Merge tag 'amd-drm-next-5.6-2020-01-09' of git://people.freedesktop.org/~agd5f/linux into drm-next

amd-drm-next-5.6-2020-01-09:

amdgpu:
- Enable DCN support on POWER
- Enable GFXOFF for Raven1 refresh
- Clean up MM engine idle handlers
- HDMI 2.0 audio fixes
- Fixes for some 10 bpc EDP panels
- Watermark fixes for renoir
- SR-IOV fixes
- Runtime pm robustness fixes
- Arcturus VCN fixes
- RAS fixes
- BACO fixes for Arcturus
- Stable pstate fixes for swSMU
- HDCP fixes
- PSP cleanup
- HDMI fixes
- Misc cleanups

amdkfd:
- Spread interrupt work across cores to reduce latency
- Topology fixes for APUs
- GPU reset improvements

UAPI:
- Enable DRIVER_SYNCOBJ_TIMELINE for vulkan
- Return better error values for kfd process ioctl

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200109230338.8022-1-alexander.deucher@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -636,9 +636,8 @@ struct amdgpu_fw_vram_usage {
 	struct amdgpu_bo *reserved_bo;
 	void *va;
 
-	/* Offset on the top of VRAM, used as c2p write buffer.
+	/* GDDR6 training support flag.
 	*/
-	u64 mem_train_fb_loc;
 	bool mem_train_support;
 };
 
@@ -994,8 +993,6 @@ struct amdgpu_device {
 
 	bool                            pm_sysfs_en;
 	bool                            ucode_sysfs_en;
-
-	bool				in_baco;
 };
 
 static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c

@@ -46,6 +46,8 @@
 #include "soc15.h"
 #include "soc15d.h"
 #include "amdgpu_amdkfd_gfx_v9.h"
+#include "gfxhub_v1_0.h"
+#include "mmhub_v9_4.h"
 
 #define HQD_N_REGS 56
 #define DUMP_REG(addr) do {				\
@@ -258,6 +260,22 @@ static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
 	return 0;
 }
 
+static void kgd_set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint64_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID %u\n",
+		       vmid);
+		return;
+	}
+
+	mmhub_v9_4_setup_vm_pt_regs(adev, vmid, page_table_base);
+
+	gfxhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base);
+}
+
 const struct kfd2kgd_calls arcturus_kfd2kgd = {
 	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
@@ -277,7 +295,7 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = {
 	.get_atc_vmid_pasid_mapping_info =
 			kgd_gfx_v9_get_atc_vmid_pasid_mapping_info,
 	.get_tile_config = kgd_gfx_v9_get_tile_config,
-	.set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base,
+	.set_vm_context_page_table_base = kgd_set_vm_context_page_table_base,
 	.invalidate_tlbs = kgd_gfx_v9_invalidate_tlbs,
 	.invalidate_tlbs_vmid = kgd_gfx_v9_invalidate_tlbs_vmid,
 	.get_hive_id = amdgpu_amdkfd_get_hive_id,

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c

@@ -40,7 +40,6 @@
 #include "soc15d.h"
 #include "mmhub_v1_0.h"
 #include "gfxhub_v1_0.h"
-#include "mmhub_v9_4.h"
 
 
 enum hqd_dequeue_request_type {
@@ -758,8 +757,8 @@ uint32_t kgd_gfx_v9_address_watch_get_offset(struct kgd_dev *kgd,
 	return 0;
 }
 
-void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+static void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd,
-		uint64_t page_table_base)
+			uint32_t vmid, uint64_t page_table_base)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 
@@ -769,14 +768,7 @@ void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmi
 		return;
 	}
 
-	/* TODO: take advantage of per-process address space size. For
+	mmhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base);
-	 * now, all processes share the same address space size, like
-	 * on GFX8 and older.
-	 */
-	if (adev->asic_type == CHIP_ARCTURUS) {
-		mmhub_v9_4_setup_vm_pt_regs(adev, vmid, page_table_base);
-	} else
-		mmhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base);
 
 	gfxhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base);
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h

@@ -57,8 +57,6 @@ uint32_t kgd_gfx_v9_address_watch_get_offset(struct kgd_dev *kgd,
 
 bool kgd_gfx_v9_get_atc_vmid_pasid_mapping_info(struct kgd_dev *kgd,
 					uint8_t vmid, uint16_t *p_pasid);
-void kgd_gfx_v9_set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
-		uint64_t page_table_base);
 int kgd_gfx_v9_invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
 int kgd_gfx_v9_invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
 int kgd_gfx_v9_get_tile_config(struct kgd_dev *kgd,

drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c

@@ -2022,7 +2022,7 @@ int amdgpu_atombios_init(struct amdgpu_device *adev)
 	if (adev->is_atom_fw) {
 		amdgpu_atomfirmware_scratch_regs_init(adev);
 		amdgpu_atomfirmware_allocate_fb_scratch(adev);
-		ret = amdgpu_atomfirmware_get_mem_train_fb_loc(adev);
+		ret = amdgpu_atomfirmware_get_mem_train_info(adev);
 		if (ret) {
 			DRM_ERROR("Failed to get mem train fb location.\n");
 			return ret;

drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c

@@ -525,16 +525,12 @@ static int gddr6_mem_train_support(struct amdgpu_device *adev)
 	return ret;
 }
 
