe(adev);
 	adev->cg_flags &= amdgpu_cg_mask;
 	adev->pg_flags &= amdgpu_pg_mask;
 
@@ -2365,7 +2365,7 @@ static int amdgpu_device_init_schedulers(struct amdgpu_device *adev)
 		}
 
 		r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
-				   ring->num_hw_submission, amdgpu_job_hang_limit,
+				   ring->num_hw_submission, 0,
 				   timeout, adev->reset_domain->wq,
 				   ring->sched_score, ring->name,
 				   adev->dev);
@@ -3305,9 +3305,11 @@ static int amdgpu_device_ip_resume(struct amdgpu_device *adev)
 {
 	int r;
 
-	r = amdgpu_amdkfd_resume_iommu(adev);
-	if (r)
-		return r;
+	if (!adev->in_s0ix) {
+		r = amdgpu_amdkfd_resume_iommu(adev);
+		if (r)
+			return r;
+	}
 
 	r = amdgpu_device_ip_resume_phase1(adev);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
index 5139334925ea..0ecce0b92b82 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@@ -1502,6 +1502,7 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(9, 4, 0):
 	case IP_VERSION(9, 4, 1):
 	case IP_VERSION(9, 4, 2):
+	case IP_VERSION(9, 4, 3):
 		amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
 		break;
 	case IP_VERSION(10, 1, 10):
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
index 7199b6b0be81..8fd11497faba 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
@@ -21,6 +21,9 @@
  *
  */
 
+#ifndef AMDGPU_DOORBELL_H
+#define AMDGPU_DOORBELL_H
+
 /*
  * GPU doorbell structures, functions & helpers
  */
@@ -29,7 +32,9 @@ struct amdgpu_doorbell {
 	resource_size_t		base;
 	resource_size_t		size;
 	u32 __iomem		*ptr;
-	u32			num_doorbells;	/* Number of doorbells actually reserved for amdgpu. */
+
+	/* Number of doorbells reserved for amdgpu kernel driver */
+	u32 num_kernel_doorbells;
 };
 
 /* Reserved doorbells for amdgpu (including multimedia).
@@ -306,3 +311,4 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
 #define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
 #define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
 
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index b4189d669b54..b1ca1ab6d6ad 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -157,7 +157,6 @@ char *amdgpu_virtual_display;
  */
 uint amdgpu_pp_feature_mask = 0xfff7bfff;
 uint amdgpu_force_long_training;
-int amdgpu_job_hang_limit;
 int amdgpu_lbpw = -1;
 int amdgpu_compute_multipipe = -1;
 int amdgpu_gpu_recovery = -1; /* auto */
@@ -521,13 +520,6 @@ MODULE_PARM_DESC(virtual_display,
 module_param_named(virtual_display, amdgpu_virtual_display, charp, 0444);
 
 /**
- * DOC: job_hang_limit (int)
- * Set how much time allow a job hang and not drop it. The default is 0.
- */
-MODULE_PARM_DESC(job_hang_limit, "how much time allow a job hang and not drop it (default 0)");
-module_param_named(job_hang_limit, amdgpu_job_hang_limit, int ,0444);
-
-/**
  * DOC: lbpw (int)
  * Override Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable). The default is -1 (auto, enabled).
  */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index c50d59855011..9d3a0542c996 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -305,6 +305,7 @@ int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
 	ring->doorbell_index = adev->doorbell_index.kiq;
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 
 	r = amdgpu_gfx_kiq_acquire(adev, ring);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index de9e7a00bb15..bfabea76d166 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -42,6 +42,8 @@
 #define AMDGPU_GFX_CG_DISABLED_MODE		0x00000004L
 #define AMDGPU_GFX_LBPW_DISABLED_MODE		0x00000008L
 
+#define AMDGPU_MAX_GC_INSTANCES		8
+
 #define AMDGPU_MAX_GFX_QUEUES KGD_MAX_QUEUES
 #define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES
 
@@ -53,6 +55,15 @@ enum amdgpu_gfx_pipe_priority {
 #define AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM  0
 #define AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM  15
 
+enum amdgpu_gfx_partition {
+	AMDGPU_SPX_PARTITION_MODE = 0,
+	AMDGPU_DPX_PARTITION_MODE = 1,
+	AMDGPU_TPX_PARTITION_MODE = 2,
+	AMDGPU_QPX_PARTITION_MODE = 3,
+	AMDGPU_CPX_PARTITION_MODE = 4,
+	AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE,
+};
+
 struct amdgpu_mec {
 	struct amdgpu_bo	*hpd_eop_obj;
 	u64			hpd_eop_gpu_addr;
@@ -323,7 +334,7 @@ struct amdgpu_gfx {
 	bool				cp_fw_write_wait;
 	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
 	unsigned			num_gfx_rings;
-	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];
+	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES];
 	unsigned			num_compute_rings;
 	struct amdgpu_irq_src		eop_irq;
 	struct amdgpu_irq_src		priv_reg_irq;
@@ -364,6 +375,10 @@ struct amdgpu_gfx {
 
 	struct amdgpu_ring		sw_gfx_ring[AMDGPU_MAX_SW_GFX_RINGS];
 	struct amdgpu_ring_mux          muxer;
+
+	enum amdgpu_gfx_partition	partition_mode;
+	uint32_t			num_xcd;
+	uint32_t			num_xcc_per_xcp;
 };
 
 #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
index 88bc7f5f46e6..4e2531758866 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
@@ -395,8 +395,21 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev,
 	while (fault->timestamp >= stamp) {
 		uint64_t tmp;
 
-		if (atomic64_read(&fault->key) == key)
-			return true;
+		if (atomic64_read(&fault->key) == key) {
+			/*
+			 * if we get a fault which is already present in
+			 * the fault_ring and the timestamp of
+			 * the fault is after the expired timestamp,
+			 * then this is a new fault that needs to be added
+			 * into the fault ring.
+			 */
+			if (fault->timestamp_expiry != 0 &&
+			    amdgpu_ih_ts_after(fault->timestamp_expiry,
+					       timestamp))
+				break;
+			else
+				return true;
+		}
 
 		tmp = fault->timestamp;
 		fault = &gmc->fault_ring[fault->next];
@@ -432,15 +445,32 @@ void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr,
 {
 	struct amdgpu_gmc *gmc = &adev->gmc;
 	uint64_t key = amdgpu_gmc_fault_key(addr, pasid);
+	struct amdgpu_ih_ring *ih;
 	struct amdgpu_gmc_fault *fault;
+	uint32_t last_wptr;
+	uint64_t last_ts;
 	uint32_t hash;
 	uint64_t tmp;
 
+	ih = adev->irq.retry_cam_enabled ? &adev->irq.ih_soft : &adev->irq.ih1;
+	/* Get the WPTR of the last entry in IH ring */
+	last_wptr = amdgpu_ih_get_wptr(adev, ih);
+	/* Order wptr with ring data. */
+	rmb();
+	/* Get the timetamp of the last entry in IH ring */
+	last_ts = amdgpu_ih_decode_iv_ts(adev, ih, last_wptr, -1);
+
 	hash = hash_64(key, AMDGPU_GMC_FAULT_HASH_ORDER);
 	fault = &gmc->fault_ring[gmc->fault_hash[hash].idx];
 	do {
-		if (atomic64_cmpxchg(&fault->key, key, 0) == key)
+		if (atomic64_read(&fault->key) == key) {
+			/*
+			 * Update the timestamp when this fault
+			 * expired.
+			 */
+			fault->timestamp_expiry = last_ts;
 			break;
+		}
 
 		tmp = fault->timestamp;
 		fault = &gmc->fault_ring[fault->next];
@@ -524,7 +554,7 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->num_rings; ++i) {
 		ring = adev->rings[i];
-		vmhub = ring->funcs->vmhub;
+		vmhub = ring->vm_hub;
 
 		if (ring == &adev->mes.ring)
 			continue;
@@ -540,7 +570,7 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 		vm_inv_engs[vmhub] &= ~(1 << ring->vm_inv_eng);
 
 		dev_info(adev->dev, "ring %s uses VM inv eng %u on hub %u\n",
-			 ring->name, ring->vm_inv_eng, ring->funcs->vmhub);
+			 ring->name, ring->vm_inv_eng, ring->vm_hub);
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
index 232523e3e270..6d105d7fb98b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -70,6 +70,7 @@ struct amdgpu_gmc_fault {
 	uint64_t	timestamp:48;
 	uint64_t	next:AMDGPU_GMC_FAULT_RING_ORDER;
 	atomic64_t	key;
+	uint64_t	timestamp_expiry:48;
 };
 
 /*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index df7eb0b7c4b9..4ff348e10e4d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -267,7 +267,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	if (r) {
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
 		if (job && job->vmid)
-			amdgpu_vmid_reset(adev, ring->funcs->vmhub, job->vmid);
+			amdgpu_vmid_reset(adev, ring->vm_hub, job->vmid);
 		amdgpu_ring_undo(ring);
 		return r;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
index 3f07b1a2ce47..c991ca0b7a1c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
@@ -202,7 +202,7 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_vm *vm,
 				 struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	struct dma_fence **fences;
 	unsigned i;
@@ -277,7 +277,7 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm,
 				     struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	uint64_t fence_context = adev->fence_context + ring->idx;
 	bool needs_flush = vm->use_cpu_for_update;
@@ -338,7 +338,7 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm,
 				 struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	uint64_t fence_context = adev->fence_context + ring->idx;
 	uint64_t updates = amdgpu_vm_tlb_seq(vm);
@@ -398,7 +398,7 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		     struct amdgpu_job *job, struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	struct amdgpu_vmid *idle = NULL;
 	struct amdgpu_vmid *id = NULL;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index e9f2c11ea416..be243adf3e65 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -98,6 +98,8 @@ struct amdgpu_irq {
 	struct irq_domain		*domain; /* GPU irq controller domain */
 	unsigned			virq[AMDGPU_MAX_IRQ_SRC_ID];
 	uint32_t                        srbm_soft_reset;
+	u32                             retry_cam_doorbell_index;
+	bool                            retry_cam_enabled;
 };
 
 void amdgpu_irq_disable_all(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
index 0e55823ef6ca..f0f00466b59f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -1434,13 +1434,31 @@ int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe)
 	struct amdgpu_firmware_info *info;
 	char ucode_prefix[30];
 	char fw_name[40];
+	bool need_retry = false;
 	int r;
 
-	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin",
-		ucode_prefix,
-		pipe == AMDGPU_MES_SCHED_PIPE ? "" : "1");
+	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix,
+				       sizeof(ucode_prefix));
+	if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(11, 0, 0)) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin",
+			 ucode_prefix,
+			 pipe == AMDGPU_MES_SCHED_PIPE ? "_2" : "1");
+		need_retry = true;
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin",
+			 ucode_prefix,
+			 pipe == AMDGPU_MES_SCHED_PIPE ? "" : "1");
+	}
+
 	r = amdgpu_ucode_request(adev, &adev->mes.fw[pipe], fw_name);
+	if (r && need_retry && pipe == AMDGPU_MES_SCHED_PIPE) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes.bin",
+			 ucode_prefix);
+		DRM_INFO("try to fall back to %s\n", fw_name);
+		r = amdgpu_ucode_request(adev, &adev->mes.fw[pipe],
+					 fw_name);
+	}
+
 	if (r)
 		goto out;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 4069bce9479f..3ab8a88789c8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2341,7 +2341,6 @@ static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev)
 	if (amdgpu_sriov_vf(adev)) {
 		switch (adev->ip_versions[MP0_HWIP][0]) {
 		case IP_VERSION(13, 0, 2):
-		case IP_VERSION(13, 0, 10):
 			return true;
 		default:
 			return false;
@@ -2430,6 +2429,13 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
 			else
 				adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
 							1 << AMDGPU_RAS_BLOCK__JPEG);
+
+			/*
+			 * XGMI RAS is not supported if xgmi num physical nodes
+			 * is zero
+			 */
+			if (!adev->gmc.xgmi.num_physical_nodes)
+				adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL);
 		} else {
 			dev_info(adev->dev, "SRAM ECC is not presented.\n");
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index ef38f4c93df0..17b3d1992e80 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -583,6 +583,10 @@ amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) {
 		return TA_RAS_BLOCK__FUSE;
 	case AMDGPU_RAS_BLOCK__MCA:
 		return TA_RAS_BLOCK__MCA;
+	case AMDGPU_RAS_BLOCK__VCN:
+		return TA_RAS_BLOCK__VCN;
+	case AMDGPU_RAS_BLOCK__JPEG:
+		return TA_RAS_BLOCK__JPEG;
 	default:
 		WARN_ONCE(1, "RAS ERROR: unexpected block id %d\n", block);
 		return TA_RAS_BLOCK__UMC;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
index 3106fa8a15ef..c2c2a7718613 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
@@ -107,47 +107,12 @@
 
 static bool __is_ras_eeprom_supported(struct amdgpu_device *adev)
 {
-	if (adev->asic_type == CHIP_IP_DISCOVERY) {
-		switch (adev->ip_versions[MP1_HWIP][0]) {
-		case IP_VERSION(13, 0, 0):
-		case IP_VERSION(13, 0, 10):
-			return true;
-		default:
-			return false;
-		}
-	}
-
-	return  adev->asic_type == CHIP_VEGA20 ||
-		adev->asic_type == CHIP_ARCTURUS ||
-		adev->asic_type == CHIP_SIENNA_CICHLID ||
-		adev->asic_type == CHIP_ALDEBARAN;
-}
-
-static bool __get_eeprom_i2c_addr_arct(struct amdgpu_device *adev,
-				       struct amdgpu_ras_eeprom_control *control)
-{
-	struct atom_context *atom_ctx = adev->mode_info.atom_context;
-
-	if (!control || !atom_ctx)
-		return false;
-
-	if (strnstr(atom_ctx->vbios_version,
-	            "D342",
-		    sizeof(atom_ctx->vbios_version)))
-		control->i2c_address = EEPROM_I2C_MADDR_0;
-	else
-		control->i2c_address = EEPROM_I2C_MADDR_4;
-
-	return true;
-}
-
-static bool __get_eeprom_i2c_addr_ip_discovery(struct amdgpu_device *adev,
-				       struct amdgpu_ras_eeprom_control *control)
-{
 	switch (adev->ip_versions[MP1_HWIP][0]) {
+	case IP_VERSION(11, 0, 2): /* VEGA20 and ARCTURUS */
+	case IP_VERSION(11, 0, 7): /* Sienna cichlid */
 	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 2): /* Aldebaran */
 	case IP_VERSION(13, 0, 10):
-		control->i2c_address = EEPROM_I2C_MADDR_4;
 		return true;
 	default:
 		return false;
@@ -178,29 +143,32 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev,
 		return true;
 	}
 
-	switch (adev->asic_type) {
-	case CHIP_VEGA20:
-		control->i2c_address = EEPROM_I2C_MADDR_0;
+	switch (adev->ip_versions[MP1_HWIP][0]) {
+	case IP_VERSION(11, 0, 2):
+		/* VEGA20 and ARCTURUS */
+		if (adev->asic_type == CHIP_VEGA20)
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		else if (strnstr(atom_ctx->vbios_version,
+				 "D342",
+				 sizeof(atom_ctx->vbios_version)))
+			control->i2c_address = EEPROM_I2C_MADDR_0;
+		else
+			control->i2c_address = EEPROM_I2C_MADDR_4;
 		return true;
-
-	case CHIP_ARCTURUS:
-		return __get_eeprom_i2c_addr_arct(adev, control);
-
-	case CHIP_SIENNA_CICHLID:
+	case IP_VERSION(11, 0, 7):
 		control->i2c_address = EEPROM_I2C_MADDR_0;
 		return true;
-
-	case CHIP_ALDEBARAN:
+	case IP_VERSION(13, 0, 2):
 		if (strnstr(atom_ctx->vbios_version, "D673",
 			    sizeof(atom_ctx->vbios_version)))
 			control->i2c_address = EEPROM_I2C_MADDR_4;
 		else
 			control->i2c_address = EEPROM_I2C_MADDR_0;
 		return true;
-
-	case CHIP_IP_DISCOVERY:
-		return __get_eeprom_i2c_addr_ip_discovery(adev, control);
-
+	case IP_VERSION(13, 0, 0):
+	case IP_VERSION(13, 0, 10):
+		control->i2c_address = EEPROM_I2C_MADDR_4;
+		return true;
 	default:
 		return false;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index 018f36b10de8..d8749444b689 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -165,7 +165,6 @@ struct amdgpu_ring_funcs {
 	bool			support_64bit_ptrs;
 	bool			no_user_fence;
 	bool			secure_submission_supported;
-	unsigned		vmhub;
 	unsigned		extra_dw;
 
 	/* ring read/write ptr handling */
@@ -250,6 +249,7 @@ struct amdgpu_ring {
 	uint64_t		ptr_mask;
 	uint32_t		buf_mask;
 	u32			idx;
+	u32			xcc_id;
 	u32			me;
 	u32			pipe;
 	u32			queue;
@@ -275,6 +275,7 @@ struct amdgpu_ring {
 	unsigned		cond_exe_offs;
 	u64			cond_exe_gpu_addr;
 	volatile u32		*cond_exe_cpu_addr;
+	unsigned		vm_hub;
 	unsigned		vm_inv_eng;
 	struct dma_fence	*vmid_wait;
 	bool			has_compute_vm_bug;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 98d91ebf5c26..525dffbe046a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -233,7 +233,7 @@ TRACE_EVENT(amdgpu_vm_grab_id,
 			   __entry->pasid = vm->pasid;
 			   __assign_str(ring, ring->name);
 			   __entry->vmid = job->vmid;
-			   __entry->vm_hub = ring->funcs->vmhub,
+			   __entry->vm_hub = ring->vm_hub,
 			   __entry->pd_addr = job->vm_pd_addr;
 			   __entry->needs_flush = job->vm_needs_flush;
 			   ),
@@ -427,7 +427,7 @@ TRACE_EVENT(amdgpu_vm_flush,
 	    TP_fast_assign(
 			   __assign_str(ring, ring->name);
 			   __entry->vmid = vmid;
-			   __entry->vm_hub = ring->funcs->vmhub;
+			   __entry->vm_hub = ring->vm_hub;
 			   __entry->pd_addr = pd_addr;
 			   ),
 	    TP_printk("ring=%s, id=%u, hub=%u, pd_addr=%010Lx",
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index 9e2e97207e53..1edf8e6aeb16 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -302,3 +302,34 @@ void amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
 
 	err_data->err_addr_cnt++;
 }
+
+int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
+			umc_func func, void *data)
+{
+	uint32_t node_inst       = 0;
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	int ret = 0;
+
+	if (adev->umc.node_inst_num) {
+		LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
+			ret = func(adev, node_inst, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev, "Node %d umc %d ch %d func returns %d\n",
+					node_inst, umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	} else {
+		LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+			ret = func(adev, 0, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev, "Umc %d ch %d func returns %d\n",
+					umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index d7f1229ff11f..86133f77a9a4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -47,6 +47,10 @@
 #define LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) \
 		LOOP_UMC_NODE_INST((node_inst)) LOOP_UMC_INST_AND_CH((umc_inst), (ch_inst))
 
+
+typedef int (*umc_func)(struct amdgpu_device *adev, uint32_t node_inst,
+			uint32_t umc_inst, uint32_t ch_inst, void *data);
+
 struct amdgpu_umc_ras {
 	struct amdgpu_ras_block_object ras_block;
 	void (*err_cnt_init)(struct amdgpu_device *adev);
@@ -104,4 +108,7 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 		struct amdgpu_iv_entry *entry);
 int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
 			uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst);
+
+int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
+			umc_func func, void *data);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 286e326bb4bd..3c0310576b3b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -483,7 +483,7 @@ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
 				  struct amdgpu_job *job)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 
 	if (job->vmid == 0)
@@ -517,7 +517,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 		    bool need_pipe_sync)
 {
 	struct amdgpu_device *adev = ring->adev;
-	unsigned vmhub = ring->funcs->vmhub;
+	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	struct amdgpu_vmid *id = &id_mgr->ids[job->vmid];
 	bool spm_update_needed = job->spm_update_needed;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index 516409989235..f5b5ce1051a2 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -4461,6 +4461,7 @@ static int gfx_v10_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
 		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
 	else
 		ring->doorbell_index = adev->doorbell_index.gfx_ring1 << 1;
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 	sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe;
@@ -4489,6 +4490,7 @@ static int gfx_v10_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
 	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
 				+ (ring_id * GFX10_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
@@ -9249,7 +9251,6 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = {
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v10_0_ring_get_rptr_gfx,
 	.get_wptr = gfx_v10_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v10_0_ring_set_wptr_gfx,
@@ -9304,7 +9305,6 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v10_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v10_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v10_0_ring_set_wptr_compute,
@@ -9340,7 +9340,6 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v10_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v10_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v10_0_ring_set_wptr_compute,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index ecf8ceb53311..a9da0486467a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -866,6 +866,7 @@ static int gfx_v11_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
 		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
 	else
 		ring->doorbell_index = adev->doorbell_index.gfx_ring1 << 1;
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 	sprintf(ring->name, "gfx_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + ring->pipe;
@@ -896,6 +897,7 @@ static int gfx_v11_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
 	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
 				+ (ring_id * GFX11_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
@@ -4671,11 +4673,24 @@ static int gfx_v11_0_post_soft_reset(void *handle)
 static uint64_t gfx_v11_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 {
 	uint64_t clock;
+	uint64_t clock_counter_lo, clock_counter_hi_pre, clock_counter_hi_after;
 
 	amdgpu_gfx_off_ctrl(adev, false);
 	mutex_lock(&adev->gfx.gpu_clock_mutex);
-	clock = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER) |
-		((uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER) << 32ULL);
+	if (amdgpu_sriov_vf(adev)) {
+		clock_counter_hi_pre = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_HI);
+		clock_counter_lo = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_LO);
+		clock_counter_hi_after = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_HI);
+		if (clock_counter_hi_pre != clock_counter_hi_after)
+			clock_counter_lo = (uint64_t)RREG32_SOC15(GC, 0, regCP_MES_MTIME_LO);
+	} else {
+		clock_counter_hi_pre = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+		clock_counter_lo = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+		clock_counter_hi_after = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_UPPER);
+		if (clock_counter_hi_pre != clock_counter_hi_after)
+			clock_counter_lo = (uint64_t)RREG32_SOC15(SMUIO, 0, regGOLDEN_TSC_COUNT_LOWER);
+	}
+	clock = clock_counter_lo | (clock_counter_hi_after << 32ULL);
 	mutex_unlock(&adev->gfx.gpu_clock_mutex);
 	amdgpu_gfx_off_ctrl(adev, true);
 	return clock;
@@ -6191,7 +6206,6 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = {
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v11_0_ring_get_rptr_gfx,
 	.get_wptr = gfx_v11_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v11_0_ring_set_wptr_gfx,
@@ -6239,7 +6253,6 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_compute = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v11_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v11_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v11_0_ring_set_wptr_compute,
@@ -6275,7 +6288,6 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_kiq = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v11_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v11_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v11_0_ring_set_wptr_compute,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
index b07a72ca25d9..068b9586a223 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
@@ -62,10 +62,18 @@ static int gfx_v11_0_3_rlc_gc_fed_irq(struct amdgpu_device *adev,
 		return -EINVAL;
 	}
 