-int amdgpu_atomfirmware_get_mem_train_fb_loc(struct amdgpu_device *adev)
+int amdgpu_atomfirmware_get_mem_train_info(struct amdgpu_device *adev)
 {
 	struct atom_context *ctx = adev->mode_info.atom_context;
-	unsigned char *bios = ctx->bios;
+	int index;
-	struct vram_reserve_block *reserved_block;
-	int index, block_number;
 	uint8_t frev, crev;
 	uint16_t data_offset, size;
-	uint32_t start_address_in_kb;
-	uint64_t offset;
 	int ret;
 
 	adev->fw_vram_usage.mem_train_support = false;
@@ -569,32 +565,6 @@ int amdgpu_atomfirmware_get_mem_train_fb_loc(struct amdgpu_device *adev)
 		return -EINVAL;
 	}
 
-	reserved_block = (struct vram_reserve_block *)
+	adev->fw_vram_usage.mem_train_support = true;
-		(bios + data_offset + sizeof(struct atom_common_table_header));
+	return 0;
-	block_number = ((unsigned int)size - sizeof(struct atom_common_table_header))
-		/ sizeof(struct vram_reserve_block);
-	reserved_block += (block_number > 0) ? block_number-1 : 0;
-	DRM_DEBUG("block_number:0x%04x, last block: 0x%08xkb sz, %dkb fw, %dkb drv.\n",
-		  block_number,
-		  le32_to_cpu(reserved_block->start_address_in_kb),
-		  le16_to_cpu(reserved_block->used_by_firmware_in_kb),
-		  le16_to_cpu(reserved_block->used_by_driver_in_kb));
-	if (reserved_block->used_by_firmware_in_kb > 0) {
-		start_address_in_kb = le32_to_cpu(reserved_block->start_address_in_kb);
-		offset = (uint64_t)start_address_in_kb * ONE_KiB;
-		if ((offset & (ONE_MiB - 1)) < (4 * ONE_KiB + 1) ) {
-			offset -= ONE_MiB;
-		}
-
-		offset &= ~(ONE_MiB - 1);
-		adev->fw_vram_usage.mem_train_fb_loc = offset;
-		adev->fw_vram_usage.mem_train_support = true;
-		DRM_DEBUG("mem_train_fb_loc:0x%09llx.\n", offset);
-		ret = 0;
-	} else {
-		DRM_ERROR("used_by_firmware_in_kb is 0!\n");
-		ret = -EINVAL;
-	}
-
-	return ret;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h

@@ -31,7 +31,7 @@ void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_get_vram_info(struct amdgpu_device *adev,
 	int *vram_width, int *vram_type, int *vram_vendor);
-int amdgpu_atomfirmware_get_mem_train_fb_loc(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_mem_train_info(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev);
 bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev);

drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c

@@ -613,7 +613,17 @@ static bool amdgpu_atpx_detect(void)
 	bool d3_supported = false;
 	struct pci_dev *parent_pdev;
 
-	while ((pdev = pci_get_class(PCI_BASE_CLASS_DISPLAY << 16, pdev)) != NULL) {
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
+		vga_count++;
+
+		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+
+		parent_pdev = pci_upstream_bridge(pdev);
+		d3_supported |= parent_pdev && parent_pdev->bridge_d3;
+		amdgpu_atpx_get_quirks(pdev);
+	}
+
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
 		vga_count++;
 
 		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);

drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c

@@ -74,7 +74,7 @@ static int amdgpu_ctx_init(struct amdgpu_device *adev,
 			   struct amdgpu_ctx *ctx)
 {
 	unsigned num_entities = amdgpu_ctx_total_num_entities();
-	unsigned i, j, k;
+	unsigned i, j;
 	int r;
 
 	if (priority < 0 || priority >= DRM_SCHED_PRIORITY_MAX)
@@ -121,72 +121,57 @@ static int amdgpu_ctx_init(struct amdgpu_device *adev,
 	ctx->override_priority = DRM_SCHED_PRIORITY_UNSET;
 
 	for (i = 0; i < AMDGPU_HW_IP_NUM; ++i) {
-		struct amdgpu_ring *rings[AMDGPU_MAX_RINGS];
+		struct drm_gpu_scheduler **scheds;
-		struct drm_sched_rq *rqs[AMDGPU_MAX_RINGS];
+		struct drm_gpu_scheduler *sched;
-		unsigned num_rings = 0;
+		unsigned num_scheds = 0;
-		unsigned num_rqs = 0;
 
 		switch (i) {
 		case AMDGPU_HW_IP_GFX:
-			rings[0] = &adev->gfx.gfx_ring[0];
+			sched = &adev->gfx.gfx_ring[0].sched;
-			num_rings = 1;
+			scheds = &sched;
+			num_scheds = 1;
 			break;
 		case AMDGPU_HW_IP_COMPUTE:
-			for (j = 0; j < adev->gfx.num_compute_rings; ++j)
+			scheds = adev->gfx.compute_sched;
-				rings[j] = &adev->gfx.compute_ring[j];
+			num_scheds = adev->gfx.num_compute_sched;
-			num_rings = adev->gfx.num_compute_rings;
 			break;
 		case AMDGPU_HW_IP_DMA:
-			for (j = 0; j < adev->sdma.num_instances; ++j)
+			scheds = adev->sdma.sdma_sched;
-				rings[j] = &adev->sdma.instance[j].ring;
+			num_scheds = adev->sdma.num_sdma_sched;
-			num_rings = adev->sdma.num_instances;
 			break;
 		case AMDGPU_HW_IP_UVD:
-			rings[0] = &adev->uvd.inst[0].ring;
+			sched = &adev->uvd.inst[0].ring.sched;
-			num_rings = 1;
+			scheds = &sched;
+			num_scheds = 1;
 			break;
 		case AMDGPU_HW_IP_VCE:
-			rings[0] = &adev->vce.ring[0];
+			sched = &adev->vce.ring[0].sched;
-			num_rings = 1;
+			scheds = &sched;
+			num_scheds = 1;
 			break;
 		case AMDGPU_HW_IP_UVD_ENC:
-			rings[0] = &adev->uvd.inst[0].ring_enc[0];
+			sched = &adev->uvd.inst[0].ring_enc[0].sched;
-			num_rings = 1;
+			scheds = &sched;
+			num_scheds = 1;
 			break;
 		case AMDGPU_HW_IP_VCN_DEC:
-			for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
+			scheds = adev->vcn.vcn_dec_sched;
-				if (adev->vcn.harvest_config & (1 << j))
+			num_scheds =  adev->vcn.num_vcn_dec_sched;
-					continue;
-				rings[num_rings++] = &adev->vcn.inst[j].ring_dec;
-			}
 			break;
 		case AMDGPU_HW_IP_VCN_ENC:
-			for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
+			scheds = adev->vcn.vcn_enc_sched;
-				if (adev->vcn.harvest_config & (1 << j))
+			num_scheds =  adev->vcn.num_vcn_enc_sched;
-					continue;
-				for (k = 0; k < adev->vcn.num_enc_rings; ++k)
-					rings[num_rings++] = &adev->vcn.inst[j].ring_enc[k];
-			}
 			break;
 		case AMDGPU_HW_IP_VCN_JPEG:
-			for (j = 0; j < adev->jpeg.num_jpeg_inst; ++j) {
+			scheds = adev->jpeg.jpeg_sched;
-				if (adev->jpeg.harvest_config & (1 << j))
+			num_scheds =  adev->jpeg.num_jpeg_sched;
-					continue;
-				rings[num_rings++] = &adev->jpeg.inst[j].ring_dec;
-			}
 			break;
 		}
 
-		for (j = 0; j < num_rings; ++j) {
-			if (!rings[j]->adev)
-				continue;
-
-			rqs[num_rqs++] = &rings[j]->sched.sched_rq[priority];
-		}
-
 		for (j = 0; j < amdgpu_ctx_num_entities[i]; ++j)
 			r = drm_sched_entity_init(&ctx->entities[i][j].entity,
-						  rqs, num_rqs, &ctx->guilty);
+						  priority, scheds,
+						  num_scheds, &ctx->guilty);
 		if (r)
 			goto error_cleanup_entities;
 	}
@@ -627,3 +612,45 @@ void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
 	idr_destroy(&mgr->ctx_handles);
 	mutex_destroy(&mgr->lock);
 }
+
+void amdgpu_ctx_init_sched(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		adev->gfx.gfx_sched[i] = &adev->gfx.gfx_ring[i].sched;
+		adev->gfx.num_gfx_sched++;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		adev->gfx.compute_sched[i] = &adev->gfx.compute_ring[i].sched;
+		adev->gfx.num_compute_sched++;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.sdma_sched[i] = &adev->sdma.instance[i].ring.sched;
+		adev->sdma.num_sdma_sched++;
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+		adev->vcn.vcn_dec_sched[adev->vcn.num_vcn_dec_sched++] =
+			&adev->vcn.inst[i].ring_dec.sched;
+	}
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+		for (j = 0; j < adev->vcn.num_enc_rings; ++j)
+			adev->vcn.vcn_enc_sched[adev->vcn.num_vcn_enc_sched++] =
+				&adev->vcn.inst[i].ring_enc[j].sched;
+	}
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+		adev->jpeg.jpeg_sched[adev->jpeg.num_jpeg_sched++] =
+			&adev->jpeg.inst[i].ring_dec.sched;
+	}
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h

@@ -87,4 +87,7 @@ void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr);
 long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout);
 void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
 
+void amdgpu_ctx_init_sched(struct amdgpu_device *adev);
+
+
 #endif

drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c

@@ -129,7 +129,7 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
 			sh_bank = 0xFFFFFFFF;
 		if (instance_bank == 0x3FF)
 			instance_bank = 0xFFFFFFFF;
-		use_bank = 1;
+		use_bank = true;
 	} else if (*pos & (1ULL << 61)) {
 
 		me = (*pos & GENMASK_ULL(33, 24)) >> 24;
@@ -137,9 +137,9 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
 		queue = (*pos & GENMASK_ULL(53, 44)) >> 44;
 		vmid = (*pos & GENMASK_ULL(58, 54)) >> 54;
 
-		use_ring = 1;
+		use_ring = true;
 	} else {
-		use_bank = use_ring = 0;
+		use_bank = use_ring = false;
 	}
 
 	*pos &= (1UL << 22) - 1;

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -66,6 +66,7 @@
 #include "amdgpu_pmu.h"
 
 #include <linux/suspend.h>
+#include <drm/task_barrier.h>
 
 MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
@@ -1031,8 +1032,6 @@ def_value:
  */
 static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 {
-	int ret = 0;
-
 	if (amdgpu_sched_jobs < 4) {
 		dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n",
 			 amdgpu_sched_jobs);
@@ -1072,7 +1071,7 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 
 	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
 
-	return ret;
+	return 0;
 }
 
 /**
@@ -1810,7 +1809,8 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev)
 		}
 	}
 
-	r = amdgpu_pm_load_smu_firmware(adev, &smu_version);
+	if (!amdgpu_sriov_vf(adev) || adev->asic_type == CHIP_TONGA)
+		r = amdgpu_pm_load_smu_firmware(adev, &smu_version);
 
 	return r;
 }
@@ -2439,7 +2439,8 @@ static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
 		AMD_IP_BLOCK_TYPE_GFX,
 		AMD_IP_BLOCK_TYPE_SDMA,
 		AMD_IP_BLOCK_TYPE_UVD,
-		AMD_IP_BLOCK_TYPE_VCE
+		AMD_IP_BLOCK_TYPE_VCE,
+		AMD_IP_BLOCK_TYPE_VCN
 	};
 