-	ih_data.head = *ras_if;
-
 	dev_warn(adev->dev, "RLC %s FED IRQ\n", ras_if->name);
-	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		ih_data.head = *ras_if;
+		amdgpu_ras_interrupt_dispatch(adev, &ih_data);
+	} else {
+		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
+			adev->virt.ops->ras_poison_handler(adev);
+		else
+			dev_warn(adev->dev,
+				"No ras_poison_handler interface in SRIOV for %s!\n", ras_if->name);
+	}
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index ae09fc1cfe6b..adbcd8127c82 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -149,6 +149,16 @@ MODULE_FIRMWARE("amdgpu/aldebaran_sjt_mec2.bin");
 #define mmGOLDEN_TSC_COUNT_LOWER_Renoir                0x0026
 #define mmGOLDEN_TSC_COUNT_LOWER_Renoir_BASE_IDX       1
 
+#define mmGOLDEN_TSC_COUNT_UPPER_Raven   0x007a
+#define mmGOLDEN_TSC_COUNT_UPPER_Raven_BASE_IDX 0
+#define mmGOLDEN_TSC_COUNT_LOWER_Raven   0x007b
+#define mmGOLDEN_TSC_COUNT_LOWER_Raven_BASE_IDX 0
+
+#define mmGOLDEN_TSC_COUNT_UPPER_Raven2   0x0068
+#define mmGOLDEN_TSC_COUNT_UPPER_Raven2_BASE_IDX 0
+#define mmGOLDEN_TSC_COUNT_LOWER_Raven2   0x0069
+#define mmGOLDEN_TSC_COUNT_LOWER_Raven2_BASE_IDX 0
+
 enum ta_ras_gfx_subblock {
 	/*CPC*/
 	TA_RAS_BLOCK__GFX_CPC_INDEX_START = 0,
@@ -1995,6 +2005,7 @@ static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
 	ring->doorbell_index = (adev->doorbell_index.mec_ring0 + ring_id) << 1;
 	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
 				+ (ring_id * GFX9_MEC_HPD_SIZE);
+	ring->vm_hub = AMDGPU_GFXHUB_0;
 	sprintf(ring->name, "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
 
 	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
@@ -2094,6 +2105,7 @@ static int gfx_v9_0_sw_init(void *handle)
 
 		/* disable scheduler on the real ring */
 		ring->no_scheduler = true;
+		ring->vm_hub = AMDGPU_GFXHUB_0;
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
 				     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP,
 				     AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -2111,6 +2123,7 @@ static int gfx_v9_0_sw_init(void *handle)
 			ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
 			ring->is_sw_ring = true;
 			hw_prio = amdgpu_sw_ring_priority(i);
+			ring->vm_hub = AMDGPU_GFXHUB_0;
 			r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
 					     AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP, hw_prio,
 					     NULL);
@@ -3988,6 +4001,36 @@ static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 		preempt_enable();
 		clock = clock_lo | (clock_hi << 32ULL);
 		break;
+	case IP_VERSION(9, 1, 0):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_UPPER_Raven);
+		clock_lo = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_LOWER_Raven);
+		hi_check = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_UPPER_Raven);
+		/* The PWR TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_LOWER_Raven);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
+	case IP_VERSION(9, 2, 2):
+		preempt_disable();
+		clock_hi = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_UPPER_Raven2);
+		clock_lo = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_LOWER_Raven2);
+		hi_check = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_UPPER_Raven2);
+		/* The PWR TSC clock frequency is 100MHz, which sets 32-bit carry over
+		 * roughly every 42 seconds.
+		 */
+		if (hi_check != clock_hi) {
+			clock_lo = RREG32_SOC15_NO_KIQ(PWR, 0, mmGOLDEN_TSC_COUNT_LOWER_Raven2);
+			clock_hi = hi_check;
+		}
+		preempt_enable();
+		clock = clock_lo | (clock_hi << 32ULL);
+		break;
 	default:
 		amdgpu_gfx_off_ctrl(adev, false);
 		mutex_lock(&adev->gfx.gpu_clock_mutex);
@@ -6750,7 +6793,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v9_0_ring_get_rptr_gfx,
 	.get_wptr = gfx_v9_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v9_0_ring_set_wptr_gfx,
@@ -6804,7 +6846,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_sw_ring_funcs_gfx = {
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = amdgpu_sw_ring_get_rptr_gfx,
 	.get_wptr = amdgpu_sw_ring_get_wptr_gfx,
 	.set_wptr = amdgpu_sw_ring_set_wptr_gfx,
@@ -6858,7 +6899,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
@@ -6897,7 +6937,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
 	.align_mask = 0xff,
 	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
 	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
new file mode 100644
index 000000000000..5f8500577c02
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -0,0 +1,430 @@
+/*
+ * Copyright 2022 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.
+ *
+ */
+#include <linux/firmware.h>
+
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "soc15.h"
+#include "soc15_common.h"
+#include "vega10_enum.h"
+
+#include "gc/gc_9_4_3_offset.h"
+#include "gc/gc_9_4_3_sh_mask.h"
+
+#include "gfx_v9_4_3.h"
+
+#define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L
+
+static uint64_t gfx_v9_4_3_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	amdgpu_gfx_off_ctrl(adev, false);
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32_SOC15(GC, 0, regRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32_SOC15(GC, 0, regRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32_SOC15(GC, 0, regRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	amdgpu_gfx_off_ctrl(adev, true);
+
+	return clock;
+}
+
+static void gfx_v9_4_3_select_se_sh(struct amdgpu_device *adev,
+				    u32 se_num,
+				    u32 sh_num,
+				    u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX,
+				     INSTANCE_INDEX, instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+				     SE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+				     SH_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32_SOC15_RLC_SHADOW_EX(reg, GC, 0, regGRBM_GFX_INDEX, data);
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15_RLC(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15_RLC(GC, 0, regSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, 0, regSQ_IND_DATA);
+}
+
+static void gfx_v9_4_3_read_wave_data(struct amdgpu_device *adev,
+				      uint32_t simd, uint32_t wave,
+				      uint32_t *dst, int *no_fields)
+{
+	/* type 1 wave data */
+	dst[(*no_fields)++] = 1;
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_MODE);
+}
+
+static void gfx_v9_4_3_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+				       uint32_t wave, uint32_t start,
+				       uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(adev, simd, wave, 0,
+		       start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_4_3_read_wave_vgprs(struct amdgpu_device *adev, uint32_t simd,
+				       uint32_t wave, uint32_t thread,
+				       uint32_t start, uint32_t size,
+				       uint32_t *dst)
+{
+	wave_read_regs(adev, simd, wave, thread,
+		       start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_4_3_select_me_pipe_q(struct amdgpu_device *adev,
+					u32 me, u32 pipe, u32 q, u32 vm)
+{
+	soc15_grbm_select(adev, me, pipe, q, vm);
+}
+
+static bool gfx_v9_4_3_is_rlc_enabled(struct amdgpu_device *adev)
+{
+	uint32_t rlc_setting;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_setting = RREG32_SOC15(GC, 0, regRLC_CNTL);
+	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return false;
+
+	return true;
+}
+
+static void gfx_v9_4_3_set_safe_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+	unsigned i;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, data);
+
+	/* wait for RLC_SAFE_MODE */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, regRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v9_4_3_unset_safe_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	data = RLC_SAFE_MODE__CMD_MASK;
+	WREG32_SOC15(GC, 0, regRLC_SAFE_MODE, data);
+}
+
+static int gfx_v9_4_3_rlc_init(struct amdgpu_device *adev)
+{
+	/* init spm vmid with 0xf */
+	if (adev->gfx.rlc.funcs->update_spm_vmid)
+		adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32_SOC15(GC, 0, regRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+			if (k == adev->usec_timeout) {
+				gfx_v9_4_3_select_se_sh(adev, 0xffffffff,
+						      0xffffffff, 0xffffffff);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
+		}
+	}
+	gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32_SOC15(GC, 0, regRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v9_4_3_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp;
+
+	/* These interrupts should be enabled to drive DS clock */
+
+	tmp = RREG32_SOC15(GC, 0, regCP_INT_CNTL_RING0);
+
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0);
+	if (adev->gfx.num_gfx_rings)
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0);
+
+	WREG32_SOC15(GC, 0, regCP_INT_CNTL_RING0, tmp);
+}
+
+static void gfx_v9_4_3_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15_PREREG(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 0);
+	gfx_v9_4_3_enable_gui_idle_interrupt(adev, false);
+	gfx_v9_4_3_wait_for_rlc_serdes(adev);
+}
+
+static void gfx_v9_4_3_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15_PREREG(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v9_4_3_rlc_start(struct amdgpu_device *adev)
+{
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+	u32 rlc_ucode_ver;
+#endif
+
+	WREG32_FIELD15_PREREG(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
+
+	/* carrizo do enable cp interrupt after cp inited */
+	if (!(adev->flags & AMD_IS_APU)) {
+		gfx_v9_4_3_enable_gui_idle_interrupt(adev, true);
+		udelay(50);
+	}
+
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+	/* RLC_GPM_GENERAL_6 : RLC Ucode version */
+	rlc_ucode_ver = RREG32_SOC15(GC, 0, regRLC_GPM_GENERAL_6);
+	if (rlc_ucode_ver == 0x108) {
+		dev_info(adev->dev,
+			 "Using rlc debug ucode. regRLC_GPM_GENERAL_6 ==0x08%x / fw_ver == %i \n",
+			 rlc_ucode_ver, adev->gfx.rlc_fw_version);
+		/* RLC_GPM_TIMER_INT_3 : Timer interval in RefCLK cycles,
+		 * default is 0x9C4 to create a 100us interval */
+		WREG32_SOC15(GC, 0, regRLC_GPM_TIMER_INT_3, 0x9C4);
+		/* RLC_GPM_GENERAL_12 : Minimum gap between wptr and rptr
+		 * to disable the page fault retry interrupts, default is
+		 * 0x100 (256) */
+		WREG32_SOC15(GC, 0, regRLC_GPM_GENERAL_12, 0x100);
+	}
+#endif
+}
+
+static int gfx_v9_4_3_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR,
+			RLCG_UCODE_LOADING_START_ADDRESS);
+	for (i = 0; i < fw_size; i++) {
+		if (amdgpu_emu_mode == 1 && i % 100 == 0) {
+			dev_info(adev->dev, "Write RLC ucode data %u DWs\n", i);
+			msleep(1);
+		}
+		WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	}
+	WREG32_SOC15(GC, 0, regRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v9_4_3_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	adev->gfx.rlc.funcs->stop(adev);
+
+	/* disable CG */
+	WREG32_SOC15(GC, 0, regRLC_CGCG_CGLS_CTRL, 0);
+
+	/* TODO: revisit pg function */
+	/* gfx_v9_4_3_init_pg(adev);*/
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		/* legacy rlc firmware loading */
+		r = gfx_v9_4_3_rlc_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	adev->gfx.rlc.funcs->start(adev);
+
+	return 0;
+}
+
+static void gfx_v9_4_3_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
+{
+	u32 reg, data;
+
+	reg = SOC15_REG_OFFSET(GC, 0, regRLC_SPM_MC_CNTL);
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		data = RREG32_NO_KIQ(reg);
+	else
+		data = RREG32(reg);
+
+	data &= ~RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK;
+	data |= (vmid & RLC_SPM_MC_CNTL__RLC_SPM_VMID_MASK) << RLC_SPM_MC_CNTL__RLC_SPM_VMID__SHIFT;
+
+	if (amdgpu_sriov_is_pp_one_vf(adev))
+		WREG32_SOC15_NO_KIQ(GC, 0, regRLC_SPM_MC_CNTL, data);
+	else
+		WREG32_SOC15(GC, 0, regRLC_SPM_MC_CNTL, data);
+}
+
+static const struct soc15_reg_rlcg rlcg_access_gc_9_4_3[] = {
+	{SOC15_REG_ENTRY(GC, 0, regGRBM_GFX_INDEX)},
+	{SOC15_REG_ENTRY(GC, 0, regSQ_IND_INDEX)},
+};
+
+static bool gfx_v9_4_3_check_rlcg_range(struct amdgpu_device *adev,
+					uint32_t offset,
+					struct soc15_reg_rlcg *entries, int arr_size)
+{
+	int i;
+	uint32_t reg;
+
+	if (!entries)
+		return false;
+
+	for (i = 0; i < arr_size; i++) {
+		const struct soc15_reg_rlcg *entry;
+
+		entry = &entries[i];
+		reg = adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+		if (offset == reg)
+			return true;
+	}
+
+	return false;
+}
+
+static bool gfx_v9_4_3_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
+{
+	return gfx_v9_4_3_check_rlcg_range(adev, offset,
+					(void *)rlcg_access_gc_9_4_3,
+					ARRAY_SIZE(rlcg_access_gc_9_4_3));
+}
+
+const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v9_4_3_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v9_4_3_select_se_sh,
+	.read_wave_data = &gfx_v9_4_3_read_wave_data,
+	.read_wave_sgprs = &gfx_v9_4_3_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v9_4_3_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v9_4_3_select_me_pipe_q,
+};
+
+const struct amdgpu_rlc_funcs gfx_v9_4_3_rlc_funcs = {
+	.is_rlc_enabled = gfx_v9_4_3_is_rlc_enabled,
+	.set_safe_mode = gfx_v9_4_3_set_safe_mode,
+	.unset_safe_mode = gfx_v9_4_3_unset_safe_mode,
+	.init = gfx_v9_4_3_rlc_init,
+	.resume = gfx_v9_4_3_rlc_resume,
+	.stop = gfx_v9_4_3_rlc_stop,
+	.reset = gfx_v9_4_3_rlc_reset,
+	.start = gfx_v9_4_3_rlc_start,
+	.update_spm_vmid = gfx_v9_4_3_update_spm_vmid,
+	.is_rlcg_access_range = gfx_v9_4_3_is_rlcg_access_range,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h
new file mode 100644
index 000000000000..84e69701b81a
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2022 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.
+ *
+ */
+
+#ifndef __GFX_V9_4_3_H__
+#define __GFX_V9_4_3_H__
+
+extern const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs;
+extern const struct amdgpu_rlc_funcs gfx_v9_4_3_rlc_funcs;
+
+#endif /* __GFX_V9_4_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c
index be0d0f47415e..13712640fa46 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0.c
@@ -417,34 +417,12 @@ static void gfxhub_v3_0_set_fault_enable_default(struct amdgpu_device *adev,
 	tmp = REG_SET_FIELD(tmp, CP_DEBUG, CPG_UTCL1_ERROR_HALT_DISABLE, 1);
 	WREG32_SOC15(GC, 0, regCP_DEBUG, tmp);
 
-	/**
-	 * Set GRBM_GFX_INDEX in broad cast mode
-	 * before programming GL1C_UTCL0_CNTL1 and SQG_CONFIG
-	 */
-	WREG32_SOC15(GC, 0, regGRBM_GFX_INDEX, regGRBM_GFX_INDEX_DEFAULT);
-
-	/**
-	 * Retry respond mode: RETRY
-	 * Error (no retry) respond mode: SUCCESS
-	 */
-	tmp = RREG32_SOC15(GC, 0, regGL1C_UTCL0_CNTL1);
-	tmp = REG_SET_FIELD(tmp, GL1C_UTCL0_CNTL1, RESP_MODE, 0);
-	tmp = REG_SET_FIELD(tmp, GL1C_UTCL0_CNTL1, RESP_FAULT_MODE, 0x2);
-	WREG32_SOC15(GC, 0, regGL1C_UTCL0_CNTL1, tmp);
-
 	/* These registers are not accessible to VF-SRIOV.
 	 * The PF will program them instead.
 	 */
 	if (amdgpu_sriov_vf(adev))
 		return;
 
-	/* Disable SQ XNACK interrupt for all VMIDs */
-	tmp = RREG32_SOC15(GC, 0, regSQG_CONFIG);
-	tmp = REG_SET_FIELD(tmp, SQG_CONFIG, XNACK_INTR_MASK,
-			    SQG_CONFIG__XNACK_INTR_MASK_MASK >>
-			    SQG_CONFIG__XNACK_INTR_MASK__SHIFT);
-	WREG32_SOC15(GC, 0, regSQG_CONFIG, tmp);
-
 	tmp = RREG32_SOC15(GC, 0, regGCVM_L2_PROTECTION_FAULT_CNTL);
 	tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
 			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index d99821692ba3..7d6f4a68f416 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -479,8 +479,8 @@ static int gmc_v10_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 static uint64_t gmc_v10_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
 					     unsigned vmid, uint64_t pd_addr)
 {
-	bool use_semaphore = gmc_v10_0_use_invalidate_semaphore(ring->adev, ring->funcs->vmhub);
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	bool use_semaphore = gmc_v10_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
 	unsigned eng = ring->vm_inv_eng;
 
@@ -534,7 +534,7 @@ static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid
 	if (ring->is_mes_queue)
 		return;
 
-	if (ring->funcs->vmhub == AMDGPU_GFXHUB_0)
+	if (ring->vm_hub == AMDGPU_GFXHUB_0)
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
 	else
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
index 9f4f28192c60..d809f2ed5600 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
@@ -378,8 +378,8 @@ static int gmc_v11_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 static uint64_t gmc_v11_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
 					     unsigned vmid, uint64_t pd_addr)
 {
-	bool use_semaphore = gmc_v11_0_use_invalidate_semaphore(ring->adev, ring->funcs->vmhub);
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	bool use_semaphore = gmc_v11_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
 	unsigned eng = ring->vm_inv_eng;
 
@@ -433,7 +433,7 @@ static void gmc_v11_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid
 	if (ring->is_mes_queue)
 		return;
 
-	if (ring->funcs->vmhub == AMDGPU_GFXHUB_0)
+	if (ring->vm_hub == AMDGPU_GFXHUB_0)
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid;
 	else
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT_MM) + vmid;
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 67ad92097a65..64ab1a306dfe 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -555,32 +555,49 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 	const char *mmhub_cid;
 	const char *hub_name;
 	u64 addr;
+	uint32_t cam_index = 0;
+	int ret;
 
 	addr = (u64)entry->src_data[0] << 12;
 	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
 
 	if (retry_fault) {
-		/* Returning 1 here also prevents sending the IV to the KFD */
+		if (adev->irq.retry_cam_enabled) {
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
+
+			cam_index = entry->src_data[2] & 0x3ff;
 
-		/* Process it onyl if it's the first fault for this address */
-		if (entry->ih != &adev->irq.ih_soft &&
-		    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+			ret = amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault);
+			WDOORBELL32(adev->irq.retry_cam_doorbell_index, cam_index);
+			if (ret)
+				return 1;
+		} else {
+			/* Process it onyl if it's the first fault for this address */
+			if (entry->ih != &adev->irq.ih_soft &&
+			    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
 					     entry->timestamp))
-			return 1;
+				return 1;
 
-		/* Delegate it to a different ring if the hardware hasn't
-		 * already done it.
-		 */
-		if (entry->ih == &adev->irq.ih) {
-			amdgpu_irq_delegate(adev, entry, 8);
-			return 1;
-		}
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
 
-		/* Try to handle the recoverable page faults by filling page
-		 * tables
-		 */
-		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
-			return 1;
+			/* Try to handle the recoverable page faults by filling page
+			 * tables
+			 */
+			if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
+				return 1;
+		}
 	}
 
 	if (!printk_ratelimit())
@@ -990,9 +1007,9 @@ static int gmc_v9_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 static uint64_t gmc_v9_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
 					    unsigned vmid, uint64_t pd_addr)
 {
-	bool use_semaphore = gmc_v9_0_use_invalidate_semaphore(ring->adev, ring->funcs->vmhub);
+	bool use_semaphore = gmc_v9_0_use_invalidate_semaphore(ring->adev, ring->vm_hub);
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_vmhub *hub = &adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &adev->vmhub[ring->vm_hub];
 	uint32_t req = gmc_v9_0_get_invalidate_req(vmid, 0);
 	unsigned eng = ring->vm_inv_eng;
 
@@ -1043,10 +1060,10 @@ static void gmc_v9_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
 	uint32_t reg;
 
 	/* Do nothing because there's no lut register for mmhub1. */
-	if (ring->funcs->vmhub == AMDGPU_MMHUB_1)
+	if (ring->vm_hub == AMDGPU_MMHUB_1)
 		return;
 
-	if (ring->funcs->vmhub == AMDGPU_GFXHUB_0)
+	if (ring->vm_hub == AMDGPU_GFXHUB_0)
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
 	else
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
index 9360204da7fb..a3076eb8af6a 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
@@ -376,7 +376,7 @@ static void jpeg_v1_0_decode_ring_emit_reg_wait(struct amdgpu_ring *ring,
 static void jpeg_v1_0_decode_ring_emit_vm_flush(struct amdgpu_ring *ring,
 		unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
@@ -485,6 +485,7 @@ int jpeg_v1_0_sw_init(void *handle)
 		return r;
 
 	ring = &adev->jpeg.inst->ring_dec;
+	ring->vm_hub = AMDGPU_MMHUB_0;
 	sprintf(ring->name, "jpeg_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq,
 			     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -548,7 +549,6 @@ static const struct amdgpu_ring_funcs jpeg_v1_0_decode_ring_vm_funcs = {
 	.nop = PACKET0(0x81ff, 0),
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.extra_dw = 64,
 	.get_rptr = jpeg_v1_0_decode_ring_get_rptr,
 	.get_wptr = jpeg_v1_0_decode_ring_get_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
index f3c1af5130ab..0eddf7c824a7 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
@@ -86,6 +86,7 @@ static int jpeg_v2_0_sw_init(void *handle)
 	ring = &adev->jpeg.inst->ring_dec;
 	ring->use_doorbell = true;
 	ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+	ring->vm_hub = AMDGPU_MMHUB_0;
 	sprintf(ring->name, "jpeg_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq,
 			     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -613,7 +614,7 @@ void jpeg_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 void jpeg_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
 				unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
@@ -762,7 +763,6 @@ static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = {
 static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_JPEG,
 	.align_mask = 0xf,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = jpeg_v2_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_0_dec_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
index 6b1887808782..b040f51d9aa9 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
@@ -127,6 +127,10 @@ static int jpeg_v2_5_sw_init(void *handle)
 
 		ring = &adev->jpeg.inst[i].ring_dec;
 		ring->use_doorbell = true;
+		if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0))
+			ring->vm_hub = AMDGPU_MMHUB_1;
+		else
+			ring->vm_hub = AMDGPU_MMHUB_0;
 		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + 8 * i;
 		sprintf(ring->name, "jpeg_dec_%d", i);
 		r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst[i].irq,
@@ -645,7 +649,6 @@ static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = {
 static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_JPEG,
 	.align_mask = 0xf,
-	.vmhub = AMDGPU_MMHUB_1,
 	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
@@ -675,7 +678,6 @@ static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
 static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_JPEG,
 	.align_mask = 0xf,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
index a1b751d9ac06..c55e09432e26 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
@@ -100,6 +100,7 @@ static int jpeg_v3_0_sw_init(void *handle)
 	ring = &adev->jpeg.inst->ring_dec;
 	ring->use_doorbell = true;
 	ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+	ring->vm_hub = AMDGPU_MMHUB_0;
 	sprintf(ring->name, "jpeg_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
 			     AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -559,7 +560,6 @@ static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = {
 static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_JPEG,
 	.align_mask = 0xf,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = jpeg_v3_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v3_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v3_0_dec_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
index 5f2a034b9ec0..a6ad678fd507 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -108,6 +108,7 @@ static int jpeg_v4_0_sw_init(void *handle)
 	ring = &adev->jpeg.inst->ring_dec;
 	ring->use_doorbell = true;
 	ring->doorbell_index = amdgpu_sriov_vf(adev) ? (((adev->doorbell_index.vcn.vcn_ring0_1) << 1) + 4) : ((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1);
+	ring->vm_hub = AMDGPU_MMHUB_0;
 
 	sprintf(ring->name, "jpeg_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
@@ -715,7 +716,6 @@ static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = {
 static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_JPEG,
 	.align_mask = 0xf,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = jpeg_v4_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v4_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v4_0_dec_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
index 5826eac270d7..45280f047180 100644
--- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
@@ -33,14 +33,19 @@
 #include "mes_v11_api_def.h"
 