 	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
@@ -2454,7 +2455,11 @@ static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
 				block->status.hw)
 				continue;
 
-			r = block->version->funcs->hw_init(adev);
+			if (block->version->type == AMD_IP_BLOCK_TYPE_SMC)
+				r = block->version->funcs->resume(adev);
+			else
+				r = block->version->funcs->hw_init(adev);
+
 			DRM_INFO("RE-INIT-late: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
 			if (r)
 				return r;
@@ -2663,14 +2668,38 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
 {
 	struct amdgpu_device *adev =
 		container_of(__work, struct amdgpu_device, xgmi_reset_work);
+	struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev, 0);
 
-	if (amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO)
+	/* It's a bug to not have a hive within this function */
-		adev->asic_reset_res = (adev->in_baco == false) ?
+	if (WARN_ON(!hive))
-				amdgpu_device_baco_enter(adev->ddev) :
+		return;
-				amdgpu_device_baco_exit(adev->ddev);
-	else
-		adev->asic_reset_res = amdgpu_asic_reset(adev);
 
+	/*
+	 * Use task barrier to synchronize all xgmi reset works across the
+	 * hive. task_barrier_enter and task_barrier_exit will block
+	 * until all the threads running the xgmi reset works reach
+	 * those points. task_barrier_full will do both blocks.
+	 */
+	if (amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) {
+
+		task_barrier_enter(&hive->tb);
+		adev->asic_reset_res = amdgpu_device_baco_enter(adev->ddev);
+
+		if (adev->asic_reset_res)
+			goto fail;
+
+		task_barrier_exit(&hive->tb);
+		adev->asic_reset_res = amdgpu_device_baco_exit(adev->ddev);
+
+		if (adev->asic_reset_res)
+			goto fail;
+	} else {
+
+		task_barrier_full(&hive->tb);
+		adev->asic_reset_res =  amdgpu_asic_reset(adev);
+	}
+
+fail:
 	if (adev->asic_reset_res)
 		DRM_WARN("ASIC reset failed with error, %d for drm dev, %s",
 			 adev->asic_reset_res, adev->ddev->unique);
@@ -2785,7 +2814,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->mman.buffer_funcs = NULL;
 	adev->mman.buffer_funcs_ring = NULL;
 	adev->vm_manager.vm_pte_funcs = NULL;
-	adev->vm_manager.vm_pte_num_rqs = 0;
+	adev->vm_manager.vm_pte_num_scheds = 0;
 	adev->gmc.gmc_funcs = NULL;
 	adev->fence_context = dma_fence_context_alloc(AMDGPU_MAX_RINGS);
 	bitmap_zero(adev->gfx.pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
@@ -3029,6 +3058,14 @@ fence_driver_init:
 		goto failed;
 	}
 
+	DRM_DEBUG("SE %d, SH per SE %d, CU per SH %d, active_cu_number %d\n",
+			adev->gfx.config.max_shader_engines,
+			adev->gfx.config.max_sh_per_se,
+			adev->gfx.config.max_cu_per_sh,
+			adev->gfx.cu_info.number);
+
+	amdgpu_ctx_init_sched(adev);
+
 	adev->accel_working = true;
 
 	amdgpu_vm_check_compute_bug(adev);
@@ -3660,8 +3697,6 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
 	if (r)
 		return r;
 
-	amdgpu_amdkfd_pre_reset(adev);
-
 	/* Resume IP prior to SMC */
 	r = amdgpu_device_ip_reinit_early_sriov(adev);
 	if (r)
@@ -3790,18 +3825,13 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 	return r;
 }
 
-static int amdgpu_do_asic_reset(struct amdgpu_device *adev,
+static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
-			       struct amdgpu_hive_info *hive,
 			       struct list_head *device_list_handle,
 			       bool *need_full_reset_arg)
 {
 	struct amdgpu_device *tmp_adev = NULL;
 	bool need_full_reset = *need_full_reset_arg, vram_lost = false;
 	int r = 0;
-	int cpu = smp_processor_id();
-	bool use_baco =
-		(amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) ?
-		true : false;
 
 	/*
 	 * ASIC reset has to be done on all HGMI hive nodes ASAP
@@ -3809,24 +3839,21 @@ static int amdgpu_do_asic_reset(struct amdgpu_device *adev,
 	 */
 	if (need_full_reset) {
 		list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
-			/*
+			/* For XGMI run all resets in parallel to speed up the process */
-			 * For XGMI run all resets in parallel to speed up the
-			 * process by scheduling the highpri wq on different
-			 * cpus. For XGMI with baco reset, all nodes must enter
-			 * baco within close proximity before anyone exit.
-			 */
 			if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
-				if (!queue_work_on(cpu, system_highpri_wq,
+				if (!queue_work(system_unbound_wq, &tmp_adev->xgmi_reset_work))
-						   &tmp_adev->xgmi_reset_work))
 					r = -EALREADY;
-				cpu = cpumask_next(cpu, cpu_online_mask);
 			} else
 				r = amdgpu_asic_reset(tmp_adev);
-			if (r)
+
+			if (r) {
+				DRM_ERROR("ASIC reset failed with error, %d for drm dev, %s",
+					 r, tmp_adev->ddev->unique);
 				break;
+			}
 		}
 