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_mes1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_1_mes1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_2_mes1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_3_mes1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_4_mes1.bin");
 
 static int mes_v11_0_hw_fini(void *handle);
@@ -1089,13 +1094,14 @@ static int mes_v11_0_sw_fini(void *handle)
 	return 0;
 }
 
-static void mes_v11_0_kiq_dequeue_sched(struct amdgpu_device *adev)
+static void mes_v11_0_kiq_dequeue(struct amdgpu_ring *ring)
 {
 	uint32_t data;
 	int i;
+	struct amdgpu_device *adev = ring->adev;
 
 	mutex_lock(&adev->srbm_mutex);
-	soc21_grbm_select(adev, 3, AMDGPU_MES_SCHED_PIPE, 0, 0);
+	soc21_grbm_select(adev, 3, ring->pipe, 0, 0);
 
 	/* disable the queue if it's active */
 	if (RREG32_SOC15(GC, 0, regCP_HQD_ACTIVE) & 1) {
@@ -1121,8 +1127,6 @@ static void mes_v11_0_kiq_dequeue_sched(struct amdgpu_device *adev)
 
 	soc21_grbm_select(adev, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
-
-	adev->mes.ring.sched.ready = false;
 }
 
 static void mes_v11_0_kiq_setting(struct amdgpu_ring *ring)
@@ -1139,6 +1143,16 @@ static void mes_v11_0_kiq_setting(struct amdgpu_ring *ring)
 	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp);
 }
 
+static void mes_v11_0_kiq_clear(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* tell RLC which is KIQ dequeue */
+	tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS);
+	tmp &= ~RLC_CP_SCHEDULERS__scheduler0_MASK;
+	WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp);
+}
+
 static int mes_v11_0_kiq_hw_init(struct amdgpu_device *adev)
 {
 	int r = 0;
@@ -1176,11 +1190,17 @@ failure:
 
 static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev)
 {
-	if (adev->mes.ring.sched.ready)
-		mes_v11_0_kiq_dequeue_sched(adev);
+	if (adev->mes.ring.sched.ready) {
+		mes_v11_0_kiq_dequeue(&adev->mes.ring);
+		adev->mes.ring.sched.ready = false;
+	}
 
-	if (!amdgpu_sriov_vf(adev))
-		mes_v11_0_enable(adev, false);
+	if (amdgpu_sriov_vf(adev)) {
+		mes_v11_0_kiq_dequeue(&adev->gfx.kiq.ring);
+		mes_v11_0_kiq_clear(adev);
+	}
+
+	mes_v11_0_enable(adev, false);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
index 19455a725939..685abf57ffdd 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -238,7 +238,7 @@ static void nbio_v7_4_ih_doorbell_range(struct amdgpu_device *adev,
 
 	if (use_doorbell) {
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
-		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 4);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 8);
 	} else
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index b5affba22156..b3cc04dd8653 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -1823,6 +1823,15 @@ static int sdma_v4_0_sw_init(void *handle)
 		/* doorbell size is 2 dwords, get DWORD offset */
 		ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
 
+		/*
+		 * On Arcturus, SDMA instance 5~7 has a different vmhub
+		 * type(AMDGPU_MMHUB_1).
+		 */
+		if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5)
+			ring->vm_hub = AMDGPU_MMHUB_1;
+		else
+			ring->vm_hub = AMDGPU_MMHUB_0;
+
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
 				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
@@ -1841,6 +1850,11 @@ static int sdma_v4_0_sw_init(void *handle)
 			ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
 			ring->doorbell_index += 0x400;
 
+			if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5)
+				ring->vm_hub = AMDGPU_MMHUB_1;
+			else
+				ring->vm_hub = AMDGPU_MMHUB_0;
+
 			sprintf(ring->name, "page%d", i);
 			r = amdgpu_ring_init(adev, ring, 1024,
 					     &adev->sdma.trap_irq,
@@ -1870,7 +1884,7 @@ static int sdma_v4_0_sw_fini(void *handle)
 			amdgpu_ring_fini(&adev->sdma.instance[i].page);
 	}
 
-	if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 0) ||
+	if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) ||
             adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 0))
 		amdgpu_sdma_destroy_inst_ctx(adev, true);
 	else
@@ -2294,44 +2308,6 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
-	.get_rptr = sdma_v4_0_ring_get_rptr,
-	.get_wptr = sdma_v4_0_ring_get_wptr,
-	.set_wptr = sdma_v4_0_ring_set_wptr,
-	.emit_frame_size =
-		6 + /* sdma_v4_0_ring_emit_hdp_flush */
-		3 + /* hdp invalidate */
-		6 + /* sdma_v4_0_ring_emit_pipeline_sync */
-		/* sdma_v4_0_ring_emit_vm_flush */
-		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
-		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
-		10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
-	.emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */
-	.emit_ib = sdma_v4_0_ring_emit_ib,
-	.emit_fence = sdma_v4_0_ring_emit_fence,
-	.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
-	.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
-	.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
-	.test_ring = sdma_v4_0_ring_test_ring,
-	.test_ib = sdma_v4_0_ring_test_ib,
-	.insert_nop = sdma_v4_0_ring_insert_nop,
-	.pad_ib = sdma_v4_0_ring_pad_ib,
-	.emit_wreg = sdma_v4_0_ring_emit_wreg,
-	.emit_reg_wait = sdma_v4_0_ring_emit_reg_wait,
-	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
-};
-
-/*
- * On Arcturus, SDMA instance 5~7 has a different vmhub type(AMDGPU_MMHUB_1).
- * So create a individual constant ring_funcs for those instances.
- */
-static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs_2nd_mmhub = {
-	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
-	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
-	.support_64bit_ptrs = true,
-	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_1,
 	.get_rptr = sdma_v4_0_ring_get_rptr,
 	.get_wptr = sdma_v4_0_ring_get_wptr,
 	.set_wptr = sdma_v4_0_ring_set_wptr,
@@ -2364,40 +2340,6 @@ static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = {
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
-	.get_rptr = sdma_v4_0_ring_get_rptr,
-	.get_wptr = sdma_v4_0_page_ring_get_wptr,
-	.set_wptr = sdma_v4_0_page_ring_set_wptr,
-	.emit_frame_size =
-		6 + /* sdma_v4_0_ring_emit_hdp_flush */
-		3 + /* hdp invalidate */
-		6 + /* sdma_v4_0_ring_emit_pipeline_sync */
-		/* sdma_v4_0_ring_emit_vm_flush */
-		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
-		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
-		10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
-	.emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */
-	.emit_ib = sdma_v4_0_ring_emit_ib,
-	.emit_fence = sdma_v4_0_ring_emit_fence,
-	.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
-	.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
-	.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
-	.test_ring = sdma_v4_0_ring_test_ring,
-	.test_ib = sdma_v4_0_ring_test_ib,
-	.insert_nop = sdma_v4_0_ring_insert_nop,
-	.pad_ib = sdma_v4_0_ring_pad_ib,
-	.emit_wreg = sdma_v4_0_ring_emit_wreg,
-	.emit_reg_wait = sdma_v4_0_ring_emit_reg_wait,
-	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
-};
-
-static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs_2nd_mmhub = {
-	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
-	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
-	.support_64bit_ptrs = true,
-	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_1,
 	.get_rptr = sdma_v4_0_ring_get_rptr,
 	.get_wptr = sdma_v4_0_page_ring_get_wptr,
 	.set_wptr = sdma_v4_0_page_ring_set_wptr,
@@ -2429,19 +2371,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev)
 	int i;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5)
-			adev->sdma.instance[i].ring.funcs =
-					&sdma_v4_0_ring_funcs_2nd_mmhub;
-		else
-			adev->sdma.instance[i].ring.funcs =
-					&sdma_v4_0_ring_funcs;
+		adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs;
 		adev->sdma.instance[i].ring.me = i;
 		if (adev->sdma.has_page_queue) {
-			if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5)
-				adev->sdma.instance[i].page.funcs =
-					&sdma_v4_0_page_ring_funcs_2nd_mmhub;
-			else
-				adev->sdma.instance[i].page.funcs =
+			adev->sdma.instance[i].page.funcs =
 					&sdma_v4_0_page_ring_funcs;
 			adev->sdma.instance[i].page.me = i;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
index 1b04700a4d55..64dcaa2670dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -1309,6 +1309,7 @@ static int sdma_v4_4_2_sw_init(void *handle)
 
 		/* doorbell size is 2 dwords, get DWORD offset */
 		ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
+		ring->vm_hub = AMDGPU_MMHUB_0;
 
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
@@ -1327,6 +1328,7 @@ static int sdma_v4_4_2_sw_init(void *handle)
 			 */
 			ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
 			ring->doorbell_index += 0x400;
+			ring->vm_hub = AMDGPU_MMHUB_0;
 
 			sprintf(ring->name, "page%d", i);
 			r = amdgpu_ring_init(adev, ring, 1024,
@@ -1741,7 +1743,6 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {
 	.align_mask = 0xf,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
 	.get_wptr = sdma_v4_4_2_ring_get_wptr,
 	.set_wptr = sdma_v4_4_2_ring_set_wptr,
@@ -1773,7 +1774,6 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = {
 	.align_mask = 0xf,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
 	.get_wptr = sdma_v4_4_2_page_ring_get_wptr,
 	.set_wptr = sdma_v4_4_2_page_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
index 1941b3b7c5d9..92e1299be021 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
@@ -1389,6 +1389,7 @@ static int sdma_v5_0_sw_init(void *handle)
 			(adev->doorbell_index.sdma_engine[0] << 1) //get DWORD offset
 			: (adev->doorbell_index.sdma_engine[1] << 1); // get DWORD offset
 
+		ring->vm_hub = AMDGPU_GFXHUB_0;
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
 				     (i == 0) ? AMDGPU_SDMA_IRQ_INSTANCE0 :
@@ -1765,7 +1766,6 @@ static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = {
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = sdma_v5_0_ring_get_rptr,
 	.get_wptr = sdma_v5_0_ring_get_wptr,
 	.set_wptr = sdma_v5_0_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
index 8e445eb9dd49..ca7e8757d78e 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
@@ -1253,6 +1253,7 @@ static int sdma_v5_2_sw_init(void *handle)
 		ring->doorbell_index =
 			(adev->doorbell_index.sdma_engine[i] << 1); //get DWORD offset
 
+		ring->vm_hub = AMDGPU_GFXHUB_0;
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024, &adev->sdma.trap_irq,
 				     AMDGPU_SDMA_IRQ_INSTANCE0 + i,
@@ -1653,7 +1654,6 @@ static const struct amdgpu_ring_funcs sdma_v5_2_ring_funcs = {
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = sdma_v5_2_ring_get_rptr,
 	.get_wptr = sdma_v5_2_ring_get_wptr,
 	.set_wptr = sdma_v5_2_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
index 40e6b22daa22..eb722830531f 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
@@ -403,15 +403,26 @@ static void sdma_v6_0_rlc_stop(struct amdgpu_device *adev)
 }
 
 /**
- * sdma_v6_0_ctx_switch_enable - stop the async dma engines context switch
+ * sdma_v6_0_ctxempty_int_enable - enable or disable context empty interrupts
  *
  * @adev: amdgpu_device pointer
- * @enable: enable/disable the DMA MEs context switch.
+ * @enable: enable/disable context switching due to queue empty conditions
  *
- * Halt or unhalt the async dma engines context switch.
+ * Enable or disable the async dma engines queue empty context switch.
  */
-static void sdma_v6_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+static void sdma_v6_0_ctxempty_int_enable(struct amdgpu_device *adev, bool enable)
 {
+	u32 f32_cntl;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			f32_cntl = RREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_CNTL));
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					CTXEMPTY_INT_ENABLE, enable ? 1 : 0);
+			WREG32(sdma_v6_0_get_reg_offset(adev, i, regSDMA0_CNTL), f32_cntl);
+		}
+	}
 }
 
 /**
@@ -579,10 +590,8 @@ static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev)
 
 		ring->sched.ready = true;
 
-		if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
-			sdma_v6_0_ctx_switch_enable(adev, true);
+		if (amdgpu_sriov_vf(adev))
 			sdma_v6_0_enable(adev, true);
-		}
 
 		r = amdgpu_ring_test_helper(ring);
 		if (r) {
@@ -778,7 +787,6 @@ static int sdma_v6_0_start(struct amdgpu_device *adev)
 	int r = 0;
 
 	if (amdgpu_sriov_vf(adev)) {
-		sdma_v6_0_ctx_switch_enable(adev, false);
 		sdma_v6_0_enable(adev, false);
 
 		/* set RB registers */
@@ -799,7 +807,7 @@ static int sdma_v6_0_start(struct amdgpu_device *adev)
 	/* unhalt the MEs */
 	sdma_v6_0_enable(adev, true);
 	/* enable sdma ring preemption */
-	sdma_v6_0_ctx_switch_enable(adev, true);
+	sdma_v6_0_ctxempty_int_enable(adev, true);
 
 	/* start the gfx rings and rlc compute queues */
 	r = sdma_v6_0_gfx_resume(adev);
@@ -1173,7 +1181,28 @@ static void sdma_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 static void sdma_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					 unsigned vmid, uint64_t pd_addr)
 {
-	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
+	uint32_t req = hub->vmhub_funcs->get_invalidate_req(vmid, 0);
+
+	/* Update the PD address for this VMID. */
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 +
+			      (hub->ctx_addr_distance * vmid),
+			      lower_32_bits(pd_addr));
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 +
+			      (hub->ctx_addr_distance * vmid),
+			      upper_32_bits(pd_addr));
+
+	/* Trigger invalidation. */
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_SUB_OP(SDMA_SUBOP_VM_INVALIDATION) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_GFX_ENG_ID(ring->vm_inv_eng) |
+			  SDMA_PKT_VM_INVALIDATION_HEADER_MM_ENG_ID(0x1f));
+	amdgpu_ring_write(ring, req);
+	amdgpu_ring_write(ring, 0xFFFFFFFF);
+	amdgpu_ring_write(ring,
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_INVALIDATEACK(1 << vmid) |
+			  SDMA_PKT_VM_INVALIDATION_ADDRESSRANGEHI_ADDRESSRANGEHI(0x1F));
 }
 
 static void sdma_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
@@ -1272,6 +1301,7 @@ static int sdma_v6_0_sw_init(void *handle)
 		ring->doorbell_index =
 			(adev->doorbell_index.sdma_engine[i] << 1); // get DWORD offset
 
+		ring->vm_hub = AMDGPU_GFXHUB_0;
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024,
 				     &adev->sdma.trap_irq,
@@ -1319,7 +1349,7 @@ static int sdma_v6_0_hw_fini(void *handle)
 		return 0;
 	}
 
-	sdma_v6_0_ctx_switch_enable(adev, false);
+	sdma_v6_0_ctxempty_int_enable(adev, false);
 	sdma_v6_0_enable(adev, false);
 
 	return 0;
@@ -1528,7 +1558,6 @@ static const struct amdgpu_ring_funcs sdma_v6_0_ring_funcs = {
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_GFXHUB_0,
 	.get_rptr = sdma_v6_0_ring_get_rptr,
 	.get_wptr = sdma_v6_0_ring_get_wptr,
 	.set_wptr = sdma_v6_0_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 7d04c39332ad..bc5dd80f10c1 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -301,11 +301,10 @@ static u32 soc15_get_xclk(struct amdgpu_device *adev)
 	u32 reference_clock = adev->clock.spll.reference_freq;
 
 	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 0) ||
-	    adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 1))
-		return 10000;
-	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(10, 0, 0) ||
+	    adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 1) ||
+	    adev->ip_versions[MP1_HWIP][0] == IP_VERSION(10, 0, 0) ||
 	    adev->ip_versions[MP1_HWIP][0] == IP_VERSION(10, 0, 1))
-		return reference_clock / 4;
+		return 10000;
 
 	return reference_clock;
 }
@@ -1101,6 +1100,11 @@ static int soc15_common_early_init(void *handle)
 		adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG;
 		adev->external_rev_id = adev->rev_id + 0x3c;
 		break;
+	case IP_VERSION(9, 4, 3):
+		adev->asic_funcs = &vega20_asic_funcs;
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		break;
 	default:
 		/* FIXME: not supported yet */
 		return -EINVAL;
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
index 509d8a1945eb..30d0482ac466 100644
--- a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
@@ -84,6 +84,8 @@ enum ta_ras_block {
 	TA_RAS_BLOCK__MP1,
 	TA_RAS_BLOCK__FUSE,
 	TA_RAS_BLOCK__MCA,
+	TA_RAS_BLOCK__VCN,
+	TA_RAS_BLOCK__JPEG,
 	TA_NUM_BLOCK_MAX
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
index e08e25a3a1a9..530549314ce4 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
@@ -160,24 +160,28 @@ static void umc_v6_7_ecc_info_querry_uncorrectable_error_count(struct amdgpu_dev
 	}
 }
 
+static int umc_v6_7_ecc_info_querry_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+	umc_v6_7_ecc_info_query_correctable_error_count(adev,
+		umc_inst, ch_inst,
+		&(err_data->ce_count));
+
+	umc_v6_7_ecc_info_querry_uncorrectable_error_count(adev,
+		umc_inst, ch_inst,
+		&(err_data->ue_count));
+
+	return 0;
+}
+
 static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
 					   void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-
-	/*TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC registers. Will add the protection */
-	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
-		umc_v6_7_ecc_info_query_correctable_error_count(adev,
-						      umc_inst, ch_inst,
-						      &(err_data->ce_count));
-		umc_v6_7_ecc_info_querry_uncorrectable_error_count(adev,
-						      umc_inst, ch_inst,
-							  &(err_data->ue_count));
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_ecc_info_querry_ecc_error_count, ras_error_status);
 }
 
 void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
@@ -215,23 +219,23 @@ void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
 	}
 }
 
-static void umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev,
-					 struct ras_err_data *err_data,
-					 uint32_t ch_inst,
-					 uint32_t umc_inst)
+static int umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint64_t mc_umc_status, err_addr;
 	uint32_t eccinfo_table_idx;
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
 
 	eccinfo_table_idx = umc_inst * adev->umc.channel_inst_num + ch_inst;
 	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
 
 	if (mc_umc_status == 0)
-		return;
+		return 0;
 
 	if (!err_data->err_addr)
-		return;
+		return 0;
 
 	/* calculate error address if ue error is detected */
 	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
@@ -243,25 +247,15 @@ static void umc_v6_7_ecc_info_query_error_address(struct amdgpu_device *adev,
 		umc_v6_7_convert_error_address(adev, err_data, err_addr,
 					ch_inst, umc_inst);
 	}
+
+	return 0;
 }
 
 static void umc_v6_7_ecc_info_query_ras_error_address(struct amdgpu_device *adev,
 					     void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-
-	/*TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC resgisters. Will add the protection
-	 * when firmware interface is ready */
-	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
-		umc_v6_7_ecc_info_query_error_address(adev,
-					     err_data,
-					     ch_inst,
-					     umc_inst);
-	}
+	amdgpu_umc_loop_channels(adev,
+	    umc_v6_7_ecc_info_query_error_address, ras_error_status);
 }
 
 static void umc_v6_7_query_correctable_error_count(struct amdgpu_device *adev,
@@ -364,11 +358,14 @@ static void umc_v6_7_querry_uncorrectable_error_count(struct amdgpu_device *adev
 	}
 }
 
-static void umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev,
-						   uint32_t umc_reg_offset)
+static int umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint32_t ecc_err_cnt_addr;
 	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
 
 	ecc_err_cnt_sel_addr =
 		SOC15_REG_OFFSET(UMC, 0,
@@ -402,58 +399,54 @@ static void umc_v6_7_reset_error_count_per_channel(struct amdgpu_device *adev,
 	/* clear higher chip error count */
 	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
 			UMC_V6_7_CE_CNT_INIT);
+
+	return 0;
 }
 
 static void umc_v6_7_reset_error_count(struct amdgpu_device *adev)
 {
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_reset_error_count_per_channel, NULL);
+}
+
+static int umc_v6_7_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
+
+	umc_v6_7_query_correctable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ce_count),
+					ch_inst, umc_inst);
 
-	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v6_7_reg_offset(adev,
-							 umc_inst,
-							 ch_inst);
+	umc_v6_7_querry_uncorrectable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ue_count));
 
-		umc_v6_7_reset_error_count_per_channel(adev,
-						       umc_reg_offset);
-	}
+	return 0;
 }
 
 static void umc_v6_7_query_ras_error_count(struct amdgpu_device *adev,
 					   void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	/*TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC registers. Will add the protection */
-	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v6_7_reg_offset(adev,
-							 umc_inst,
-							 ch_inst);
-		umc_v6_7_query_correctable_error_count(adev,
-						       umc_reg_offset,
-						       &(err_data->ce_count),
-						       ch_inst, umc_inst);
-		umc_v6_7_querry_uncorrectable_error_count(adev,
-							  umc_reg_offset,
-							  &(err_data->ue_count));
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_query_ecc_error_count, ras_error_status);
 
 	umc_v6_7_reset_error_count(adev);
 }
 
-static void umc_v6_7_query_error_address(struct amdgpu_device *adev,
-					 struct ras_err_data *err_data,
-					 uint32_t umc_reg_offset, uint32_t ch_inst,
-					 uint32_t umc_inst)
+static int umc_v6_7_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint32_t mc_umc_status_addr;
 	uint64_t mc_umc_status = 0, mc_umc_addrt0, err_addr;
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v6_7_reg_offset(adev, umc_inst, ch_inst);
 
 	mc_umc_status_addr =
 		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
@@ -463,12 +456,12 @@ static void umc_v6_7_query_error_address(struct amdgpu_device *adev,
 	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
 
 	if (mc_umc_status == 0)
-		return;
+		return 0;
 
 	if (!err_data->err_addr) {
 		/* clear umc status */
 		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
-		return;
+		return 0;
 	}
 