-		/* For XGMI wait for all work to complete before proceed */
+		/* For XGMI wait for all resets to complete before proceed */
 		if (!r) {
 			list_for_each_entry(tmp_adev, device_list_handle,
 					    gmc.xgmi.head) {
@@ -3835,53 +3862,9 @@ static int amdgpu_do_asic_reset(struct amdgpu_device *adev,
 					r = tmp_adev->asic_reset_res;
 					if (r)
 						break;
-					if (use_baco)
-						tmp_adev->in_baco = true;
 				}
 			}
 		}
-
-		/*
-		 * For XGMI with baco reset, need exit baco phase by scheduling
-		 * xgmi_reset_work one more time. PSP reset and sGPU skips this
-		 * phase. Not assume the situation that PSP reset and baco reset
-		 * coexist within an XGMI hive.
-		 */
-
-		if (!r && use_baco) {
-			cpu = smp_processor_id();
-			list_for_each_entry(tmp_adev, device_list_handle,
-					    gmc.xgmi.head) {
-				if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
-					if (!queue_work_on(cpu,
-						system_highpri_wq,
-						&tmp_adev->xgmi_reset_work))
-						r = -EALREADY;
-					if (r)
-						break;
-					cpu = cpumask_next(cpu, cpu_online_mask);
-				}
-			}
-		}
-
-		if (!r && use_baco) {
-			list_for_each_entry(tmp_adev, device_list_handle,
-					    gmc.xgmi.head) {
-				if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
-					flush_work(&tmp_adev->xgmi_reset_work);
-					r = tmp_adev->asic_reset_res;
-					if (r)
-						break;
-					tmp_adev->in_baco = false;
-				}
-			}
-		}
-
-		if (r) {
-			DRM_ERROR("ASIC reset failed with error, %d for drm dev, %s",
-				 r, tmp_adev->ddev->unique);
-			goto end;
-		}
 	}
 
 	if (!r && amdgpu_ras_intr_triggered())
@@ -3974,7 +3957,7 @@ static bool amdgpu_device_lock_adev(struct amdgpu_device *adev, bool trylock)
 		mutex_lock(&adev->lock_reset);
 
 	atomic_inc(&adev->gpu_reset_counter);
-	adev->in_gpu_reset = 1;
+	adev->in_gpu_reset = true;
 	switch (amdgpu_asic_reset_method(adev)) {
 	case AMD_RESET_METHOD_MODE1:
 		adev->mp1_state = PP_MP1_STATE_SHUTDOWN;
@@ -3994,7 +3977,7 @@ static void amdgpu_device_unlock_adev(struct amdgpu_device *adev)
 {
 	amdgpu_vf_error_trans_all(adev);
 	adev->mp1_state = PP_MP1_STATE_NONE;
-	adev->in_gpu_reset = 0;
+	adev->in_gpu_reset = false;
 	mutex_unlock(&adev->lock_reset);
 }
 
@@ -4175,8 +4158,7 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 		if (r)
 			adev->asic_reset_res = r;
 	} else {
-		r  = amdgpu_do_asic_reset(adev, hive, device_list_handle,
+		r  = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset);
-					  &need_full_reset);
 		if (r && r == -EAGAIN)
 			goto retry;
 	}

drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c

@@ -951,16 +951,31 @@ int amdgpu_dpm_set_powergating_by_smu(struct amdgpu_device *adev, uint32_t block
 	case AMD_IP_BLOCK_TYPE_VCN:
 	case AMD_IP_BLOCK_TYPE_VCE:
 	case AMD_IP_BLOCK_TYPE_SDMA:
+		if (swsmu) {
+			ret = smu_dpm_set_power_gate(&adev->smu, block_type, gate);
+		} else {
+			if (adev->powerplay.pp_funcs &&
+			    adev->powerplay.pp_funcs->set_powergating_by_smu) {
+				mutex_lock(&adev->pm.mutex);
+				ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
+					(adev)->powerplay.pp_handle, block_type, gate));
+				mutex_unlock(&adev->pm.mutex);
+			}
+		}
+		break;
+	case AMD_IP_BLOCK_TYPE_JPEG:
 		if (swsmu)
 			ret = smu_dpm_set_power_gate(&adev->smu, block_type, gate);
-		else
-			ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
-				(adev)->powerplay.pp_handle, block_type, gate));
 		break;
 	case AMD_IP_BLOCK_TYPE_GMC:
 	case AMD_IP_BLOCK_TYPE_ACP:
-		ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
+		if (adev->powerplay.pp_funcs &&
+		    adev->powerplay.pp_funcs->set_powergating_by_smu) {
+			mutex_lock(&adev->pm.mutex);
+			ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
 				(adev)->powerplay.pp_handle, block_type, gate));
+			mutex_unlock(&adev->pm.mutex);
+		}
 		break;
 	default:
 		break;

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -142,7 +142,7 @@ int amdgpu_async_gfx_ring = 1;
 int amdgpu_mcbp = 0;
 int amdgpu_discovery = -1;
 int amdgpu_mes = 0;
-int amdgpu_noretry = 1;
+int amdgpu_noretry;
 int amdgpu_force_asic_type = -1;
 
 struct amdgpu_mgpu_info mgpu_info = {
@@ -588,7 +588,7 @@ MODULE_PARM_DESC(mes,
 module_param_named(mes, amdgpu_mes, int, 0444);
 
 MODULE_PARM_DESC(noretry,
-	"Disable retry faults (0 = retry enabled, 1 = retry disabled (default))");
+	"Disable retry faults (0 = retry enabled (default), 1 = retry disabled)");
 module_param_named(noretry, amdgpu_noretry, int, 0644);
 