 	/* calculate error address if ue error is detected */
@@ -484,29 +477,15 @@ static void umc_v6_7_query_error_address(struct amdgpu_device *adev,
 
 	/* clear umc status */
 	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+	return 0;
 }
 
 static void umc_v6_7_query_ras_error_address(struct amdgpu_device *adev,
 					     void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	/*TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC resgisters. Will add the protection
-	 * when firmware interface is ready */
-	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v6_7_reg_offset(adev,
-							 umc_inst,
-							 ch_inst);
-		umc_v6_7_query_error_address(adev,
-					     err_data,
-					     umc_reg_offset, ch_inst,
-					     umc_inst);
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v6_7_query_error_address, ras_error_status);
 }
 
 static uint32_t umc_v6_7_query_ras_poison_mode_per_channel(
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c
index fb55e8cb9967..d51ae0bc36f7 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c
@@ -76,10 +76,13 @@ static inline uint32_t get_umc_v8_10_reg_offset(struct amdgpu_device *adev,
 		UMC_8_NODE_DIST * node_inst;
 }
 
-static void umc_v8_10_clear_error_count_per_channel(struct amdgpu_device *adev,
-					uint32_t umc_reg_offset)
+static int umc_v8_10_clear_error_count_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
 
 	ecc_err_cnt_addr =
 		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_GeccErrCnt);
@@ -87,24 +90,14 @@ static void umc_v8_10_clear_error_count_per_channel(struct amdgpu_device *adev,
 	/* clear error count */
 	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4,
 			UMC_V8_10_CE_CNT_INIT);
+
+	return 0;
 }
 
 static void umc_v8_10_clear_error_count(struct amdgpu_device *adev)
 {
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v8_10_reg_offset(adev,
-						node_inst,
-						umc_inst,
-						ch_inst);
-
-		umc_v8_10_clear_error_count_per_channel(adev,
-						umc_reg_offset);
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_clear_error_count_per_channel, NULL);
 }
 
 static void umc_v8_10_query_correctable_error_count(struct amdgpu_device *adev,
@@ -147,29 +140,29 @@ static void umc_v8_10_query_uncorrectable_error_count(struct amdgpu_device *adev
 		*error_count += 1;
 }
 
+static int umc_v8_10_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
+
+	umc_v8_10_query_correctable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ce_count));
+	umc_v8_10_query_uncorrectable_error_count(adev,
+					umc_reg_offset,
+					&(err_data->ue_count));
+
+	return 0;
+}
+
 static void umc_v8_10_query_ras_error_count(struct amdgpu_device *adev,
 					   void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v8_10_reg_offset(adev,
-						node_inst,
-						umc_inst,
-						ch_inst);
-
-		umc_v8_10_query_correctable_error_count(adev,
-						umc_reg_offset,
-						&(err_data->ce_count));
-		umc_v8_10_query_uncorrectable_error_count(adev,
-						umc_reg_offset,
-						&(err_data->ue_count));
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_query_ecc_error_count, ras_error_status);
 
 	umc_v8_10_clear_error_count(adev);
 }
@@ -248,28 +241,28 @@ static void umc_v8_10_convert_error_address(struct amdgpu_device *adev,
 	}
 }
 
-static void umc_v8_10_query_error_address(struct amdgpu_device *adev,
-					 struct ras_err_data *err_data,
-					 uint32_t umc_reg_offset,
-					 uint32_t node_inst,
-					 uint32_t ch_inst,
-					 uint32_t umc_inst)
+static int umc_v8_10_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint64_t mc_umc_status_addr;
 	uint64_t mc_umc_status, err_addr;
 	uint64_t mc_umc_addrt0;
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
 
 	mc_umc_status_addr =
 		SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
 	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
 
 	if (mc_umc_status == 0)
-		return;
+		return 0;
 
 	if (!err_data->err_addr) {
 		/* clear umc status */
 		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
-		return;
+		return 0;
 	}
 
 	/* calculate error address if ue error is detected */
@@ -287,37 +280,25 @@ static void umc_v8_10_query_error_address(struct amdgpu_device *adev,
 
 	/* clear umc status */
 	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
+
+	return 0;
 }
 
 static void umc_v8_10_query_ras_error_address(struct amdgpu_device *adev,
 					     void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v8_10_reg_offset(adev,
-						node_inst,
-						umc_inst,
-						ch_inst);
-
-		umc_v8_10_query_error_address(adev,
-					err_data,
-					umc_reg_offset,
-					node_inst,
-					ch_inst,
-					umc_inst);
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_query_error_address, ras_error_status);
 }
 
-static void umc_v8_10_err_cnt_init_per_channel(struct amdgpu_device *adev,
-					      uint32_t umc_reg_offset)
+static int umc_v8_10_err_cnt_init_per_channel(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
 	uint32_t ecc_err_cnt_addr;
+	uint32_t umc_reg_offset =
+		get_umc_v8_10_reg_offset(adev, node_inst, umc_inst, ch_inst);
 
 	ecc_err_cnt_sel_addr =
 		SOC15_REG_OFFSET(UMC, 0, regUMCCH0_0_GeccErrCntSel);
@@ -332,23 +313,14 @@ static void umc_v8_10_err_cnt_init_per_channel(struct amdgpu_device *adev,
 	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
 	/* set error count to initial value */
 	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V8_10_CE_CNT_INIT);
+
+	return 0;
 }
 
 static void umc_v8_10_err_cnt_init(struct amdgpu_device *adev)
 {
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-	uint32_t umc_reg_offset  = 0;
-
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_reg_offset = get_umc_v8_10_reg_offset(adev,
-						node_inst,
-						umc_inst,
-						ch_inst);
-
-		umc_v8_10_err_cnt_init_per_channel(adev, umc_reg_offset);
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_err_cnt_init_per_channel, NULL);
 }
 
 static bool umc_v8_10_query_ras_poison_mode(struct amdgpu_device *adev)
@@ -406,37 +378,35 @@ static void umc_v8_10_ecc_info_query_uncorrectable_error_count(struct amdgpu_dev
 	}
 }
 
+static int umc_v8_10_ecc_info_query_ecc_error_count(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
+{
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+	umc_v8_10_ecc_info_query_correctable_error_count(adev,
+					node_inst, umc_inst, ch_inst,
+					&(err_data->ce_count));
+	umc_v8_10_ecc_info_query_uncorrectable_error_count(adev,
+					node_inst, umc_inst, ch_inst,
+					&(err_data->ue_count));
+	return 0;
+}
+
 static void umc_v8_10_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
 					void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-
-	/* TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC registers. Will add the protection
-	 */
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_v8_10_ecc_info_query_correctable_error_count(adev,
-							node_inst, umc_inst, ch_inst,
-							&(err_data->ce_count));
-		umc_v8_10_ecc_info_query_uncorrectable_error_count(adev,
-							node_inst, umc_inst, ch_inst,
-							&(err_data->ue_count));
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_ecc_info_query_ecc_error_count, ras_error_status);
 }
 
-static void umc_v8_10_ecc_info_query_error_address(struct amdgpu_device *adev,
-					struct ras_err_data *err_data,
-					uint32_t ch_inst,
-					uint32_t umc_inst,
-					uint32_t node_inst)
+static int umc_v8_10_ecc_info_query_error_address(struct amdgpu_device *adev,
+					uint32_t node_inst, uint32_t umc_inst,
+					uint32_t ch_inst, void *data)
 {
 	uint32_t eccinfo_table_idx;
 	uint64_t mc_umc_status, err_addr;
-
+	struct ras_err_data *err_data = (struct ras_err_data *)data;
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 
 	eccinfo_table_idx = node_inst * adev->umc.umc_inst_num *
@@ -447,10 +417,10 @@ static void umc_v8_10_ecc_info_query_error_address(struct amdgpu_device *adev,
 	mc_umc_status = ras->umc_ecc.ecc[eccinfo_table_idx].mca_umc_status;
 
 	if (mc_umc_status == 0)
-		return;
+		return 0;
 
 	if (!err_data->err_addr)
-		return;
+		return 0;
 
 	/* calculate error address if ue error is detected */
 	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
@@ -463,28 +433,15 @@ static void umc_v8_10_ecc_info_query_error_address(struct amdgpu_device *adev,
 		umc_v8_10_convert_error_address(adev, err_data, err_addr,
 					ch_inst, umc_inst, node_inst, mc_umc_status);
 	}
+
+	return 0;
 }
 
 static void umc_v8_10_ecc_info_query_ras_error_address(struct amdgpu_device *adev,
 					void *ras_error_status)
 {
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
-
-	uint32_t node_inst       = 0;
-	uint32_t umc_inst        = 0;
-	uint32_t ch_inst         = 0;
-
-	/* TODO: driver needs to toggle DF Cstate to ensure
-	 * safe access of UMC resgisters. Will add the protection
-	 * when firmware interface is ready
-	 */
-	LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
-		umc_v8_10_ecc_info_query_error_address(adev,
-						err_data,
-						ch_inst,
-						umc_inst,
-						node_inst);
-	}
+	amdgpu_umc_loop_channels(adev,
+		umc_v8_10_ecc_info_query_error_address, ras_error_status);
 }
 
 const struct amdgpu_ras_block_hw_ops umc_v8_10_ras_hw_ops = {
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
index e407be6cb63c..e32b656b3dab 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
@@ -444,6 +444,7 @@ static int uvd_v7_0_sw_init(void *handle)
 			continue;
 		if (!amdgpu_sriov_vf(adev)) {
 			ring = &adev->uvd.inst[j].ring;
+			ring->vm_hub = AMDGPU_MMHUB_0;
 			sprintf(ring->name, "uvd_%d", ring->me);
 			r = amdgpu_ring_init(adev, ring, 512,
 					     &adev->uvd.inst[j].irq, 0,
@@ -454,6 +455,7 @@ static int uvd_v7_0_sw_init(void *handle)
 
 		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
 			ring = &adev->uvd.inst[j].ring_enc[i];
+			ring->vm_hub = AMDGPU_MMHUB_0;
 			sprintf(ring->name, "uvd_enc_%d.%d", ring->me, i);
 			if (amdgpu_sriov_vf(adev)) {
 				ring->use_doorbell = true;
@@ -1397,7 +1399,7 @@ static void uvd_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 static void uvd_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
@@ -1440,7 +1442,7 @@ static void uvd_v7_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
 static void uvd_v7_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					    unsigned int vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
@@ -1802,7 +1804,6 @@ static const struct amdgpu_ring_funcs uvd_v7_0_ring_vm_funcs = {
 	.align_mask = 0xf,
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = uvd_v7_0_ring_get_rptr,
 	.get_wptr = uvd_v7_0_ring_get_wptr,
 	.set_wptr = uvd_v7_0_ring_set_wptr,
@@ -1835,7 +1836,6 @@ static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
 	.nop = HEVC_ENC_CMD_NO_OP,
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = uvd_v7_0_enc_ring_get_rptr,
 	.get_wptr = uvd_v7_0_enc_ring_get_wptr,
 	.set_wptr = uvd_v7_0_enc_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
index 66cd3d11aa4b..57b85bb6a1e4 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
@@ -466,6 +466,7 @@ static int vce_v4_0_sw_init(void *handle)
 		enum amdgpu_ring_priority_level hw_prio = amdgpu_vce_get_ring_prio(i);
 
 		ring = &adev->vce.ring[i];
+		ring->vm_hub = AMDGPU_MMHUB_0;
 		sprintf(ring->name, "vce%d", i);
 		if (amdgpu_sriov_vf(adev)) {
 			/* DOORBELL only works under SRIOV */
@@ -1021,7 +1022,7 @@ static void vce_v4_0_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 static void vce_v4_0_emit_vm_flush(struct amdgpu_ring *ring,
 				   unsigned int vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
@@ -1103,7 +1104,6 @@ static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = {
 	.nop = VCE_CMD_NO_OP,
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vce_v4_0_ring_get_rptr,
 	.get_wptr = vce_v4_0_ring_get_wptr,
 	.set_wptr = vce_v4_0_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c
index 1ceda3d0cd5b..2b9ddb3d2fe1 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_sw_ring.c
@@ -65,7 +65,7 @@ void vcn_dec_sw_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 void vcn_dec_sw_ring_emit_vm_flush(struct amdgpu_ring *ring,
 	uint32_t vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
index c305b2cb8490..761c28fa6ec1 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
@@ -120,6 +120,7 @@ static int vcn_v1_0_sw_init(void *handle)
 		return r;
 
 	ring = &adev->vcn.inst->ring_dec;
+	ring->vm_hub = AMDGPU_MMHUB_0;
 	sprintf(ring->name, "vcn_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
 			     AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -141,6 +142,7 @@ static int vcn_v1_0_sw_init(void *handle)
 		enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(i);
 
 		ring = &adev->vcn.inst->ring_enc[i];
+		ring->vm_hub = AMDGPU_MMHUB_0;
 		sprintf(ring->name, "vcn_enc%d", i);
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
 				     hw_prio, NULL);
@@ -1548,7 +1550,7 @@ static void vcn_v1_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring,
 static void vcn_v1_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					    unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
@@ -1693,7 +1695,7 @@ static void vcn_v1_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
 static void vcn_v1_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					    unsigned int vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
@@ -1977,7 +1979,6 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v1_0_dec_ring_get_rptr,
 	.get_wptr = vcn_v1_0_dec_ring_get_wptr,
 	.set_wptr = vcn_v1_0_dec_ring_set_wptr,
@@ -2012,7 +2013,6 @@ static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
 	.nop = VCN_ENC_CMD_NO_OP,
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v1_0_enc_ring_get_rptr,
 	.get_wptr = vcn_v1_0_enc_ring_get_wptr,
 	.set_wptr = vcn_v1_0_enc_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
index 4b4cd88414e0..7c2b3aa48083 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
@@ -129,6 +129,7 @@ static int vcn_v2_0_sw_init(void *handle)
 
 	ring->use_doorbell = true;
 	ring->doorbell_index = adev->doorbell_index.vcn.vcn_ring0_1 << 1;
+	ring->vm_hub = AMDGPU_MMHUB_0;
 
 	sprintf(ring->name, "vcn_dec");
 	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
@@ -159,6 +160,7 @@ static int vcn_v2_0_sw_init(void *handle)
 
 		ring = &adev->vcn.inst->ring_enc[i];
 		ring->use_doorbell = true;
+		ring->vm_hub = AMDGPU_MMHUB_0;
 		if (!amdgpu_sriov_vf(adev))
 			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + i;
 		else
@@ -1511,7 +1513,7 @@ void vcn_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 void vcn_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
 				unsigned vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 	uint32_t data0, data1, mask;
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
@@ -1671,7 +1673,7 @@ void vcn_v2_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 void vcn_v2_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
 				unsigned int vmid, uint64_t pd_addr)
 {
-	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
 
 	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
@@ -2014,7 +2016,6 @@ static const struct amdgpu_ring_funcs vcn_v2_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_DEC,
 	.align_mask = 0xf,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v2_0_dec_ring_get_rptr,
 	.get_wptr = vcn_v2_0_dec_ring_get_wptr,
 	.set_wptr = vcn_v2_0_dec_ring_set_wptr,
@@ -2045,7 +2046,6 @@ static const struct amdgpu_ring_funcs vcn_v2_0_enc_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_ENC,
 	.align_mask = 0x3f,
 	.nop = VCN_ENC_CMD_NO_OP,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v2_0_enc_ring_get_rptr,
 	.get_wptr = vcn_v2_0_enc_ring_get_wptr,
 	.set_wptr = vcn_v2_0_enc_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
index 223e7dfe4618..ab0b45d0ead1 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
@@ -186,6 +186,12 @@ static int vcn_v2_5_sw_init(void *handle)
 
 		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
 				(amdgpu_sriov_vf(adev) ? 2*j : 8*j);
+
+		if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0))
+			ring->vm_hub = AMDGPU_MMHUB_1;
+		else
+			ring->vm_hub = AMDGPU_MMHUB_0;
+
 		sprintf(ring->name, "vcn_dec_%d", j);
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq,
 				     0, AMDGPU_RING_PRIO_DEFAULT, NULL);
@@ -201,6 +207,11 @@ static int vcn_v2_5_sw_init(void *handle)
 			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
 					(amdgpu_sriov_vf(adev) ? (1 + i + 2*j) : (2 + i + 8*j));
 
+			if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0))
+				ring->vm_hub = AMDGPU_MMHUB_1;
+			else
+				ring->vm_hub = AMDGPU_MMHUB_0;
+
 			sprintf(ring->name, "vcn_enc_%d.%d", j, i);
 			r = amdgpu_ring_init(adev, ring, 512,
 					     &adev->vcn.inst[j].irq, 0,
@@ -1562,38 +1573,6 @@ static const struct amdgpu_ring_funcs vcn_v2_5_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_DEC,
 	.align_mask = 0xf,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_1,
-	.get_rptr = vcn_v2_5_dec_ring_get_rptr,
-	.get_wptr = vcn_v2_5_dec_ring_get_wptr,
-	.set_wptr = vcn_v2_5_dec_ring_set_wptr,
-	.emit_frame_size =
-		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
-		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
-		8 + /* vcn_v2_0_dec_ring_emit_vm_flush */
-		14 + 14 + /* vcn_v2_0_dec_ring_emit_fence x2 vm fence */
-		6,
-	.emit_ib_size = 8, /* vcn_v2_0_dec_ring_emit_ib */
-	.emit_ib = vcn_v2_0_dec_ring_emit_ib,
-	.emit_fence = vcn_v2_0_dec_ring_emit_fence,
-	.emit_vm_flush = vcn_v2_0_dec_ring_emit_vm_flush,
-	.test_ring = vcn_v2_0_dec_ring_test_ring,
-	.test_ib = amdgpu_vcn_dec_ring_test_ib,
-	.insert_nop = vcn_v2_0_dec_ring_insert_nop,
-	.insert_start = vcn_v2_0_dec_ring_insert_start,
-	.insert_end = vcn_v2_0_dec_ring_insert_end,
-	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
-	.end_use = amdgpu_vcn_ring_end_use,
-	.emit_wreg = vcn_v2_0_dec_ring_emit_wreg,
-	.emit_reg_wait = vcn_v2_0_dec_ring_emit_reg_wait,
-	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
-};
-
-static const struct amdgpu_ring_funcs vcn_v2_6_dec_ring_vm_funcs = {
-	.type = AMDGPU_RING_TYPE_VCN_DEC,
-	.align_mask = 0xf,
-	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v2_5_dec_ring_get_rptr,
 	.get_wptr = vcn_v2_5_dec_ring_get_wptr,
 	.set_wptr = vcn_v2_5_dec_ring_set_wptr,
@@ -1693,7 +1672,6 @@ static const struct amdgpu_ring_funcs vcn_v2_5_enc_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_ENC,
 	.align_mask = 0x3f,
 	.nop = VCN_ENC_CMD_NO_OP,
-	.vmhub = AMDGPU_MMHUB_1,
 	.get_rptr = vcn_v2_5_enc_ring_get_rptr,
 	.get_wptr = vcn_v2_5_enc_ring_get_wptr,
 	.set_wptr = vcn_v2_5_enc_ring_set_wptr,
@@ -1719,36 +1697,6 @@ static const struct amdgpu_ring_funcs vcn_v2_5_enc_ring_vm_funcs = {
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
-static const struct amdgpu_ring_funcs vcn_v2_6_enc_ring_vm_funcs = {
-        .type = AMDGPU_RING_TYPE_VCN_ENC,
-        .align_mask = 0x3f,
-        .nop = VCN_ENC_CMD_NO_OP,
-        .vmhub = AMDGPU_MMHUB_0,
-        .get_rptr = vcn_v2_5_enc_ring_get_rptr,
-        .get_wptr = vcn_v2_5_enc_ring_get_wptr,
-        .set_wptr = vcn_v2_5_enc_ring_set_wptr,
-        .emit_frame_size =
-                SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
-                SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
-                4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
-                5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
-                1, /* vcn_v2_0_enc_ring_insert_end */
-        .emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
-        .emit_ib = vcn_v2_0_enc_ring_emit_ib,
-        .emit_fence = vcn_v2_0_enc_ring_emit_fence,
-        .emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
-        .test_ring = amdgpu_vcn_enc_ring_test_ring,
-        .test_ib = amdgpu_vcn_enc_ring_test_ib,
-        .insert_nop = amdgpu_ring_insert_nop,
-        .insert_end = vcn_v2_0_enc_ring_insert_end,
-        .pad_ib = amdgpu_ring_generic_pad_ib,
-        .begin_use = amdgpu_vcn_ring_begin_use,
-        .end_use = amdgpu_vcn_ring_end_use,
-        .emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
-        .emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
-        .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
-};
-
 static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev)
 {
 	int i;
@@ -1756,10 +1704,7 @@ static void vcn_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev)
 	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
 		if (adev->vcn.harvest_config & (1 << i))
 			continue;
-		if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0))
-			adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_5_dec_ring_vm_funcs;
-		else /* CHIP_ALDEBARAN */
-			adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_6_dec_ring_vm_funcs;
+		adev->vcn.inst[i].ring_dec.funcs = &vcn_v2_5_dec_ring_vm_funcs;
 		adev->vcn.inst[i].ring_dec.me = i;
 		DRM_INFO("VCN(%d) decode is enabled in VM mode\n", i);
 	}
@@ -1773,10 +1718,7 @@ static void vcn_v2_5_set_enc_ring_funcs(struct amdgpu_device *adev)
 		if (adev->vcn.harvest_config & (1 << j))
 			continue;
 		for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
-			if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0))
-				adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_5_enc_ring_vm_funcs;
-			else /* CHIP_ALDEBARAN */
-				adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_6_enc_ring_vm_funcs;
+			adev->vcn.inst[j].ring_enc[i].funcs = &vcn_v2_5_enc_ring_vm_funcs;
 			adev->vcn.inst[j].ring_enc[i].me = j;
 		}
 		DRM_INFO("VCN(%d) encode is enabled in VM mode\n", j);
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
index 66439388faee..3eab186261aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
@@ -189,6 +189,7 @@ static int vcn_v3_0_sw_init(void *handle)
 		} else {
 			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i;
 		}
+		ring->vm_hub = AMDGPU_MMHUB_0;
 		sprintf(ring->name, "vcn_dec_%d", i);
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
 				     AMDGPU_RING_PRIO_DEFAULT,
@@ -212,6 +213,7 @@ static int vcn_v3_0_sw_init(void *handle)
 			} else {
 				ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + j + 8 * i;
 			}
+			ring->vm_hub = AMDGPU_MMHUB_0;
 			sprintf(ring->name, "vcn_enc_%d.%d", i, j);
 			r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
 					     hw_prio, &adev->vcn.inst[i].sched_score);
@@ -1738,7 +1740,6 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = {
 	.align_mask = 0x3f,
 	.nop = VCN_DEC_SW_CMD_NO_OP,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v3_0_dec_ring_get_rptr,
 	.get_wptr = vcn_v3_0_dec_ring_get_wptr,
 	.set_wptr = vcn_v3_0_dec_ring_set_wptr,
@@ -1899,7 +1900,6 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_DEC,
 	.align_mask = 0xf,
 	.secure_submission_supported = true,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v3_0_dec_ring_get_rptr,
 	.get_wptr = vcn_v3_0_dec_ring_get_wptr,
 	.set_wptr = vcn_v3_0_dec_ring_set_wptr,
@@ -2000,7 +2000,6 @@ static const struct amdgpu_ring_funcs vcn_v3_0_enc_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_ENC,
 	.align_mask = 0x3f,
 	.nop = VCN_ENC_CMD_NO_OP,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v3_0_enc_ring_get_rptr,
 	.get_wptr = vcn_v3_0_enc_ring_get_wptr,
 	.set_wptr = vcn_v3_0_enc_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
index 720ab36f9c92..bf0674039598 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
@@ -149,7 +149,7 @@ static int vcn_v4_0_sw_init(void *handle)
 			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + i * (adev->vcn.num_enc_rings + 1) + 1;
 		else
 			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + 8 * i;
-
+		ring->vm_hub = AMDGPU_MMHUB_0;
 		sprintf(ring->name, "vcn_unified_%d", i);
 