 /**
@@ -1203,13 +1203,23 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
 	struct amdgpu_device *adev = drm_dev->dev_private;
-	int ret;
+	int ret, i;
 
 	if (!adev->runpm) {
 		pm_runtime_forbid(dev);
 		return -EBUSY;
 	}
 
+	/* wait for all rings to drain before suspending */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (ring && ring->sched.ready) {
+			ret = amdgpu_fence_wait_empty(ring);
+			if (ret)
+				return -EBUSY;
+		}
+	}
+
 	if (amdgpu_device_supports_boco(drm_dev))
 		drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 	drm_kms_helper_poll_disable(drm_dev);
@@ -1381,7 +1391,8 @@ static struct drm_driver kms_driver = {
 	.driver_features =
 	    DRIVER_USE_AGP | DRIVER_ATOMIC |
 	    DRIVER_GEM |
-	    DRIVER_RENDER | DRIVER_MODESET | DRIVER_SYNCOBJ,
+	    DRIVER_RENDER | DRIVER_MODESET | DRIVER_SYNCOBJ |
+	    DRIVER_SYNCOBJ_TIMELINE,
 	.load = amdgpu_driver_load_kms,
 	.open = amdgpu_driver_open_kms,
 	.postclose = amdgpu_driver_postclose_kms,

drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c

@@ -34,6 +34,7 @@
 #include <linux/kref.h>
 #include <linux/slab.h>
 #include <linux/firmware.h>
+#include <linux/pm_runtime.h>
 
 #include <drm/drm_debugfs.h>
 
@@ -154,7 +155,7 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
 		       seq);
 	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
 			       seq, flags | AMDGPU_FENCE_FLAG_INT);
-
+	pm_runtime_get_noresume(adev->ddev->dev);
 	ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask];
 	if (unlikely(rcu_dereference_protected(*ptr, 1))) {
 		struct dma_fence *old;
@@ -234,6 +235,7 @@ static void amdgpu_fence_schedule_fallback(struct amdgpu_ring *ring)
 bool amdgpu_fence_process(struct amdgpu_ring *ring)
 {
 	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	struct amdgpu_device *adev = ring->adev;
 	uint32_t seq, last_seq;
 	int r;
 
@@ -274,6 +276,8 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring)
 			BUG();
 
 		dma_fence_put(fence);
+		pm_runtime_mark_last_busy(adev->ddev->dev);
+		pm_runtime_put_autosuspend(adev->ddev->dev);
 	} while (last_seq != seq);
 
 	return true;

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c

@@ -641,7 +641,7 @@ int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev,
 		kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
 		if (adev->gfx.funcs->query_ras_error_count)
 			adev->gfx.funcs->query_ras_error_count(adev, err_data);
-		amdgpu_ras_reset_gpu(adev, 0);
+		amdgpu_ras_reset_gpu(adev);
 	}
 	return AMDGPU_RAS_SUCCESS;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h

@@ -269,8 +269,12 @@ struct amdgpu_gfx {
 	bool				me_fw_write_wait;
 	bool				cp_fw_write_wait;
 	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
+	struct drm_gpu_scheduler	*gfx_sched[AMDGPU_MAX_GFX_RINGS];
+	uint32_t			num_gfx_sched;
 	unsigned			num_gfx_rings;
 	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];
+	struct drm_gpu_scheduler	*compute_sched[AMDGPU_MAX_COMPUTE_RINGS];
+	uint32_t			num_compute_sched;
 	unsigned			num_compute_rings;
 	struct amdgpu_irq_src		eop_irq;
 	struct amdgpu_irq_src		priv_reg_irq;

drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c

@@ -223,7 +223,7 @@ void amdgpu_gmc_agp_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc)
 	u64 size_af, size_bf;
 
 	if (amdgpu_sriov_vf(adev)) {
-		mc->agp_start = 0xffffffff;
+		mc->agp_start = 0xffffffffffff;
 		mc->agp_end = 0x0;
 		mc->agp_size = 0;
 
@@ -333,3 +333,43 @@ void amdgpu_gmc_ras_fini(struct amdgpu_device *adev)
 	amdgpu_mmhub_ras_fini(adev);
 	amdgpu_xgmi_ras_fini(adev);
 }
+
+	/*
+	 * The latest engine allocation on gfx9/10 is:
+	 * Engine 2, 3: firmware
+	 * Engine 0, 1, 4~16: amdgpu ring,
+	 *                    subject to change when ring number changes
+	 * Engine 17: Gart flushes
+	 */
+#define GFXHUB_FREE_VM_INV_ENGS_BITMAP		0x1FFF3
+#define MMHUB_FREE_VM_INV_ENGS_BITMAP		0x1FFF3
+
+int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	unsigned vm_inv_engs[AMDGPU_MAX_VMHUBS] =
+		{GFXHUB_FREE_VM_INV_ENGS_BITMAP, MMHUB_FREE_VM_INV_ENGS_BITMAP,
+		GFXHUB_FREE_VM_INV_ENGS_BITMAP};
+	unsigned i;
+	unsigned vmhub, inv_eng;
+
+	for (i = 0; i < adev->num_rings; ++i) {
+		ring = adev->rings[i];
+		vmhub = ring->funcs->vmhub;
+
+		inv_eng = ffs(vm_inv_engs[vmhub]);
+		if (!inv_eng) {
+			dev_err(adev->dev, "no VM inv eng for ring %s\n",
+				ring->name);
+			return -EINVAL;
+		}
+
+		ring->vm_inv_eng = inv_eng - 1;
+		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);
+	}
+
+	return 0;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h

@@ -267,5 +267,6 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
 			      uint16_t pasid, uint64_t timestamp);
 int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev);
 void amdgpu_gmc_ras_fini(struct amdgpu_device *adev);
+int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev);
 