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
@@ -1798,7 +1798,6 @@ static const struct amdgpu_ring_funcs vcn_v4_0_unified_ring_vm_funcs = {
 	.type = AMDGPU_RING_TYPE_VCN_ENC,
 	.align_mask = 0x3f,
 	.nop = VCN_ENC_CMD_NO_OP,
-	.vmhub = AMDGPU_MMHUB_0,
 	.get_rptr = vcn_v4_0_unified_ring_get_rptr,
 	.get_wptr = vcn_v4_0_unified_ring_get_wptr,
 	.set_wptr = vcn_v4_0_unified_ring_set_wptr,
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
index 827e2768f867..536128447b71 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
@@ -38,6 +38,11 @@
 #define mmIH_CHICKEN_ALDEBARAN			0x18d
 #define mmIH_CHICKEN_ALDEBARAN_BASE_IDX		0
 
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN		0x00ea
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN_BASE_IDX	0
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE__SHIFT	0x10
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE_MASK	0x00010000L
+
 static void vega20_ih_set_interrupt_funcs(struct amdgpu_device *adev);
 
 /**
@@ -251,36 +256,14 @@ static int vega20_ih_enable_ring(struct amdgpu_device *adev,
 	return 0;
 }
 
-/**
- * vega20_ih_reroute_ih - reroute VMC/UTCL2 ih to an ih ring
- *
- * @adev: amdgpu_device pointer
- *
- * Reroute VMC and UMC interrupts on primary ih ring to
- * ih ring 1 so they won't lose when bunches of page faults
- * interrupts overwhelms the interrupt handler(VEGA20)
- */
-static void vega20_ih_reroute_ih(struct amdgpu_device *adev)
+static uint32_t vega20_setup_retry_doorbell(u32 doorbell_index)
 {
-	uint32_t tmp;
+	u32 val = 0;
 
-	/* vega20 ih reroute will go through psp this
-	 * function is used for newer asics starting arcturus
-	 */
-	if (adev->ip_versions[OSSSYS_HWIP][0] >= IP_VERSION(4, 2, 1)) {
-		/* Reroute to IH ring 1 for VMC */
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x12);
-		tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, CLIENT_TYPE, 1);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA, tmp);
-
-		/* Reroute IH ring 1 for UTCL2 */
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x1B);
-		tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA, tmp);
-	}
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, OFFSET, doorbell_index);
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, ENABLE, 1);
+
+	return val;
 }
 
 /**
@@ -333,8 +316,6 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev)
 
 	for (i = 0; i < ARRAY_SIZE(ih); i++) {
 		if (ih[i]->ring_size) {
-			if (i == 1)
-				vega20_ih_reroute_ih(adev);
 			ret = vega20_ih_enable_ring(adev, ih[i]);
 			if (ret)
 				return ret;
@@ -347,6 +328,20 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev)
 
 	pci_set_master(adev->pdev);
 
+	/* Allocate the doorbell for IH Retry CAM */
+	adev->irq.retry_cam_doorbell_index = (adev->doorbell_index.ih + 3) << 1;
+	WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RETRY_CAM,
+		vega20_setup_retry_doorbell(adev->irq.retry_cam_doorbell_index));
+
+	/* Enable IH Retry CAM */
+	if (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 0))
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL_ALDEBARAN,
+			       ENABLE, 1);
+	else
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL, ENABLE, 1);
+
+	adev->irq.retry_cam_enabled = true;
+
 	/* enable interrupts */
 	ret = vega20_ih_toggle_interrupts(adev, true);
 	if (ret)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 0bf8d75950f7..00f528eb9812 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -204,6 +204,14 @@ static void kfd_device_info_init(struct kfd_dev *kfd,
 			/* Navi1x+ */
 			if (gc_version >= IP_VERSION(10, 1, 1))
 				kfd->device_info.needs_pci_atomics = true;
+		} else if (gc_version < IP_VERSION(12, 0, 0)) {
+			/*
+			 * PCIe atomics support acknowledgment in GFX11 RS64 CPFW requires
+			 * MEC version >= 509. Prior RS64 CPFW versions (and all F32) require
+			 * PCIe atomics support.
+			 */
+			kfd->device_info.needs_pci_atomics = true;
+			kfd->device_info.no_atomic_fw_version = kfd->adev->gfx.rs64_enable ? 509 : 0;
 		}
 	} else {
 		kfd->device_info.doorbell_size = 4;
@@ -315,10 +323,13 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf)
 			break;
 		/* Aldebaran */
 		case IP_VERSION(9, 4, 2):
-		case IP_VERSION(9, 4, 3):
 			gfx_target_version = 90010;
 			f2g = &aldebaran_kfd2kgd;
 			break;
+		case IP_VERSION(9, 4, 3):
+			gfx_target_version = 90400;
+			f2g = &aldebaran_kfd2kgd;
+			break;
 		/* Navi10 */
 		case IP_VERSION(10, 1, 10):
 			gfx_target_version = 100100;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c
index 4a9af800b1f1..5aa75f72caa1 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v11.c
@@ -143,6 +143,13 @@ static void init_mqd(struct mqd_manager *mm, void **mqd,
 			1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
 			1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
 
+	/*
+	 * GFX11 RS64 CPFW version >= 509 supports PCIe atomics support
+	 * acknowledgment.
+	 */
+	if (amdgpu_amdkfd_have_atomics_support(mm->dev->adev))
+		m->cp_hqd_hq_status0 |= 1 << 29;
+
 	if (q->format == KFD_QUEUE_FORMAT_AQL) {
 		m->cp_hqd_aql_control =
 			1 << CP_HQD_AQL_CONTROL__CONTROL0__SHIFT;
@@ -350,6 +357,10 @@ static void update_mqd_sdma(struct mqd_manager *mm, void *mqd,
 	m->sdmax_rlcx_doorbell_offset =
 		q->doorbell_off << SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT;
 
+	m->sdmax_rlcx_sched_cntl = (amdgpu_sdma_phase_quantum
+		<< SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM__SHIFT)
+		 & SDMA0_QUEUE0_SCHEDULE_CNTL__CONTEXT_QUANTUM_MASK;
+
 	m->sdma_engine_id = q->sdma_engine_id;
 	m->sdma_queue_id = q->sdma_queue_id;
 	m->sdmax_rlcx_dummy_reg = SDMA_RLC_DUMMY_DEFAULT;
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 4dfae19714ab..fdbfd725841f 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
@@ -224,6 +224,7 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 			struct queue_properties *q,
 			struct mqd_update_info *minfo)
 {
+	struct amdgpu_device *adev = (struct amdgpu_device *)mm->dev->adev;
 	struct v9_mqd *m;
 
 	m = get_mqd(mqd);
@@ -269,10 +270,13 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 	m->cp_hqd_vmid = q->vmid;
 
 	if (q->format == KFD_QUEUE_FORMAT_AQL) {
-		m->cp_hqd_pq_control |= CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK |
+		m->cp_hqd_pq_control |=
 				2 << CP_HQD_PQ_CONTROL__SLOT_BASED_WPTR__SHIFT |
 				1 << CP_HQD_PQ_CONTROL__QUEUE_FULL_EN__SHIFT |
 				1 << CP_HQD_PQ_CONTROL__WPP_CLAMP_EN__SHIFT;
+		if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 3))
+			m->cp_hqd_pq_control |=
+				 CP_HQD_PQ_CONTROL__NO_UPDATE_RPTR_MASK;
 		m->cp_hqd_pq_doorbell_control |= 1 <<
 			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_BIF_DROP__SHIFT;
 	}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index dc6fd6967050..96a138a39515 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -2172,7 +2172,15 @@ restart:
 		pr_debug("drain retry fault gpu %d svms %p\n", i, svms);
 
 		amdgpu_ih_wait_on_checkpoint_process_ts(pdd->dev->adev,
-						     &pdd->dev->adev->irq.ih1);
+				pdd->dev->adev->irq.retry_cam_enabled ?
+				&pdd->dev->adev->irq.ih :
+				&pdd->dev->adev->irq.ih1);
+
+		if (pdd->dev->adev->irq.retry_cam_enabled)
+			amdgpu_ih_wait_on_checkpoint_process_ts(pdd->dev->adev,
+				&pdd->dev->adev->irq.ih_soft);
+
+
 		pr_debug("drain retry fault gpu %d svms 0x%p done\n", i, svms);
 	}
 	if (atomic_cmpxchg(&svms->drain_pagefaults, drain, 0) != drain)
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 e29655a26dd4..6cacb76f389e 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -1890,7 +1890,8 @@ static void amdgpu_dm_fini(struct amdgpu_device *adev)
 	if (adev->dm.dc)
 		dc_deinit_callbacks(adev->dm.dc);
 
-	dc_dmub_srv_destroy(&adev->dm.dc->ctx->dmub_srv);
+	if (adev->dm.dc)
+		dc_dmub_srv_destroy(&adev->dm.dc->ctx->dmub_srv);
 
 	if (dc_enable_dmub_notifications(adev->dm.dc)) {
 		kfree(adev->dm.dmub_notify);
diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
index f0f948501e9a..cce47d3f1a13 100644
--- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
+++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
@@ -515,11 +515,8 @@ static enum bp_result get_gpio_i2c_info(
 	info->i2c_slave_address = record->i2c_slave_addr;
 
 	/* TODO: check how to get register offset for en, Y, etc. */
-	info->gpio_info.clk_a_register_index =
-			le16_to_cpu(
-			header->gpio_pin[table_index].data_a_reg_index);
-	info->gpio_info.clk_a_shift =
-			header->gpio_pin[table_index].gpio_bitshift;
+	info->gpio_info.clk_a_register_index = le16_to_cpu(pin->data_a_reg_index);
+	info->gpio_info.clk_a_shift = pin->gpio_bitshift;
 
 	return BP_RESULT_OK;
 }
diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c
index af108f88b112..ea753f8fa175 100644
--- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c
+++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c
@@ -399,6 +399,23 @@ static void dcn32_update_clocks_update_dentist(
 
 }
 
+static int dcn32_get_dispclk_from_dentist(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dispclk_wdivider;
+	int disp_divider;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, &dispclk_wdivider);
+	disp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+
+	/* Return DISPCLK freq in Khz */
+	if (disp_divider)
+		return (DENTIST_DIVIDER_RANGE_SCALE_FACTOR * clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+
+	return 0;
+}
+
+
 static void dcn32_update_clocks(struct clk_mgr *clk_mgr_base,
 			struct dc_state *context,
 			bool safe_to_lower)
@@ -852,6 +869,7 @@ static struct clk_mgr_funcs dcn32_funcs = {
 		.are_clock_states_equal = dcn32_are_clock_states_equal,
 		.enable_pme_wa = dcn32_enable_pme_wa,
 		.is_smu_present = dcn32_is_smu_present,
+		.get_dispclk_from_dentist = dcn32_get_dispclk_from_dentist,
 };
 
 void dcn32_clk_mgr_construct(
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c
index 41198c729d90..30c0644d4418 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c
@@ -622,7 +622,6 @@ bool link_enc_cfg_validate(struct dc *dc, struct dc_state *state)
 	int i, j;
 	uint8_t valid_count = 0;
 	uint8_t dig_stream_count = 0;
-	int matching_stream_ptrs = 0;
 	int eng_ids_per_ep_id[MAX_PIPES] = {0};
 	int ep_ids_per_eng_id[MAX_PIPES] = {0};
 	int valid_bitmap = 0;
@@ -645,9 +644,7 @@ bool link_enc_cfg_validate(struct dc *dc, struct dc_state *state)
 		struct link_enc_assignment assignment = state->res_ctx.link_enc_cfg_ctx.link_enc_assignments[i];
 
 		if (assignment.valid) {
-			if (assignment.stream == state->streams[i])
-				matching_stream_ptrs++;
-			else
+			if (assignment.stream != state->streams[i])
 				valid_stream_ptrs = false;
 		}
 	}
diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index e363a3c88250..23ee63b98dcd 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -45,7 +45,7 @@ struct aux_payload;
 struct set_config_cmd_payload;
 struct dmub_notification;
 
-#define DC_VER "3.2.229"
+#define DC_VER "3.2.230"
 
 #define MAX_SURFACES 3
 #define MAX_PLANES 6
@@ -405,6 +405,7 @@ struct dc_config {
 	bool force_bios_enable_lttpr;
 	uint8_t force_bios_fixed_vs;
 	int sdpif_request_limit_words_per_umc;
+	bool use_old_fixed_vs_sequence;
 	bool disable_subvp_drr;
 };
 
@@ -875,6 +876,9 @@ struct dc_debug_options {
 	bool override_dispclk_programming;
 	bool disable_fpo_optimizations;
 	bool support_eDP1_5;
+	uint32_t fpo_vactive_margin_us;
+	bool disable_fpo_vactive;
+	bool disable_boot_optimizations;
 };
 
 struct gpu_info_soc_bounding_box_v1_0;
diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c
index dd6f643254fe..a9b9490a532c 100644
--- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c
+++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c
@@ -327,6 +327,7 @@ bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, stru
 	int i = 0, k = 0;
 	int ramp_up_num_steps = 1; // TODO: Ramp is currently disabled. Reenable it.
 	uint8_t visual_confirm_enabled;
+	int pipe_idx = 0;
 
 	if (dc == NULL)
 		return false;
@@ -339,6 +340,25 @@ bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, stru
 	cmd.fw_assisted_mclk_switch.config_data.fams_enabled = should_manage_pstate;
 	cmd.fw_assisted_mclk_switch.config_data.visual_confirm_enabled = visual_confirm_enabled;
 
+	if (should_manage_pstate) {
+		for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+			struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+			if (!pipe->stream)
+				continue;
+
+			/* If FAMS is being used to support P-State and there is a stream
+			 * that does not use FAMS, we are in an FPO + VActive scenario.
+			 * Assign vactive stretch margin in this case.
+			 */
+			if (!pipe->stream->fpo_in_use) {
+				cmd.fw_assisted_mclk_switch.config_data.vactive_stretch_margin_us = dc->debug.fpo_vactive_margin_us;
+				break;
+			}
+			pipe_idx++;
+		}
+	}
+
 	for (i = 0, k = 0; context && i < dc->res_pool->pipe_count; i++) {
 		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
 
diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c
index a142a00bc432..bf399819ca80 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c
@@ -755,8 +755,8 @@ bool hubp1_is_flip_pending(struct hubp *hubp)
 	return false;
 }
 
-uint32_t aperture_default_system = 1;
-uint32_t context0_default_system; /* = 0;*/
+static uint32_t aperture_default_system = 1;
+static uint32_t context0_default_system; /* = 0;*/
 
 static void hubp1_set_vm_system_aperture_settings(struct hubp *hubp,
 		struct vm_system_aperture_param *apt)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.h
index 893c0809cd4e..7bdc146f7cb5 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.h
@@ -205,6 +205,11 @@
 	type PHYDSYMCLK_GATE_DISABLE; \
 	type PHYESYMCLK_GATE_DISABLE;
 
+#define DCCG314_REG_FIELD_LIST(type) \
+	type DSCCLK3_DTO_PHASE;\
+	type DSCCLK3_DTO_MODULO;\
+	type DSCCLK3_DTO_ENABLE;
+
 #define DCCG32_REG_FIELD_LIST(type) \
 	type DPSTREAMCLK0_EN;\
 	type DPSTREAMCLK1_EN;\
@@ -237,6 +242,7 @@ struct dccg_shift {
 	DCCG_REG_FIELD_LIST(uint8_t)
 	DCCG3_REG_FIELD_LIST(uint8_t)
 	DCCG31_REG_FIELD_LIST(uint8_t)
+	DCCG314_REG_FIELD_LIST(uint8_t)
 	DCCG32_REG_FIELD_LIST(uint8_t)
 };
 
@@ -244,6 +250,7 @@ struct dccg_mask {
 	DCCG_REG_FIELD_LIST(uint32_t)
 	DCCG3_REG_FIELD_LIST(uint32_t)
 	DCCG31_REG_FIELD_LIST(uint32_t)
+	DCCG314_REG_FIELD_LIST(uint32_t)
 	DCCG32_REG_FIELD_LIST(uint32_t)
 };
 
@@ -273,6 +280,7 @@ struct dccg_registers {
 	uint32_t DSCCLK0_DTO_PARAM;
 	uint32_t DSCCLK1_DTO_PARAM;
 	uint32_t DSCCLK2_DTO_PARAM;
+	uint32_t DSCCLK3_DTO_PARAM;
 	uint32_t DPSTREAMCLK_ROOT_GATE_DISABLE;
 	uint32_t DPSTREAMCLK_GATE_DISABLE;
 	uint32_t DCCG_GATE_DISABLE_CNTL;
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c
index 7d2b982506fd..4c2fdfea162f 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c
@@ -360,6 +360,15 @@ void dccg31_disable_dscclk(struct dccg *dccg, int inst)
 				DSCCLK2_DTO_PHASE, 0,
 				DSCCLK2_DTO_MODULO, 1);
 		break;
+	case 3:
+		if (REG(DSCCLK3_DTO_PARAM)) {
+			REG_UPDATE(DSCCLK_DTO_CTRL,
+					DSCCLK3_DTO_ENABLE, 1);
+			REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+					DSCCLK3_DTO_PHASE, 0,
+					DSCCLK3_DTO_MODULO, 1);
+		}
+		break;
 	default:
 		BREAK_TO_DEBUGGER();
 		return;
@@ -395,6 +404,15 @@ void dccg31_enable_dscclk(struct dccg *dccg, int inst)
 		REG_UPDATE(DSCCLK_DTO_CTRL,
 				DSCCLK2_DTO_ENABLE, 0);
 		break;
+	case 3:
+		if (REG(DSCCLK3_DTO_PARAM)) {
+			REG_UPDATE(DSCCLK_DTO_CTRL,
+					DSCCLK3_DTO_ENABLE, 0);
+			REG_UPDATE_2(DSCCLK3_DTO_PARAM,
+					DSCCLK3_DTO_PHASE, 0,
+					DSCCLK3_DTO_MODULO, 0);
+		}
+		break;
 	default:
 		BREAK_TO_DEBUGGER();
 		return;
diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
index eaaa2e01f6d0..ff8cd5076434 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c
@@ -1965,6 +1965,8 @@ static bool dcn31_resource_construct(
 	dc->caps.color.mpc.ogam_rom_caps.hlg = 0;
 	dc->caps.color.mpc.ocsc = 1;
 
+	dc->config.use_old_fixed_vs_sequence = true;
+
 	/* Use pipe context based otg sync logic */
 	dc->config.use_pipe_ctx_sync_logic = true;
 
diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c
index 081ce168f621..6f879265ad9c 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c
@@ -274,6 +274,32 @@ static void dccg314_set_dpstreamclk(
 	}
 }
 
+void dccg314_init(struct dccg *dccg)
+{
+	int otg_inst;
+
+	/* Set HPO stream encoder to use refclk to avoid case where PHY is
+	 * disabled and SYMCLK32 for HPO SE is sourced from PHYD32CLK which
+	 * will cause DCN to hang.
+	 */
+	for (otg_inst = 0; otg_inst < 4; otg_inst++)
+		dccg31_disable_symclk32_se(dccg, otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.symclk32_le)
+		for (otg_inst = 0; otg_inst < 2; otg_inst++)
+			dccg31_disable_symclk32_le(dccg, otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.dpstream)
+		for (otg_inst = 0; otg_inst < 4; otg_inst++)
+			dccg314_set_dpstreamclk(dccg, REFCLK, otg_inst,
+						otg_inst);
+
+	if (dccg->ctx->dc->debug.root_clock_optimization.bits.physymclk)
+		for (otg_inst = 0; otg_inst < 5; otg_inst++)
+			dccg31_set_physymclk(dccg, otg_inst,
+					     PHYSYMCLK_FORCE_SRC_SYMCLK, false);
+}
+
 static void dccg314_set_valid_pixel_rate(
 		struct dccg *dccg,
 		int ref_dtbclk_khz,
@@ -315,7 +341,7 @@ static const struct dccg_funcs dccg314_funcs = {
 	.update_dpp_dto = dccg31_update_dpp_dto,
 	.dpp_root_clock_control = dccg314_dpp_root_clock_control,
 	.get_dccg_ref_freq = dccg31_get_dccg_ref_freq,
-	.dccg_init = dccg31_init,
+	.dccg_init = dccg314_init,
 	.set_dpstreamclk = dccg314_set_dpstreamclk,
 	.enable_symclk32_se = dccg31_enable_symclk32_se,
 	.disable_symclk32_se = dccg31_disable_symclk32_se,
diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.h b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.h
index 6a35986307af..90687a9e8fdd 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.h
@@ -68,6 +68,7 @@
 	SR(DSCCLK0_DTO_PARAM),\
 	SR(DSCCLK1_DTO_PARAM),\
 	SR(DSCCLK2_DTO_PARAM),\
+	SR(DSCCLK3_DTO_PARAM),\
 	SR(DSCCLK_DTO_CTRL),\
 	SR(DCCG_GATE_DISABLE_CNTL2),\
 	SR(DCCG_GATE_DISABLE_CNTL3),\
@@ -149,12 +150,20 @@
 	DCCG_SF(DSCCLK1_DTO_PARAM, DSCCLK1_DTO_MODULO, mask_sh),\
 	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_PHASE, mask_sh),\
 	DCCG_SF(DSCCLK2_DTO_PARAM, DSCCLK2_DTO_MODULO, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_PHASE, mask_sh),\
+	DCCG_SF(DSCCLK3_DTO_PARAM, DSCCLK3_DTO_MODULO, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE0_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE1_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE2_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_SE3_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE0_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_ROOT_LE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE1_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE2_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_SE3_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE0_GATE_DISABLE, mask_sh),\
+	DCCG_SF(DCCG_GATE_DISABLE_CNTL3, SYMCLK32_LE1_GATE_DISABLE, mask_sh),\
 	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_PHASE, mask_sh),\
 	DCCG_SF(HDMISTREAMCLK0_DTO_PARAM, HDMISTREAMCLK0_DTO_MODULO, mask_sh)
 
@@ -178,6 +187,7 @@
 	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK0_DTO_ENABLE, mask_sh),\
 	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK1_DTO_ENABLE, mask_sh),\
 	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK2_DTO_ENABLE, mask_sh),\
+	DCCG_SF(DSCCLK_DTO_CTRL, DSCCLK3_DTO_ENABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYASYMCLK_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYBSYMCLK_GATE_DISABLE, mask_sh),\
 	DCCG_SF(DCCG_GATE_DISABLE_CNTL2, PHYCSYMCLK_GATE_DISABLE, mask_sh),\
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c
index b3824287c224..db0974fe58ab 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c
@@ -721,10 +721,19 @@ static void dcn32_initialize_min_clocks(struct dc *dc)
 	clocks->socclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].socclk_mhz * 1000;
 	clocks->dramclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].memclk_mhz * 1000;
 	clocks->dppclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].dppclk_mhz * 1000;
-	clocks->dispclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].dispclk_mhz * 1000;
-	clocks->ref_dtbclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].dtbclk_mhz * 1000;
-	clocks->fclk_p_state_change_support = true;
-	clocks->p_state_change_support = true;
+	if (dc->debug.disable_boot_optimizations) {
+		clocks->dispclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].dispclk_mhz * 1000;
+	} else {
+		/* Even though DPG_EN = 1 for the connected display, it still requires the
+		 * correct timing so we cannot set DISPCLK to min freq or it could cause
+		 * audio corruption. Read current DISPCLK from DENTIST and request the same
+		 * freq to ensure that the timing is valid and unchanged.
+		 */
+		clocks->dispclk_khz = dc->clk_mgr->funcs->get_dispclk_from_dentist(dc->clk_mgr);
+		clocks->ref_dtbclk_khz = dc->clk_mgr->bw_params->clk_table.entries[0].dtbclk_mhz * 1000;
+		clocks->fclk_p_state_change_support = true;
+		clocks->p_state_change_support = true;
+	}
 
 	dc->clk_mgr->funcs->update_clocks(
 			dc->clk_mgr,
@@ -823,7 +832,14 @@ void dcn32_init_hw(struct dc *dc)
 	 * everything down.
 	 */
 	if (dcb->funcs->is_accelerated_mode(dcb) || !dc->config.seamless_boot_edp_requested) {
-		hws->funcs.init_pipes(dc, dc->current_state);
+		/* Disable boot optimizations means power down everything including PHY, DIG,
+		 * and OTG (i.e. the boot is not optimized because we do a full power down).
+		 */
+		if (dc->hwss.enable_accelerated_mode && dc->debug.disable_boot_optimizations)
+			dc->hwss.enable_accelerated_mode(dc, dc->current_state);
+		else
+			hws->funcs.init_pipes(dc, dc->current_state);
+
 		if (dc->res_pool->hubbub->funcs->allow_self_refresh_control)
 			dc->res_pool->hubbub->funcs->allow_self_refresh_control(dc->res_pool->hubbub,
 					!dc->res_pool->hubbub->ctx->dc->debug.disable_stutter);
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
index 502f990346b1..e30d1f60695d 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
@@ -726,6 +726,9 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.disable_unbounded_requesting = false,
 	.override_dispclk_programming = true,
 	.disable_fpo_optimizations = false,
+	.fpo_vactive_margin_us = 2000, // 2000us
+	.disable_fpo_vactive = true,
+	.disable_boot_optimizations = false,
 };
 
 static const struct dc_debug_options debug_defaults_diags = {
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h
index 10a3350376e4..3937dbc1e552 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h
@@ -39,6 +39,7 @@
 #define DCN3_2_MBLK_HEIGHT_8BPE 64
 #define DCN3_2_VMIN_DISPCLK_HZ 717000000
 #define DCN3_2_DCFCLK_DS_INIT_KHZ 10000 // Choose 10Mhz for init DCFCLK DS freq
+#define DCN3_2_MIN_ACTIVE_SWITCH_MARGIN_FPO_US 100 // Only allow FPO + Vactive if active margin >= 100
 
 #define TO_DCN32_RES_POOL(pool)\
 	container_of(pool, struct dcn32_resource_pool, base)
@@ -146,6 +147,8 @@ void dcn32_restore_mall_state(struct dc *dc,
 		struct dc_state *context,
 		struct mall_temp_config *temp_config);
 
+struct dc_stream_state *dcn32_can_support_mclk_switch_using_fw_based_vblank_stretch(struct dc *dc, const struct dc_state *context);
+
 bool dcn32_allow_subvp_with_active_margin(struct pipe_ctx *pipe);
 
 unsigned int dcn32_calc_num_avail_chans_for_mall(struct dc *dc, int num_chans);
@@ -472,6 +475,7 @@ double dcn32_determine_max_vratio_prefetch(struct dc *dc, struct dc_state *conte
       SRI_ARR(OTG_H_BLANK, DSCL, id), SRI_ARR(OTG_V_BLANK, DSCL, id),          \
       SRI_ARR(SCL_MODE, DSCL, id), SRI_ARR(LB_DATA_FORMAT, DSCL, id),          \
       SRI_ARR(LB_MEMORY_CTRL, DSCL, id), SRI_ARR(DSCL_AUTOCAL, DSCL, id),      \
+      SRI_ARR(DSCL_CONTROL, DSCL, id),                                         \
       SRI_ARR(SCL_TAP_CONTROL, DSCL, id),                                      \
       SRI_ARR(SCL_COEF_RAM_TAP_SELECT, DSCL, id),                              \
       SRI_ARR(SCL_COEF_RAM_TAP_DATA, DSCL, id),                                \
diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c
index 47fa51c1d3f4..eeca16faf31a 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c
@@ -27,6 +27,7 @@
 #include "dcn32_resource.h"
 #include "dcn20/dcn20_resource.h"
 #include "dml/dcn32/display_mode_vba_util_32.h"
+#include "dml/dcn32/dcn32_fpu.h"
 
 static bool is_dual_plane(enum surface_pixel_format format)
 {
@@ -500,3 +501,158 @@ void dcn32_restore_mall_state(struct dc *dc,
 			pipe->plane_state->is_phantom = temp_config->is_phantom_plane[i];
 	}
 }
+
+#define MAX_STRETCHED_V_BLANK 1000 // in micro-seconds (must ensure to match value in FW)
+/*
+ * Scaling factor for v_blank stretch calculations considering timing in
+ * micro-seconds and pixel clock in 100hz.
+ * Note: the parenthesis are necessary to ensure the correct order of
+ * operation where V_SCALE is used.
+ */
+#define V_SCALE (10000 / MAX_STRETCHED_V_BLANK)
+
+static int get_frame_rate_at_max_stretch_100hz(
+		struct dc_stream_state *fpo_candidate_stream,
+		uint32_t fpo_vactive_margin_us)
+{
+	struct dc_crtc_timing *timing = NULL;
+	uint32_t sec_per_100_lines;
+	uint32_t max_v_blank;
+	uint32_t curr_v_blank;
+	uint32_t v_stretch_max;
+	uint32_t stretched_frame_pix_cnt;
+	uint32_t scaled_stretched_frame_pix_cnt;
+	uint32_t scaled_refresh_rate;
+	uint32_t v_scale;
+
+	if (fpo_candidate_stream == NULL)
+		return 0;
+
+	/* check if refresh rate at least 120hz */
+	timing = &fpo_candidate_stream->timing;
+	if (timing == NULL)
+		return 0;
+
+	v_scale = 10000 / (MAX_STRETCHED_V_BLANK + fpo_vactive_margin_us);
+
+	sec_per_100_lines = timing->pix_clk_100hz / timing->h_total + 1;
+	max_v_blank = sec_per_100_lines / v_scale + 1;
+	curr_v_blank = timing->v_total - timing->v_addressable;
+	v_stretch_max = (max_v_blank > curr_v_blank) ? (max_v_blank - curr_v_blank) : (0);
+	stretched_frame_pix_cnt = (v_stretch_max + timing->v_total) * timing->h_total;
+	scaled_stretched_frame_pix_cnt = stretched_frame_pix_cnt / 10000;
+	scaled_refresh_rate = (timing->pix_clk_100hz) / scaled_stretched_frame_pix_cnt + 1;
+
+	return scaled_refresh_rate;
+
+}
+
+static bool is_refresh_rate_support_mclk_switch_using_fw_based_vblank_stretch(
+		struct dc_stream_state *fpo_candidate_stream, uint32_t fpo_vactive_margin_us)
+{
+	int refresh_rate_max_stretch_100hz;
+	int min_refresh_100hz;
+
+	if (fpo_candidate_stream == NULL)
+		return false;
+
+	refresh_rate_max_stretch_100hz = get_frame_rate_at_max_stretch_100hz(fpo_candidate_stream, fpo_vactive_margin_us);
+	min_refresh_100hz = fpo_candidate_stream->timing.min_refresh_in_uhz / 10000;
+
+	if (refresh_rate_max_stretch_100hz < min_refresh_100hz)
+		return false;
+
+	return true;
+}
+
+static int get_refresh_rate(struct dc_stream_state *fpo_candidate_stream)
+{
+	int refresh_rate = 0;
+	int h_v_total = 0;
+	struct dc_crtc_timing *timing = NULL;
+
+	if (fpo_candidate_stream == NULL)
+		return 0;
+
+	/* check if refresh rate at least 120hz */
+	timing = &fpo_candidate_stream->timing;
+	if (timing == NULL)
+		return 0;
+
+	h_v_total = timing->h_total * timing->v_total;
+	if (h_v_total == 0)
+		return 0;
+
+	refresh_rate = ((timing->pix_clk_100hz * 100) / (h_v_total)) + 1;
+	return refresh_rate;
+}
+
+/**
+ * dcn32_can_support_mclk_switch_using_fw_based_vblank_stretch - Determines if config can support FPO
+ *
+ * @param [in]: dc - current dc state
+ * @param [in]: context - new dc state
+ *
+ * Return: Pointer to FPO stream candidate if config can support FPO, otherwise NULL
+ */
+struct dc_stream_state *dcn32_can_support_mclk_switch_using_fw_based_vblank_stretch(struct dc *dc, const struct dc_state *context)
+{
+	int refresh_rate = 0;
+	const int minimum_refreshrate_supported = 120;
+	struct dc_stream_state *fpo_candidate_stream = NULL;
+	bool is_fpo_vactive = false;
+	uint32_t fpo_vactive_margin_us = 0;
+
+	if (context == NULL)
+		return NULL;
+
+	if (dc->debug.disable_fams)
+		return NULL;
+
+	if (!dc->caps.dmub_caps.mclk_sw)
+		return NULL;
+
+	if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching_shut_down)
+		return NULL;
+
+	/* For FPO we can support up to 2 display configs if:
+	 * - first display uses FPO
+	 * - Second display switches in VACTIVE */
+	if (context->stream_count > 2)
+		return NULL;
+	else if (context->stream_count == 2) {
+		DC_FP_START();
+		dcn32_assign_fpo_vactive_candidate(dc, context, &fpo_candidate_stream);
+		DC_FP_END();
+
+		DC_FP_START();
+		is_fpo_vactive = dcn32_find_vactive_pipe(dc, context, DCN3_2_MIN_ACTIVE_SWITCH_MARGIN_FPO_US);
+		DC_FP_END();
+		if (!is_fpo_vactive || dc->debug.disable_fpo_vactive)
+			return NULL;
+	} else
+		fpo_candidate_stream = context->streams[0];
+
+	if (!fpo_candidate_stream)
+		return NULL;
+
+	if (fpo_candidate_stream->sink->edid_caps.panel_patch.disable_fams)
+		return NULL;
+
+	refresh_rate = get_refresh_rate(fpo_candidate_stream);
+	if (refresh_rate < minimum_refreshrate_supported)
+		return NULL;
+
+	fpo_vactive_margin_us = is_fpo_vactive ? dc->debug.fpo_vactive_margin_us : 0; // For now hardcode the FPO + Vactive stretch margin to be 2000us
+	if (!is_refresh_rate_support_mclk_switch_using_fw_based_vblank_stretch(fpo_candidate_stream, fpo_vactive_margin_us))
+		return NULL;
+
+	// check if freesync enabled
+	if (!fpo_candidate_stream->allow_freesync)
+		return NULL;
+
+	if (fpo_candidate_stream->vrr_active_variable)
+		return NULL;
+
+	return fpo_candidate_stream;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c
index 79664ba7e7af..e5ab7f3077c4 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c
@@ -724,6 +724,9 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.disable_unbounded_requesting = false,
 	.override_dispclk_programming = true,
 	.disable_fpo_optimizations = false,
+	.fpo_vactive_margin_us = 2000, // 2000us
+	.disable_fpo_vactive = true,
+	.disable_boot_optimizations = false,
 };
 
 static const struct dc_debug_options debug_defaults_diags = {
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c b/drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c
index d0303173ce80..7d0626e42ea6 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c
@@ -4864,7 +4864,7 @@ void dml30_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 							v->DETBufferSizeCThisState[k],
 							&v->UrgentBurstFactorCursorPre[k],
 							&v->UrgentBurstFactorLumaPre[k],
-							&v->UrgentBurstFactorChroma[k],
+							&v->UrgentBurstFactorChromaPre[k],
 							&v->NoUrgentLatencyHidingPre[k]);
 				}
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
index 536a63624595..bd674dc30df3 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c
@@ -5191,7 +5191,7 @@ void dml31_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 							v->DETBufferSizeCThisState[k],
 							&v->UrgentBurstFactorCursorPre[k],
 							&v->UrgentBurstFactorLumaPre[k],
-							&v->UrgentBurstFactorChroma[k],
+							&v->UrgentBurstFactorChromaPre[k],
 							&v->NotUrgentLatencyHidingPre[k]);
 				}
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
index c52b76610bd2..44082f65de1f 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c
@@ -97,7 +97,7 @@ struct _vcs_dpi_ip_params_st dcn3_14_ip = {
 	.dcc_supported = true,
 };
 
-struct _vcs_dpi_soc_bounding_box_st dcn3_14_soc = {
+static struct _vcs_dpi_soc_bounding_box_st dcn3_14_soc = {
 		/*TODO: correct dispclk/dppclk voltage level determination*/
 	.clock_limits = {
 		{
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
index daf319370190..7eb2173b7691 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
@@ -5288,7 +5288,7 @@ void dml314_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_
 							v->DETBufferSizeCThisState[k],
 							&v->UrgentBurstFactorCursorPre[k],
 							&v->UrgentBurstFactorLumaPre[k],
-							&v->UrgentBurstFactorChroma[k],
+							&v->UrgentBurstFactorChromaPre[k],
 							&v->NotUrgentLatencyHidingPre[k]);
 				}
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
index 1e26adf987cc..4548320217fc 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
@@ -1927,6 +1927,7 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
 	unsigned int min_dram_speed_mts_margin;
 	bool need_fclk_lat_as_dummy = false;
 	bool is_subvp_p_drr = false;
+	struct dc_stream_state *fpo_candidate_stream = NULL;
 
 	dc_assert_fp_enabled();
 
@@ -1968,8 +1969,11 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
 	if (!pstate_en || (!dc->debug.disable_fpo_optimizations &&
 			pstate_en && vlevel != 0)) {
 		/* only when the mclk switch can not be natural, is the fw based vblank stretch attempted */
-		context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching =
-			dcn30_can_support_mclk_switch_using_fw_based_vblank_stretch(dc, context);
+		fpo_candidate_stream = dcn32_can_support_mclk_switch_using_fw_based_vblank_stretch(dc, context);
+		if (fpo_candidate_stream) {
+			fpo_candidate_stream->fpo_in_use = true;
+			context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = true;
+		}
 
 		if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) {
 			dummy_latency_index = dcn32_find_dummy_latency_index_for_fw_based_mclk_switch(dc,
@@ -2002,6 +2006,10 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
 				 * voltage level)
 				 */
 				context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = false;
+				for (i = 0; i < context->stream_count; i++) {
+					if (context->streams[i])
+						context->streams[i]->fpo_in_use = false;
+				}
 				context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.fclk_change_latency_us;
 				dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false);
 			}
@@ -2157,7 +2165,13 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
 		 * DCFCLK: Min, as reported by PM FW, when available
 		 * UCLK: Min, as reported by PM FW, when available
 		 */
-		dc->res_pool->funcs->update_soc_for_wm_a(dc, context);
+
+		/* For set A set the correct latency values (i.e. non-dummy values) unconditionally
+		 */
+		context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us;
+		context->bw_ctx.dml.soc.sr_enter_plus_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.sr_enter_plus_exit_time_us;
+		context->bw_ctx.dml.soc.sr_exit_time_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.sr_exit_time_us;
+
 		context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
 		context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
 		context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000;
@@ -2792,3 +2806,68 @@ double dcn32_determine_max_vratio_prefetch(struct dc *dc, struct dc_state *conte
 	}
 	return max_vratio_pre;
 }
+
+/**
+ * dcn32_assign_fpo_vactive_candidate - Assign the FPO stream candidate for FPO + VActive case
+ *
+ * This function chooses the FPO candidate stream for FPO + VActive cases (2 stream config).
+ * For FPO + VAtive cases, the assumption is that one display has ActiveMargin > 0, and the
+ * other display has ActiveMargin <= 0. This function will choose the pipe/stream that has
+ * ActiveMargin <= 0 to be the FPO stream candidate if found.
+ *
+ *
+ * @param [in]: dc - current dc state
+ * @param [in]: context - new dc state
+ * @param [out]: fpo_candidate_stream - pointer to FPO stream candidate if one is found
+ *
+ * Return: void
+ */
+void dcn32_assign_fpo_vactive_candidate(struct dc *dc, const struct dc_state *context, struct dc_stream_state **fpo_candidate_stream)
+{
+	unsigned int i, pipe_idx;
+	const struct vba_vars_st *vba = &context->bw_ctx.dml.vba;
+
+	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+		const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+
+		if (vba->ActiveDRAMClockChangeLatencyMarginPerState[vba->VoltageLevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]] <= 0) {
+			*fpo_candidate_stream = pipe->stream;
+			break;
+		}
+		pipe_idx++;
+	}
+}
+
+/**
+ * dcn32_find_vactive_pipe - Determines if the config has a pipe that can switch in VACTIVE
+ *
+ * @param [in]: dc - current dc state
+ * @param [in]: context - new dc state
+ * @param [in]: vactive_margin_req_us - The vactive marign required for a vactive pipe to be
+ *                                      considered "found"
+ *
+ * Return: True if VACTIVE display is found, false otherwise
+ */
+bool dcn32_find_vactive_pipe(struct dc *dc, const struct dc_state *context, uint32_t vactive_margin_req_us)
+{
+	unsigned int i, pipe_idx;
+	const struct vba_vars_st *vba = &context->bw_ctx.dml.vba;
+	bool vactive_found = false;
+
+	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
+		const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+
+		if (vba->ActiveDRAMClockChangeLatencyMarginPerState[vba->VoltageLevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]] >= vactive_margin_req_us) {
+			vactive_found = true;
+			break;
+		}
+		pipe_idx++;
+	}
+	return vactive_found;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h
index ab010e7e840b..9a0806a0e2ef 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h
@@ -76,4 +76,8 @@ void dcn32_patch_dpm_table(struct clk_bw_params *bw_params);
 void dcn32_zero_pipe_dcc_fraction(display_e2e_pipe_params_st *pipes,
 				  int pipe_cnt);
 
+void dcn32_assign_fpo_vactive_candidate(struct dc *dc, const struct dc_state *context, struct dc_stream_state **fpo_candidate_stream);
+
+bool dcn32_find_vactive_pipe(struct dc *dc, const struct dc_state *context, uint32_t vactive_margin_req);
+
 #endif
diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
index f74730c2abbd..13c7e7394b1c 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c
@@ -3354,7 +3354,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 							/* Output */
 							&mode_lib->vba.UrgentBurstFactorCursorPre[k],
 							&mode_lib->vba.UrgentBurstFactorLumaPre[k],
-							&mode_lib->vba.UrgentBurstFactorChroma[k],
+							&mode_lib->vba.UrgentBurstFactorChromaPre[k],
 							&mode_lib->vba.NotUrgentLatencyHidingPre[k]);
 				}
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_lib.c b/drivers/gpu/drm/amd/display/dc/dml/display_mode_lib.c
index 4125d3d111d1..bdf3ac6cadd5 100644
--- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_lib.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_lib.c
@@ -41,51 +41,51 @@
 #include "dcn32/display_rq_dlg_calc_32.h"
 #include "dml_logger.h"
 
-const struct dml_funcs dml20_funcs = {
+static const struct dml_funcs dml20_funcs = {
 	.validate = dml20_ModeSupportAndSystemConfigurationFull,
 	.recalculate = dml20_recalculate,
 	.rq_dlg_get_dlg_reg = dml20_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg = dml20_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml20v2_funcs = {
+static const struct dml_funcs dml20v2_funcs = {
 	.validate = dml20v2_ModeSupportAndSystemConfigurationFull,
 	.recalculate = dml20v2_recalculate,
 	.rq_dlg_get_dlg_reg = dml20v2_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg = dml20v2_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml21_funcs = {
-        .validate = dml21_ModeSupportAndSystemConfigurationFull,
-        .recalculate = dml21_recalculate,
-        .rq_dlg_get_dlg_reg = dml21_rq_dlg_get_dlg_reg,
-        .rq_dlg_get_rq_reg = dml21_rq_dlg_get_rq_reg
+static const struct dml_funcs dml21_funcs = {
+	.validate = dml21_ModeSupportAndSystemConfigurationFull,
+	.recalculate = dml21_recalculate,
+	.rq_dlg_get_dlg_reg = dml21_rq_dlg_get_dlg_reg,
+	.rq_dlg_get_rq_reg = dml21_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml30_funcs = {
+static const struct dml_funcs dml30_funcs = {
 	.validate = dml30_ModeSupportAndSystemConfigurationFull,
 	.recalculate = dml30_recalculate,
 	.rq_dlg_get_dlg_reg = dml30_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg = dml30_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml31_funcs = {
+static const struct dml_funcs dml31_funcs = {
 	.validate = dml31_ModeSupportAndSystemConfigurationFull,
 	.recalculate = dml31_recalculate,
 	.rq_dlg_get_dlg_reg = dml31_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg = dml31_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml314_funcs = {
+static const struct dml_funcs dml314_funcs = {
 	.validate = dml314_ModeSupportAndSystemConfigurationFull,
 	.recalculate = dml314_recalculate,
 	.rq_dlg_get_dlg_reg = dml314_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg = dml314_rq_dlg_get_rq_reg
 };
 