 #endif

drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h

@@ -43,6 +43,8 @@ struct amdgpu_jpeg {
 	uint8_t	num_jpeg_inst;
 	struct amdgpu_jpeg_inst inst[AMDGPU_MAX_JPEG_INSTANCES];
 	struct amdgpu_jpeg_reg internal;
+	struct drm_gpu_scheduler *jpeg_sched[AMDGPU_MAX_JPEG_INSTANCES];
+	uint32_t num_jpeg_sched;
 	unsigned harvest_config;
 	struct delayed_work idle_work;
 	enum amd_powergating_state cur_state;

drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c

@@ -2762,17 +2762,12 @@ static void amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev)
 void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
 {
 	int ret = 0;
-	if (is_support_sw_smu(adev)) {
+
-	    ret = smu_dpm_set_power_gate(&adev->smu, AMD_IP_BLOCK_TYPE_UVD, enable);
+	ret = amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_UVD, !enable);
-	    if (ret)
+	if (ret)
-		DRM_ERROR("[SW SMU]: dpm enable uvd failed, state = %s, ret = %d. \n",
+		DRM_ERROR("Dpm %s uvd failed, ret = %d. \n",
-			  enable ? "true" : "false", ret);
+			  enable ? "enable" : "disable", ret);
-	} else if (adev->powerplay.pp_funcs->set_powergating_by_smu) {
+
-		/* enable/disable UVD */
-		mutex_lock(&adev->pm.mutex);
-		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_UVD, !enable);
-		mutex_unlock(&adev->pm.mutex);
-	}
 	/* enable/disable Low Memory PState for UVD (4k videos) */
 	if (adev->asic_type == CHIP_STONEY &&
 		adev->uvd.decode_image_width >= WIDTH_4K) {
@@ -2789,17 +2784,11 @@ void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
 void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable)
 {
 	int ret = 0;
-	if (is_support_sw_smu(adev)) {
+
-	    ret = smu_dpm_set_power_gate(&adev->smu, AMD_IP_BLOCK_TYPE_VCE, enable);
+	ret = amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_VCE, !enable);
-	    if (ret)
+	if (ret)
-		DRM_ERROR("[SW SMU]: dpm enable vce failed, state = %s, ret = %d. \n",
+		DRM_ERROR("Dpm %s vce failed, ret = %d. \n",
-			  enable ? "true" : "false", ret);
+			  enable ? "enable" : "disable", ret);
-	} else if (adev->powerplay.pp_funcs->set_powergating_by_smu) {
-		/* enable/disable VCE */
-		mutex_lock(&adev->pm.mutex);
-		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_VCE, !enable);
-		mutex_unlock(&adev->pm.mutex);
-	}
 }
 
 void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
@@ -2818,12 +2807,10 @@ void amdgpu_dpm_enable_jpeg(struct amdgpu_device *adev, bool enable)
 {
 	int ret = 0;
 
-	if (is_support_sw_smu(adev)) {
+	ret = amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_JPEG, !enable);
-		ret = smu_dpm_set_power_gate(&adev->smu, AMD_IP_BLOCK_TYPE_JPEG, enable);
+	if (ret)
-		if (ret)
+		DRM_ERROR("Dpm %s jpeg failed, ret = %d. \n",
-			DRM_ERROR("[SW SMU]: dpm enable jpeg failed, state = %s, ret = %d. \n",
+			  enable ? "enable" : "disable", ret);
-				  enable ? "true" : "false", ret);
-	}
 }
 
 int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_version)

drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c

@@ -107,7 +107,7 @@ static void amdgpu_perf_read(struct perf_event *event)
 		default:
 			count = 0;
 			break;
-		};
+		}
 	} while (local64_cmpxchg(&hwc->prev_count, prev, count) != prev);
 
 	local64_add(count - prev, &event->count);
@@ -130,7 +130,7 @@ static void amdgpu_perf_stop(struct perf_event *event, int flags)
 		break;
 	default:
 		break;
-	};
+	}
 
 	WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
 	hwc->state |= PERF_HES_STOPPED;
@@ -160,7 +160,7 @@ static int amdgpu_perf_add(struct perf_event *event, int flags)
 		break;
 	default:
 		return 0;
-	};
+	}
 
 	if (retval)
 		return retval;
@@ -188,7 +188,7 @@ static void amdgpu_perf_del(struct perf_event *event, int flags)
 		break;
 	default:
 		break;
-	};
+	}
 
 	perf_event_update_userpage(event);
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c

@@ -191,9 +191,9 @@ psp_cmd_submit_buf(struct psp_context *psp,
 		if (ucode)
 			DRM_WARN("failed to load ucode id (%d) ",
 				  ucode->ucode_id);
-		DRM_DEBUG_DRIVER("psp command (0x%X) failed and response status is (0x%X)\n",
+		DRM_WARN("psp command (0x%X) failed and response status is (0x%X)\n",
 			 psp->cmd_buf_mem->cmd_id,
-			 psp->cmd_buf_mem->resp.status & GFX_CMD_STATUS_MASK);
+			 psp->cmd_buf_mem->resp.status);
 		if (!timeout) {
 			mutex_unlock(&psp->mutex);
 			return -EINVAL;
@@ -365,11 +365,11 @@ static int psp_asd_load(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_asd_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+static void psp_prep_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					uint32_t asd_session_id)
+				       uint32_t session_id)
 {
 	cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA;
-	cmd->cmd.cmd_unload_ta.session_id = asd_session_id;
+	cmd->cmd.cmd_unload_ta.session_id = session_id;
 }
 
 static int psp_asd_unload(struct psp_context *psp)
@@ -387,7 +387,7 @@ static int psp_asd_unload(struct psp_context *psp)
 	if (!cmd)
 		return -ENOMEM;
 