-const struct dml_funcs dml32_funcs = {
+static const struct dml_funcs dml32_funcs = {
 	.validate = dml32_ModeSupportAndSystemConfigurationFull,
-    .recalculate = dml32_recalculate,
+	.recalculate = dml32_recalculate,
 	.rq_dlg_get_dlg_reg_v2 = dml32_rq_dlg_get_dlg_reg,
 	.rq_dlg_get_rq_reg_v2 = dml32_rq_dlg_get_rq_reg
 };
diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h
index 591ab1389e3b..bef843cc32a1 100644
--- a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h
@@ -293,6 +293,9 @@ struct clk_mgr_funcs {
 
 	/* Get SMU present */
 	bool (*is_smu_present)(struct clk_mgr *clk_mgr);
+
+	int (*get_dispclk_from_dentist)(struct clk_mgr *clk_mgr_base);
+
 };
 
 struct clk_mgr {
diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
index f6c5ee2d639b..027ad1f0144d 100644
--- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
+++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
@@ -2035,6 +2035,12 @@ static enum dc_status enable_link_dp(struct dc_state *state,
 	uint32_t post_oui_delay = 30; // 30ms
 	/* Reduce link bandwidth between failed link training attempts. */
 	bool do_fallback = false;
+	int lt_attempts = LINK_TRAINING_ATTEMPTS;
+
+	// Increase retry count if attempting DP1.x on FIXED_VS link
+	if ((link->chip_caps & EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) &&
+			link_dp_get_encoding_format(link_settings) == DP_8b_10b_ENCODING)
+		lt_attempts = 10;
 
 	// check for seamless boot
 	for (i = 0; i < state->stream_count; i++) {
@@ -2099,7 +2105,7 @@ static enum dc_status enable_link_dp(struct dc_state *state,
 
 	if (perform_link_training_with_retries(link_settings,
 					       skip_video_pattern,
-					       LINK_TRAINING_ATTEMPTS,
+					       lt_attempts,
 					       pipe_ctx,
 					       pipe_ctx->stream->signal,
 					       do_fallback)) {
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c
index c840ef17802e..ba98013fecd0 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c
@@ -1043,6 +1043,9 @@ static enum dc_status wake_up_aux_channel(struct dc_link *link)
 				DP_SET_POWER,
 				&dpcd_power_state,
 				sizeof(dpcd_power_state));
+		if (status < 0)
+			DC_LOG_DC("%s: Failed to power up sink: %s\n", __func__,
+				  dpcd_power_state == DP_SET_POWER_D0 ? "D0" : "D3");
 		return DC_ERROR_UNEXPECTED;
 	}
 
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
index f301c9eaf2f9..170f33835930 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c
@@ -212,27 +212,36 @@ enum dpcd_training_patterns
 
 	switch (pattern) {
 	case DP_TRAINING_PATTERN_SEQUENCE_1:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP training pattern TPS1\n", __func__);
 		dpcd_tr_pattern = DPCD_TRAINING_PATTERN_1;
 		break;
 	case DP_TRAINING_PATTERN_SEQUENCE_2:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP training pattern TPS2\n", __func__);
 		dpcd_tr_pattern = DPCD_TRAINING_PATTERN_2;
 		break;
 	case DP_TRAINING_PATTERN_SEQUENCE_3:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP training pattern TPS3\n", __func__);
 		dpcd_tr_pattern = DPCD_TRAINING_PATTERN_3;
 		break;
 	case DP_TRAINING_PATTERN_SEQUENCE_4:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP training pattern TPS4\n", __func__);
 		dpcd_tr_pattern = DPCD_TRAINING_PATTERN_4;
 		break;
 	case DP_128b_132b_TPS1:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP 128b/132b training pattern TPS1\n", __func__);
 		dpcd_tr_pattern = DPCD_128b_132b_TPS1;
 		break;
 	case DP_128b_132b_TPS2:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP 128b/132b training pattern TPS2\n", __func__);
 		dpcd_tr_pattern = DPCD_128b_132b_TPS2;
 		break;
 	case DP_128b_132b_TPS2_CDS:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP 128b/132b training pattern TPS2 CDS\n",
+					__func__);
 		dpcd_tr_pattern = DPCD_128b_132b_TPS2_CDS;
 		break;
 	case DP_TRAINING_PATTERN_VIDEOIDLE:
+		DC_LOG_HW_LINK_TRAINING("%s: Using DP training pattern videoidle\n", __func__);
 		dpcd_tr_pattern = DPCD_TRAINING_PATTERN_VIDEOIDLE;
 		break;
 	default:
@@ -1496,7 +1505,10 @@ enum link_training_result dp_perform_link_training(
 	 * Non-LT AUX transactions inside training mode.
 	 */
 	if ((link->chip_caps & EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) && encoding == DP_8b_10b_ENCODING)
-		status = dp_perform_fixed_vs_pe_training_sequence(link, link_res, &lt_settings);
+		if (link->dc->config.use_old_fixed_vs_sequence)
+			status = dp_perform_fixed_vs_pe_training_sequence_legacy(link, link_res, &lt_settings);
+		else
+			status = dp_perform_fixed_vs_pe_training_sequence(link, link_res, &lt_settings);
 	else if (encoding == DP_8b_10b_ENCODING)
 		status = dp_perform_8b_10b_link_training(link, link_res, &lt_settings);
 	else if (encoding == DP_128b_132b_ENCODING)
@@ -1557,9 +1569,10 @@ bool perform_link_training_with_retries(
 	j = 0;
 	while (j < attempts && fail_count < (attempts * 10)) {
 
-		DC_LOG_HW_LINK_TRAINING("%s: Beginning link(%d) training attempt %u of %d @ rate(%d) x lane(%d)\n",
-			__func__, link->link_index, (unsigned int)j + 1, attempts, cur_link_settings.link_rate,
-			cur_link_settings.lane_count);
+		DC_LOG_HW_LINK_TRAINING("%s: Beginning link(%d) training attempt %u of %d @ rate(%d) x lane(%d) @ spread = %x\n",
+					__func__, link->link_index, (unsigned int)j + 1, attempts,
+				       cur_link_settings.link_rate, cur_link_settings.lane_count,
+				       cur_link_settings.link_spread);
 
 		dp_enable_link_phy(
 			link,
@@ -1637,9 +1650,10 @@ bool perform_link_training_with_retries(
 				break;
 		}
 
-		DC_LOG_WARNING("%s: Link(%d) training attempt %u of %d failed @ rate(%d) x lane(%d) : fail reason:(%d)\n",
-			__func__, link->link_index, (unsigned int)j + 1, attempts, cur_link_settings.link_rate,
-			cur_link_settings.lane_count, status);
+		DC_LOG_WARNING("%s: Link(%d) training attempt %u of %d failed @ rate(%d) x lane(%d) @ spread = %x : fail reason:(%d)\n",
+			       __func__, link->link_index, (unsigned int)j + 1, attempts,
+			      cur_link_settings.link_rate, cur_link_settings.lane_count,
+			      cur_link_settings.link_spread, status);
 
 		dp_disable_link_phy(link, &pipe_ctx->link_res, signal);
 
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c
index 14b98e096d39..3889ebb2256b 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c
@@ -225,8 +225,10 @@ enum link_training_result perform_8b_10b_clock_recovery_sequence(
 				offset);
 
 		/* 5. check CR done*/
-		if (dp_is_cr_done(lane_count, dpcd_lane_status))
+		if (dp_is_cr_done(lane_count, dpcd_lane_status)) {
+			DC_LOG_HW_LINK_TRAINING("%s: Clock recovery OK\n", __func__);
 			return LINK_TRAINING_SUCCESS;
+		}
 
 		/* 6. max VS reached*/
 		if ((link_dp_get_encoding_format(&lt_settings->link_settings) ==
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_dpia.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_dpia.c
index ab4aafdb5e5c..4f4e899e5c46 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_dpia.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_dpia.c
@@ -401,6 +401,7 @@ static enum link_training_result dpia_training_cr_non_transparent(
 
 		/* Check if clock recovery successful. */
 		if (dp_is_cr_done(lane_count, dpcd_lane_status)) {
+			DC_LOG_HW_LINK_TRAINING("%s: Clock recovery OK\n", __func__);
 			result = LINK_TRAINING_SUCCESS;
 			break;
 		}
@@ -508,6 +509,7 @@ static enum link_training_result dpia_training_cr_transparent(
 
 		/* Check if clock recovery successful. */
 		if (dp_is_cr_done(lane_count, dpcd_lane_status)) {
+			DC_LOG_HW_LINK_TRAINING("%s: Clock recovery OK\n", __func__);
 			result = LINK_TRAINING_SUCCESS;
 			break;
 		}
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
index a4071d2959a0..5731c4b61f9f 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c
@@ -223,7 +223,7 @@ static enum link_training_result perform_fixed_vs_pe_nontransparent_training_seq
 }
 
 
-enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence_legacy(
 	struct dc_link *link,
 	const struct link_resource *link_res,
 	struct link_training_settings *lt_settings)
@@ -577,3 +577,379 @@ enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
 
 	return status;
 }
+
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
+	struct dc_link *link,
+	const struct link_resource *link_res,
+	struct link_training_settings *lt_settings)
+{
+	const uint8_t vendor_lttpr_write_data_reset[4] = {0x1, 0x50, 0x63, 0xFF};
+	const uint8_t offset = dp_parse_lttpr_repeater_count(
+			link->dpcd_caps.lttpr_caps.phy_repeater_cnt);
+	const uint8_t vendor_lttpr_write_data_intercept_en[4] = {0x1, 0x55, 0x63, 0x0};
+	const uint8_t vendor_lttpr_write_data_intercept_dis[4] = {0x1, 0x55, 0x63, 0x6E};
+	const uint8_t vendor_lttpr_write_data_adicora_eq1[4] = {0x1, 0x55, 0x63, 0x2E};
+	const uint8_t vendor_lttpr_write_data_adicora_eq2[4] = {0x1, 0x55, 0x63, 0x01};
+	const uint8_t vendor_lttpr_write_data_adicora_eq3[4] = {0x1, 0x55, 0x63, 0x68};
+	uint32_t pre_disable_intercept_delay_ms = link->dc->debug.fixed_vs_aux_delay_config_wa;
+	uint8_t vendor_lttpr_write_data_vs[4] = {0x1, 0x51, 0x63, 0x0};
+	uint8_t vendor_lttpr_write_data_pe[4] = {0x1, 0x52, 0x63, 0x0};
+
+	uint32_t vendor_lttpr_write_address = 0xF004F;
+	enum link_training_result status = LINK_TRAINING_SUCCESS;
+	uint8_t lane = 0;
+	union down_spread_ctrl downspread = {0};
+	union lane_count_set lane_count_set = {0};
+	uint8_t toggle_rate;
+	uint8_t rate;
+
+	/* Only 8b/10b is supported */
+	ASSERT(link_dp_get_encoding_format(&lt_settings->link_settings) ==
+			DP_8b_10b_ENCODING);
+
+	if (lt_settings->lttpr_mode == LTTPR_MODE_NON_TRANSPARENT) {
+		status = perform_fixed_vs_pe_nontransparent_training_sequence(link, link_res, lt_settings);
+		return status;
+	}
+
+	if (offset != 0xFF) {
+		vendor_lttpr_write_address +=
+				((DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE) * (offset - 1));
+
+		/* Certain display and cable configuration require extra delay */
+		if (offset > 2)
+			pre_disable_intercept_delay_ms = link->dc->debug.fixed_vs_aux_delay_config_wa * 2;
+	}
+
+	/* Vendor specific: Reset lane settings */
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_reset[0],
+			sizeof(vendor_lttpr_write_data_reset));
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_vs[0],
+			sizeof(vendor_lttpr_write_data_vs));
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_pe[0],
+			sizeof(vendor_lttpr_write_data_pe));
+
+	/* Vendor specific: Enable intercept */
+	core_link_write_dpcd(
+			link,
+			vendor_lttpr_write_address,
+			&vendor_lttpr_write_data_intercept_en[0],
+			sizeof(vendor_lttpr_write_data_intercept_en));
+
+	/* 1. set link rate, lane count and spread. */
+
+	downspread.raw = (uint8_t)(lt_settings->link_settings.link_spread);
+
+	lane_count_set.bits.LANE_COUNT_SET =
+	lt_settings->link_settings.lane_count;
+
+	lane_count_set.bits.ENHANCED_FRAMING = lt_settings->enhanced_framing;
+	lane_count_set.bits.POST_LT_ADJ_REQ_GRANTED = 0;
+
+
+	if (lt_settings->pattern_for_eq < DP_TRAINING_PATTERN_SEQUENCE_4) {
+		lane_count_set.bits.POST_LT_ADJ_REQ_GRANTED =
+				link->dpcd_caps.max_ln_count.bits.POST_LT_ADJ_REQ_SUPPORTED;
+	}
+
+	core_link_write_dpcd(link, DP_DOWNSPREAD_CTRL,
+		&downspread.raw, sizeof(downspread));
+
+	core_link_write_dpcd(link, DP_LANE_COUNT_SET,
+		&lane_count_set.raw, 1);
+
+	rate = get_dpcd_link_rate(&lt_settings->link_settings);
+
+	/* Vendor specific: Toggle link rate */
+	toggle_rate = (rate == 0x6) ? 0xA : 0x6;
+
+	if (link->vendor_specific_lttpr_link_rate_wa == rate) {
+		core_link_write_dpcd(
+				link,
+				DP_LINK_BW_SET,
+				&toggle_rate,
+				1);
+	}
+
+	link->vendor_specific_lttpr_link_rate_wa = rate;
+
+	core_link_write_dpcd(link, DP_LINK_BW_SET, &rate, 1);
+
+	DC_LOG_HW_LINK_TRAINING("%s\n %x rate = %x\n %x lane = %x framing = %x\n %x spread = %x\n",
+		__func__,
+		DP_LINK_BW_SET,
+		lt_settings->link_settings.link_rate,
+		DP_LANE_COUNT_SET,
+		lt_settings->link_settings.lane_count,
+		lt_settings->enhanced_framing,
+		DP_DOWNSPREAD_CTRL,
+		lt_settings->link_settings.link_spread);
+
+	/* 2. Perform link training */
+
+	/* Perform Clock Recovery Sequence */
+	if (status == LINK_TRAINING_SUCCESS) {
+		const uint8_t max_vendor_dpcd_retries = 10;
+		uint32_t retries_cr;
+		uint32_t retry_count;
+		uint32_t wait_time_microsec;
+		enum dc_lane_count lane_count = lt_settings->link_settings.lane_count;
+		union lane_status dpcd_lane_status[LANE_COUNT_DP_MAX];
+		union lane_align_status_updated dpcd_lane_status_updated;
+		union lane_adjust dpcd_lane_adjust[LANE_COUNT_DP_MAX] = {0};
+		enum dc_status dpcd_status = DC_OK;
+		uint8_t i = 0;
+
+		retries_cr = 0;
+		retry_count = 0;
+
+		memset(&dpcd_lane_status, '\0', sizeof(dpcd_lane_status));
+		memset(&dpcd_lane_status_updated, '\0',
+		sizeof(dpcd_lane_status_updated));
+
+		while ((retries_cr < LINK_TRAINING_MAX_RETRY_COUNT) &&
+			(retry_count < LINK_TRAINING_MAX_CR_RETRY)) {
+
+
+			/* 1. call HWSS to set lane settings */
+			dp_set_hw_lane_settings(
+					link,
+					link_res,
+					lt_settings,
+					0);
+
+			/* 2. update DPCD of the receiver */
+			if (!retry_count) {
+				/* EPR #361076 - write as a 5-byte burst,
+				 * but only for the 1-st iteration.
+				 */
+				dpcd_set_lt_pattern_and_lane_settings(
+						link,
+						lt_settings,
+						lt_settings->pattern_for_cr,
+						0);
+				/* Vendor specific: Disable intercept */
+				for (i = 0; i < max_vendor_dpcd_retries; i++) {
+					msleep(pre_disable_intercept_delay_ms);
+					dpcd_status = core_link_write_dpcd(
+							link,
+							vendor_lttpr_write_address,
+							&vendor_lttpr_write_data_intercept_dis[0],
+							sizeof(vendor_lttpr_write_data_intercept_dis));
+
+					if (dpcd_status == DC_OK)
+						break;
+
+					core_link_write_dpcd(
+							link,
+							vendor_lttpr_write_address,
+							&vendor_lttpr_write_data_intercept_en[0],
+							sizeof(vendor_lttpr_write_data_intercept_en));
+				}
+			} else {
+				vendor_lttpr_write_data_vs[3] = 0;
+				vendor_lttpr_write_data_pe[3] = 0;
+
+				for (lane = 0; lane < lane_count; lane++) {
+					vendor_lttpr_write_data_vs[3] |=
+							lt_settings->dpcd_lane_settings[lane].bits.VOLTAGE_SWING_SET << (2 * lane);
+					vendor_lttpr_write_data_pe[3] |=
+							lt_settings->dpcd_lane_settings[lane].bits.PRE_EMPHASIS_SET << (2 * lane);
+				}
+
+				/* Vendor specific: Update VS and PE to DPRX requested value */
+				core_link_write_dpcd(
+						link,
+						vendor_lttpr_write_address,
+						&vendor_lttpr_write_data_vs[0],
+						sizeof(vendor_lttpr_write_data_vs));
+				core_link_write_dpcd(
+						link,
+						vendor_lttpr_write_address,
+						&vendor_lttpr_write_data_pe[0],
+						sizeof(vendor_lttpr_write_data_pe));
+
+				dpcd_set_lane_settings(
+						link,
+						lt_settings,
+						0);
+			}
+
+			/* 3. wait receiver to lock-on*/
+			wait_time_microsec = lt_settings->cr_pattern_time;
+
+			dp_wait_for_training_aux_rd_interval(
+					link,
+					wait_time_microsec);
+
+			/* 4. Read lane status and requested drive
+			 * settings as set by the sink
+			 */
+			dp_get_lane_status_and_lane_adjust(
+					link,
+					lt_settings,
+					dpcd_lane_status,
+					&dpcd_lane_status_updated,
+					dpcd_lane_adjust,
+					0);
+
+			/* 5. check CR done*/
+			if (dp_is_cr_done(lane_count, dpcd_lane_status)) {
+				status = LINK_TRAINING_SUCCESS;
+				break;
+			}
+
+			/* 6. max VS reached*/
+			if (dp_is_max_vs_reached(lt_settings))
+				break;
+
+			/* 7. same lane settings */
+			/* Note: settings are the same for all lanes,
+			 * so comparing first lane is sufficient
+			 */
+			if (lt_settings->dpcd_lane_settings[0].bits.VOLTAGE_SWING_SET ==
+					dpcd_lane_adjust[0].bits.VOLTAGE_SWING_LANE)
+				retries_cr++;
+			else
+				retries_cr = 0;
+
+			/* 8. update VS/PE/PC2 in lt_settings*/
+			dp_decide_lane_settings(lt_settings, dpcd_lane_adjust,
+					lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings);
+			retry_count++;
+		}
+
+		if (retry_count >= LINK_TRAINING_MAX_CR_RETRY) {
+			ASSERT(0);
+			DC_LOG_ERROR("%s: Link Training Error, could not get CR after %d tries. Possibly voltage swing issue",
+				__func__,
+				LINK_TRAINING_MAX_CR_RETRY);
+
+		}
+
+		status = dp_get_cr_failure(lane_count, dpcd_lane_status);
+	}
+
+	/* Perform Channel EQ Sequence */
+	if (status == LINK_TRAINING_SUCCESS) {
+		enum dc_dp_training_pattern tr_pattern;
+		uint32_t retries_ch_eq;
+		uint32_t wait_time_microsec;
+		enum dc_lane_count lane_count = lt_settings->link_settings.lane_count;
+		union lane_align_status_updated dpcd_lane_status_updated = {0};
+		union lane_status dpcd_lane_status[LANE_COUNT_DP_MAX] = {0};
+		union lane_adjust dpcd_lane_adjust[LANE_COUNT_DP_MAX] = {0};
+
+		core_link_write_dpcd(
+				link,
+				vendor_lttpr_write_address,
+				&vendor_lttpr_write_data_adicora_eq1[0],
+				sizeof(vendor_lttpr_write_data_adicora_eq1));
+		core_link_write_dpcd(
+				link,
+				vendor_lttpr_write_address,
+				&vendor_lttpr_write_data_adicora_eq2[0],
+				sizeof(vendor_lttpr_write_data_adicora_eq2));
+
+		/* Note: also check that TPS4 is a supported feature*/
+		tr_pattern = lt_settings->pattern_for_eq;
+
+		dp_set_hw_training_pattern(link, link_res, tr_pattern, 0);
+
+		status = LINK_TRAINING_EQ_FAIL_EQ;
+
+		for (retries_ch_eq = 0; retries_ch_eq <= LINK_TRAINING_MAX_RETRY_COUNT;
+			retries_ch_eq++) {
+
+			dp_set_hw_lane_settings(link, link_res, lt_settings, 0);
+
+			vendor_lttpr_write_data_vs[3] = 0;
+			vendor_lttpr_write_data_pe[3] = 0;
+
+			for (lane = 0; lane < lane_count; lane++) {
+				vendor_lttpr_write_data_vs[3] |=
+						lt_settings->dpcd_lane_settings[lane].bits.VOLTAGE_SWING_SET << (2 * lane);
+				vendor_lttpr_write_data_pe[3] |=
+						lt_settings->dpcd_lane_settings[lane].bits.PRE_EMPHASIS_SET << (2 * lane);
+			}
+
+			/* Vendor specific: Update VS and PE to DPRX requested value */
+			core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_vs[0],
+					sizeof(vendor_lttpr_write_data_vs));
+			core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_pe[0],
+					sizeof(vendor_lttpr_write_data_pe));
+
+			/* 2. update DPCD*/
+			if (!retries_ch_eq) {
+				/* EPR #361076 - write as a 5-byte burst,
+				 * but only for the 1-st iteration
+				 */
+
+				dpcd_set_lt_pattern_and_lane_settings(
+					link,
+					lt_settings,
+					tr_pattern, 0);
+
+				core_link_write_dpcd(
+					link,
+					vendor_lttpr_write_address,
+					&vendor_lttpr_write_data_adicora_eq3[0],
+					sizeof(vendor_lttpr_write_data_adicora_eq3));
+			} else
+				dpcd_set_lane_settings(link, lt_settings, 0);
+
+			/* 3. wait for receiver to lock-on*/
+			wait_time_microsec = lt_settings->eq_pattern_time;
+
+			dp_wait_for_training_aux_rd_interval(
+					link,
+					wait_time_microsec);
+
+			/* 4. Read lane status and requested
+			 * drive settings as set by the sink
+			 */
+			dp_get_lane_status_and_lane_adjust(
+				link,
+				lt_settings,
+				dpcd_lane_status,
+				&dpcd_lane_status_updated,
+				dpcd_lane_adjust,
+				0);
+
+			/* 5. check CR done*/
+			if (!dp_is_cr_done(lane_count, dpcd_lane_status)) {
+				status = LINK_TRAINING_EQ_FAIL_CR;
+				break;
+			}
+
+			/* 6. check CHEQ done*/
+			if (dp_is_ch_eq_done(lane_count, dpcd_lane_status) &&
+					dp_is_symbol_locked(lane_count, dpcd_lane_status) &&
+					dp_is_interlane_aligned(dpcd_lane_status_updated)) {
+				status = LINK_TRAINING_SUCCESS;
+				break;
+			}
+
+			/* 7. update VS/PE/PC2 in lt_settings*/
+			dp_decide_lane_settings(lt_settings, dpcd_lane_adjust,
+					lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings);
+		}
+	}
+
+	return status;
+}
diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
index e61970e27661..c0d6ea329504 100644
--- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
+++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.h
@@ -28,6 +28,11 @@
 #define __DC_LINK_DP_FIXED_VS_PE_RETIMER_H__
 #include "link_dp_training.h"
 
+enum link_training_result dp_perform_fixed_vs_pe_training_sequence_legacy(
+	struct dc_link *link,
+	const struct link_resource *link_res,
+	struct link_training_settings *lt_settings);
+
 enum link_training_result dp_perform_fixed_vs_pe_training_sequence(
 	struct dc_link *link,
 	const struct link_resource *link_res,
diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h
index 15d26222597a..598fa1de54ce 100644
--- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h
+++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h
@@ -362,7 +362,7 @@ union dmub_fw_boot_status {
 		uint32_t defer_load : 1; /**< 1 if VBIOS data is deferred programmed */
 		uint32_t reserved : 1;
 		uint32_t detection_required: 1; /**<  if detection need to be triggered by driver */
-
+		uint32_t hw_power_init_done: 1; /**< 1 if hw power init is completed */
 	} bits; /**< status bits */
 	uint32_t all; /**< 32-bit access to status bits */
 };
@@ -377,6 +377,7 @@ enum dmub_fw_boot_status_bit {
 	DMUB_FW_BOOT_STATUS_BIT_RESTORE_REQUIRED = (1 << 3), /**< 1 if driver should call restore */
 	DMUB_FW_BOOT_STATUS_BIT_DEFERRED_LOADED = (1 << 4), /**< 1 if VBIOS data is deferred programmed */
 	DMUB_FW_BOOT_STATUS_BIT_DETECTION_REQUIRED = (1 << 6), /**< 1 if detection need to be triggered by driver*/
+	DMUB_FW_BOOT_STATUS_BIT_HW_POWER_INIT_DONE = (1 << 7), /**< 1 if hw power init is completed */
 };
 
 /* Register bit definition for SCRATCH5 */
@@ -1104,7 +1105,12 @@ enum dmub_cmd_idle_opt_type {
 	/**
 	 * DCN hardware save.
 	 */
-	DMUB_CMD__IDLE_OPT_DCN_SAVE_INIT = 1
+	DMUB_CMD__IDLE_OPT_DCN_SAVE_INIT = 1,
+
+	/**
+	 * DCN hardware notify idle.
+	 */
+	DMUB_CMD__IDLE_OPT_DCN_NOTIFY_IDLE = 2
 };
 
 /**
@@ -1115,6 +1121,24 @@ struct dmub_rb_cmd_idle_opt_dcn_restore {
 };
 
 /**
+ * struct dmub_dcn_notify_idle_cntl_data - Data passed to FW in a DMUB_CMD__IDLE_OPT_DCN_NOTIFY_IDLE command.
+ */
+struct dmub_dcn_notify_idle_cntl_data {
+	uint8_t driver_idle;
+	uint8_t d3_entry;
+	uint8_t trigger;
+	uint8_t pad[1];
+};
+
+/**
+ * struct dmub_rb_cmd_idle_opt_dcn_notify_idle - Data passed to FW in a DMUB_CMD__IDLE_OPT_DCN_NOTIFY_IDLE command.
+ */
+struct dmub_rb_cmd_idle_opt_dcn_notify_idle {
+	struct dmub_cmd_header header; /**< header */
+	struct dmub_dcn_notify_idle_cntl_data cntl_data;
+};
+
+/**
  * struct dmub_clocks - Clock update notification.
  */
 struct dmub_clocks {
diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
index 315da61ee897..5c41a4751db4 100644
--- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
+++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
@@ -1129,7 +1129,6 @@ void mod_freesync_handle_preflip(struct mod_freesync *mod_freesync,
 {
 	struct core_freesync *core_freesync = NULL;
 	unsigned int last_render_time_in_us = 0;
-	unsigned int average_render_time_in_us = 0;
 
 	if (mod_freesync == NULL)
 		return;
@@ -1138,7 +1137,6 @@ void mod_freesync_handle_preflip(struct mod_freesync *mod_freesync,
 
 	if (in_out_vrr->supported &&
 			in_out_vrr->state == VRR_STATE_ACTIVE_VARIABLE) {
-		unsigned int i = 0;
 		unsigned int oldest_index = plane->time.index + 1;
 
 		if (oldest_index >= DC_PLANE_UPDATE_TIMES_MAX)
@@ -1147,18 +1145,6 @@ void mod_freesync_handle_preflip(struct mod_freesync *mod_freesync,
 		last_render_time_in_us = curr_time_stamp_in_us -
 				plane->time.prev_update_time_in_us;
 
-		/* Sum off all entries except oldest one */
-		for (i = 0; i < DC_PLANE_UPDATE_TIMES_MAX; i++) {
-			average_render_time_in_us +=
-					plane->time.time_elapsed_in_us[i];
-		}
-		average_render_time_in_us -=
-				plane->time.time_elapsed_in_us[oldest_index];
-
-		/* Add render time for current flip */
-		average_render_time_in_us += last_render_time_in_us;
-		average_render_time_in_us /= DC_PLANE_UPDATE_TIMES_MAX;
-
 		if (in_out_vrr->btr.btr_enabled) {
 			apply_below_the_range(core_freesync,
 					stream,
diff --git a/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_offset.h
index bd129266ebfd..a84a7cfaf71e 100644
--- a/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_offset.h
+++ b/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_offset.h
@@ -135,6 +135,8 @@
 #define mmIH_RB_WPTR_ADDR_LO_BASE_IDX                                                                  0
 #define mmIH_DOORBELL_RPTR                                                                             0x0087
 #define mmIH_DOORBELL_RPTR_BASE_IDX                                                                    0
+#define mmIH_DOORBELL_RETRY_CAM                                                                        0x0088
+#define mmIH_DOORBELL_RETRY_CAM_BASE_IDX                                                               0
 #define mmIH_RB_CNTL_RING1                                                                             0x008c
 #define mmIH_RB_CNTL_RING1_BASE_IDX                                                                    0
 #define mmIH_RB_BASE_RING1                                                                             0x008d
@@ -159,6 +161,8 @@
 #define mmIH_RB_WPTR_RING2_BASE_IDX                                                                    0
 #define mmIH_DOORBELL_RPTR_RING2                                                                       0x009f
 #define mmIH_DOORBELL_RPTR_RING2_BASE_IDX                                                              0
+#define mmIH_RETRY_CAM_ACK                                                                             0x00a4
+#define mmIH_RETRY_CAM_ACK_BASE_IDX                                                                    0
 #define mmIH_VERSION                                                                                   0x00a5
 #define mmIH_VERSION_BASE_IDX                                                                          0
 #define mmIH_CNTL                                                                                      0x00c0
@@ -235,6 +239,8 @@
 #define mmIH_MMHUB_ERROR_BASE_IDX                                                                      0
 #define mmIH_MEM_POWER_CTRL                                                                            0x00e8
 #define mmIH_MEM_POWER_CTRL_BASE_IDX                                                                   0
+#define mmIH_RETRY_INT_CAM_CNTL                                                                        0x00e9
+#define mmIH_RETRY_INT_CAM_CNTL_BASE_IDX                                                               0
 #define mmIH_REGISTER_LAST_PART2                                                                       0x00ff
 #define mmIH_REGISTER_LAST_PART2_BASE_IDX                                                              0
 #define mmSEM_CLK_CTRL                                                                                 0x0100
diff --git a/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_sh_mask.h
index 3ea83ea9ce3a..75c04fc275a0 100644
--- a/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_sh_mask.h
+++ b/drivers/gpu/drm/amd/include/asic_reg/oss/osssys_4_2_0_sh_mask.h
@@ -349,6 +349,17 @@
 #define IH_DOORBELL_RPTR_RING2__ENABLE__SHIFT                                                                 0x1c
 #define IH_DOORBELL_RPTR_RING2__OFFSET_MASK                                                                   0x03FFFFFFL
 #define IH_DOORBELL_RPTR_RING2__ENABLE_MASK                                                                   0x10000000L
+//IH_RETRY_INT_CAM_CNTL
+#define IH_RETRY_INT_CAM_CNTL__CAM_SIZE__SHIFT                                                                0x0
+#define IH_RETRY_INT_CAM_CNTL__BACK_PRESSURE_SKID_VALUE__SHIFT                                                0x8
+#define IH_RETRY_INT_CAM_CNTL__ENABLE__SHIFT                                                                  0x10
+#define IH_RETRY_INT_CAM_CNTL__BACK_PRESSURE_ENABLE__SHIFT                                                    0x11
+#define IH_RETRY_INT_CAM_CNTL__PER_VF_ENTRY_SIZE__SHIFT                                                       0x14
+#define IH_RETRY_INT_CAM_CNTL__CAM_SIZE_MASK                                                                  0x0000001FL
+#define IH_RETRY_INT_CAM_CNTL__BACK_PRESSURE_SKID_VALUE_MASK                                                  0x00003F00L
+#define IH_RETRY_INT_CAM_CNTL__ENABLE_MASK                                                                    0x00010000L
+#define IH_RETRY_INT_CAM_CNTL__BACK_PRESSURE_ENABLE_MASK                                                      0x00020000L
+#define IH_RETRY_INT_CAM_CNTL__PER_VF_ENTRY_SIZE_MASK                                                         0x00300000L
 //IH_VERSION
 #define IH_VERSION__MINVER__SHIFT                                                                             0x0
 #define IH_VERSION__MAJVER__SHIFT                                                                             0x8
diff --git a/drivers/gpu/drm/amd/include/v11_structs.h b/drivers/gpu/drm/amd/include/v11_structs.h
index b8ff7456ae0b..f8008270f813 100644
--- a/drivers/gpu/drm/amd/include/v11_structs.h
+++ b/drivers/gpu/drm/amd/include/v11_structs.h
@@ -25,14 +25,14 @@
 #define V11_STRUCTS_H_
 
 struct v11_gfx_mqd {
-	uint32_t reserved_0; // offset: 0  (0x0)
-	uint32_t reserved_1; // offset: 1  (0x1)
-	uint32_t reserved_2; // offset: 2  (0x2)
-	uint32_t reserved_3; // offset: 3  (0x3)
-	uint32_t reserved_4; // offset: 4  (0x4)
-	uint32_t reserved_5; // offset: 5  (0x5)
-	uint32_t reserved_6; // offset: 6  (0x6)
-	uint32_t reserved_7; // offset: 7  (0x7)
+	uint32_t shadow_base_lo; // offset: 0  (0x0)
+	uint32_t shadow_base_hi; // offset: 1  (0x1)
+	uint32_t gds_bkup_base_lo; // offset: 2  (0x2)
+	uint32_t gds_bkup_base_hi; // offset: 3  (0x3)
+	uint32_t fw_work_area_base_lo; // offset: 4  (0x4)
+	uint32_t fw_work_area_base_hi; // offset: 5  (0x5)
+	uint32_t shadow_initialized; // offset: 6  (0x6)
+	uint32_t ib_vmid; // offset: 7  (0x7)
 	uint32_t reserved_8; // offset: 8  (0x8)
 	uint32_t reserved_9; // offset: 9  (0x9)
 	uint32_t reserved_10; // offset: 10  (0xA)
diff --git a/drivers/gpu/drm/amd/pm/amdgpu_pm.c b/drivers/gpu/drm/amd/pm/amdgpu_pm.c
index e011041e3ec6..58c2246918fd 100644
--- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c
@@ -3395,7 +3395,7 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
 	      attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't manage state */
 		effective_mode &= ~S_IWUSR;
 
-	/* not implemented yet for GC 10.3.1 APUs */
+	/* In the case of APUs, this is only implemented on Vangogh */
 	if (((adev->family == AMDGPU_FAMILY_SI) ||
 	     ((adev->flags & AMD_IS_APU) && (gc_ver != IP_VERSION(10, 3, 1)))) &&
 	    (attr == &sensor_dev_attr_power1_cap_max.dev_attr.attr ||
@@ -3404,7 +3404,7 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
 	     attr == &sensor_dev_attr_power1_cap_default.dev_attr.attr))
 		return 0;
 
-	/* not implemented yet for APUs having <= GC 9.3.0 */
+	/* not implemented yet for APUs having < GC 9.3.0 (Renoir) */
 	if (((adev->family == AMDGPU_FAMILY_SI) ||
 	     ((adev->flags & AMD_IS_APU) && (gc_ver < IP_VERSION(9, 3, 0)))) &&
 	    (attr == &sensor_dev_attr_power1_average.dev_attr.attr))
diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
index 056ac2b512eb..5633c5797e85 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
@@ -1712,8 +1712,6 @@ static int smu_display_configuration_change(void *handle,
 					    const struct amd_pp_display_configuration *display_config)
 {
 	struct smu_context *smu = handle;
-	int index = 0;
-	int num_of_active_display = 0;
 
 	if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled)
 		return -EOPNOTSUPP;
@@ -1724,11 +1722,6 @@ static int smu_display_configuration_change(void *handle,
 	smu_set_min_dcef_deep_sleep(smu,
 				    display_config->min_dcef_deep_sleep_set_clk / 100);
 
-	for (index = 0; index < display_config->num_path_including_non_display; index++) {
-		if (display_config->displays[index].controller_id != 0)
-			num_of_active_display++;
-	}
-
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
index 0ef37837b164..7944ce80e5c3 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
+++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
@@ -62,6 +62,12 @@
 #define CTF_OFFSET_HOTSPOT		5
 #define CTF_OFFSET_MEM			5
 
+static const int pmfw_decoded_link_speed[5] = {1, 2, 3, 4, 5};
+static const int pmfw_decoded_link_width[7] = {0, 1, 2, 4, 8, 12, 16};
+
+#define DECODE_GEN_SPEED(gen_speed_idx)		(pmfw_decoded_link_speed[gen_speed_idx])
+#define DECODE_LANE_WIDTH(lane_width_idx)	(pmfw_decoded_link_width[lane_width_idx])
+
 struct smu_13_0_max_sustainable_clocks {
 	uint32_t display_clock;
 	uint32_t phy_clock;
diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
index e9766fe5656e..09405ef1e3c8 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
@@ -1144,8 +1144,8 @@ static int smu_v13_0_0_print_clk_levels(struct smu_context *smu,
 					(pcie_table->pcie_lane[i] == 5) ? "x12" :
 					(pcie_table->pcie_lane[i] == 6) ? "x16" : "",
 					pcie_table->clk_freq[i],
-					((gen_speed - 1) == pcie_table->pcie_gen[i]) &&
-					(lane_width == link_width[pcie_table->pcie_lane[i]]) ?
+					(gen_speed == DECODE_GEN_SPEED(pcie_table->pcie_gen[i])) &&
+					(lane_width == DECODE_LANE_WIDTH(link_width[pcie_table->pcie_lane[i]])) ?
 					"*" : "");
 		break;
 
diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c
index 1b2c82449f20..3d9ff46706fb 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c
@@ -575,6 +575,14 @@ static int smu_v13_0_7_set_default_dpm_table(struct smu_context *smu)
 						     dpm_table);
 		if (ret)
 			return ret;
+
+		if (skutable->DriverReportedClocks.GameClockAc &&
+			(dpm_table->dpm_levels[dpm_table->count - 1].value >
+			skutable->DriverReportedClocks.GameClockAc)) {
+			dpm_table->dpm_levels[dpm_table->count - 1].value =
+				skutable->DriverReportedClocks.GameClockAc;
+			dpm_table->max = skutable->DriverReportedClocks.GameClockAc;
+		}
 	} else {
 		dpm_table->count = 1;
 		dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
@@ -828,6 +836,57 @@ static int smu_v13_0_7_get_smu_metrics_data(struct smu_context *smu,
 	return ret;
 }
 
+static int smu_v13_0_7_get_dpm_ultimate_freq(struct smu_context *smu,
+					     enum smu_clk_type clk_type,
+					     uint32_t *min,
+					     uint32_t *max)
+{
+	struct smu_13_0_dpm_context *dpm_context =
+		smu->smu_dpm.dpm_context;
+	struct smu_13_0_dpm_table *dpm_table;
+
+	switch (clk_type) {
+	case SMU_MCLK:
+	case SMU_UCLK:
+		/* uclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.uclk_table;
+		break;
+	case SMU_GFXCLK:
+	case SMU_SCLK:
+		/* gfxclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.gfx_table;
+		break;
+	case SMU_SOCCLK:
+		/* socclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.soc_table;
+		break;
+	case SMU_FCLK:
+		/* fclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.fclk_table;
+		break;
+	case SMU_VCLK:
+	case SMU_VCLK1:
+		/* vclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.vclk_table;
+		break;
+	case SMU_DCLK:
+	case SMU_DCLK1:
+		/* dclk dpm table */
+		dpm_table = &dpm_context->dpm_tables.dclk_table;
+		break;
+	default:
+		dev_err(smu->adev->dev, "Unsupported clock type!\n");
+		return -EINVAL;
+	}
+
+	if (min)
+		*min = dpm_table->min;
+	if (max)
+		*max = dpm_table->max;
+
+	return 0;
+}
+
 static int smu_v13_0_7_read_sensor(struct smu_context *smu,
 				   enum amd_pp_sensors sensor,
 				   void *data,
@@ -1074,8 +1133,8 @@ static int smu_v13_0_7_print_clk_levels(struct smu_context *smu,
 					(pcie_table->pcie_lane[i] == 5) ? "x12" :
 					(pcie_table->pcie_lane[i] == 6) ? "x16" : "",
 					pcie_table->clk_freq[i],
-					(gen_speed == pcie_table->pcie_gen[i]) &&
-					(lane_width == pcie_table->pcie_lane[i]) ?
+					(gen_speed == DECODE_GEN_SPEED(pcie_table->pcie_gen[i])) &&
+					(lane_width == DECODE_LANE_WIDTH(pcie_table->pcie_lane[i])) ?
 					"*" : "");
 		break;
 
@@ -1329,9 +1388,17 @@ static int smu_v13_0_7_populate_umd_state_clk(struct smu_context *smu)
 				&dpm_context->dpm_tables.fclk_table;
 	struct smu_umd_pstate_table *pstate_table =
 				&smu->pstate_table;
+	struct smu_table_context *table_context = &smu->smu_table;
+	PPTable_t *pptable = table_context->driver_pptable;
+	DriverReportedClocks_t driver_clocks =
+		pptable->SkuTable.DriverReportedClocks;
 
 	pstate_table->gfxclk_pstate.min = gfx_table->min;
-	pstate_table->gfxclk_pstate.peak = gfx_table->max;
+	if (driver_clocks.GameClockAc &&
+		(driver_clocks.GameClockAc < gfx_table->max))
+		pstate_table->gfxclk_pstate.peak = driver_clocks.GameClockAc;
+	else
+		pstate_table->gfxclk_pstate.peak = gfx_table->max;
 
 	pstate_table->uclk_pstate.min = mem_table->min;
 	pstate_table->uclk_pstate.peak = mem_table->max;
@@ -1348,12 +1415,12 @@ static int smu_v13_0_7_populate_umd_state_clk(struct smu_context *smu)
 	pstate_table->fclk_pstate.min = fclk_table->min;
 	pstate_table->fclk_pstate.peak = fclk_table->max;
 
-	/*
-	 * For now, just use the mininum clock frequency.
-	 * TODO: update them when the real pstate settings available
-	 */
-	pstate_table->gfxclk_pstate.standard = gfx_table->min;
-	pstate_table->uclk_pstate.standard = mem_table->min;
+	if (driver_clocks.BaseClockAc &&
+		driver_clocks.BaseClockAc < gfx_table->max)
+		pstate_table->gfxclk_pstate.standard = driver_clocks.BaseClockAc;
+	else
+		pstate_table->gfxclk_pstate.standard = gfx_table->max;
+	pstate_table->uclk_pstate.standard = mem_table->max;
 	pstate_table->socclk_pstate.standard = soc_table->min;
 	pstate_table->vclk_pstate.standard = vclk_table->min;
 	pstate_table->dclk_pstate.standard = dclk_table->min;
@@ -1678,7 +1745,7 @@ static const struct pptable_funcs smu_v13_0_7_ppt_funcs = {
 	.dpm_set_jpeg_enable = smu_v13_0_set_jpeg_enable,
 	.init_pptable_microcode = smu_v13_0_init_pptable_microcode,
 	.populate_umd_state_clk = smu_v13_0_7_populate_umd_state_clk,
-	.get_dpm_ultimate_freq = smu_v13_0_get_dpm_ultimate_freq,
+	.get_dpm_ultimate_freq = smu_v13_0_7_get_dpm_ultimate_freq,
 	.get_vbios_bootup_values = smu_v13_0_get_vbios_bootup_values,
 	.read_sensor = smu_v13_0_7_read_sensor,
 	.feature_is_enabled = smu_cmn_feature_is_enabled,
author	Dave Airlie <airlied@redhat.com>	2023-04-17 10:54:58 +1000
committer	Dave Airlie <airlied@redhat.com>	2023-04-17 10:54:59 +1000
commit	e82c98f2ca439356d5595ba8c9cd782f993f6f8c (patch)
tree	e07718a52c12f78e247c83f2d6ba14a0058fd819 /drivers/gpu
parent	afa351a15d80993f8ba6ae28652cc23127237e37 (diff)
parent	541372bb62f289f4402cf55be51fb9cec7373627 (diff)
download	lwn-e82c98f2ca439356d5595ba8c9cd782f993f6f8c.tar.gz lwn-e82c98f2ca439356d5595ba8c9cd782f993f6f8c.zip