-	psp_prep_asd_unload_cmd_buf(cmd, psp->asd_context.session_id);
+	psp_prep_ta_unload_cmd_buf(cmd, psp->asd_context.session_id);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 				 psp->fence_buf_mc_addr);
@@ -427,18 +427,20 @@ int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg,
 	return ret;
 }
 
-static void psp_prep_xgmi_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+static void psp_prep_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					  uint64_t xgmi_ta_mc, uint64_t xgmi_mc_shared,
+				     uint64_t ta_bin_mc,
-					  uint32_t xgmi_ta_size, uint32_t shared_size)
+				     uint32_t ta_bin_size,
+				     uint64_t ta_shared_mc,
+				     uint32_t ta_shared_size)
 {
-        cmd->cmd_id = GFX_CMD_ID_LOAD_TA;
+	cmd->cmd_id 				= GFX_CMD_ID_LOAD_TA;
-        cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(xgmi_ta_mc);
+	cmd->cmd.cmd_load_ta.app_phy_addr_lo 	= lower_32_bits(ta_bin_mc);
-        cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(xgmi_ta_mc);
+	cmd->cmd.cmd_load_ta.app_phy_addr_hi 	= upper_32_bits(ta_bin_mc);
-        cmd->cmd.cmd_load_ta.app_len = xgmi_ta_size;
+	cmd->cmd.cmd_load_ta.app_len 		= ta_bin_size;
 
-        cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(xgmi_mc_shared);
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(ta_shared_mc);
-        cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(xgmi_mc_shared);
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(ta_shared_mc);
-        cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
+	cmd->cmd.cmd_load_ta.cmd_buf_len 	 = ta_shared_size;
 }
 
 static int psp_xgmi_init_shared_buf(struct psp_context *psp)
@@ -458,6 +460,36 @@ static int psp_xgmi_init_shared_buf(struct psp_context *psp)
 	return ret;
 }
 
+static void psp_prep_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				       uint32_t ta_cmd_id,
+				       uint32_t session_id)
+{
+	cmd->cmd_id 				= GFX_CMD_ID_INVOKE_CMD;
+	cmd->cmd.cmd_invoke_cmd.session_id 	= session_id;
+	cmd->cmd.cmd_invoke_cmd.ta_cmd_id 	= ta_cmd_id;
+}
+
+int psp_ta_invoke(struct psp_context *psp,
+		  uint32_t ta_cmd_id,
+		  uint32_t session_id)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	if (!cmd)
+		return -ENOMEM;
+
+	psp_prep_ta_invoke_cmd_buf(cmd, ta_cmd_id, session_id);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+
+	kfree(cmd);
+
+	return ret;
+}
+
 static int psp_xgmi_load(struct psp_context *psp)
 {
 	int ret;
@@ -466,8 +498,6 @@ static int psp_xgmi_load(struct psp_context *psp)
 	/*
 	 * TODO: bypass the loading in sriov for now
 	 */
-	if (amdgpu_sriov_vf(psp->adev))
-		return 0;
 
 	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
 	if (!cmd)
@@ -476,9 +506,11 @@ static int psp_xgmi_load(struct psp_context *psp)
 	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
 	memcpy(psp->fw_pri_buf, psp->ta_xgmi_start_addr, psp->ta_xgmi_ucode_size);
 
-	psp_prep_xgmi_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr,
+	psp_prep_ta_load_cmd_buf(cmd,
-				      psp->xgmi_context.xgmi_shared_mc_addr,
+				 psp->fw_pri_mc_addr,
-				      psp->ta_xgmi_ucode_size, PSP_XGMI_SHARED_MEM_SIZE);
+				 psp->ta_xgmi_ucode_size,
+				 psp->xgmi_context.xgmi_shared_mc_addr,
+				 PSP_XGMI_SHARED_MEM_SIZE);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 				 psp->fence_buf_mc_addr);
@@ -493,13 +525,6 @@ static int psp_xgmi_load(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_xgmi_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					    uint32_t xgmi_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA;
-	cmd->cmd.cmd_unload_ta.session_id = xgmi_session_id;
-}
-
 static int psp_xgmi_unload(struct psp_context *psp)
 {
 	int ret;
@@ -508,14 +533,12 @@ static int psp_xgmi_unload(struct psp_context *psp)
 	/*
 	 * TODO: bypass the unloading in sriov for now
 	 */
-	if (amdgpu_sriov_vf(psp->adev))
-		return 0;
 
 	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
 	if (!cmd)
 		return -ENOMEM;
 
-	psp_prep_xgmi_ta_unload_cmd_buf(cmd, psp->xgmi_context.session_id);
+	psp_prep_ta_unload_cmd_buf(cmd, psp->xgmi_context.session_id);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 				 psp->fence_buf_mc_addr);
@@ -525,40 +548,9 @@ static int psp_xgmi_unload(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_xgmi_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					    uint32_t ta_cmd_id,
-					    uint32_t xgmi_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD;
-	cmd->cmd.cmd_invoke_cmd.session_id = xgmi_session_id;
-	cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id;
-	/* Note: cmd_invoke_cmd.buf is not used for now */
-}
-
 int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
 {
-	int ret;
+	return psp_ta_invoke(psp, ta_cmd_id, psp->xgmi_context.session_id);
-	struct psp_gfx_cmd_resp *cmd;
-
-	/*
-	 * TODO: bypass the loading in sriov for now
-	*/
-	if (amdgpu_sriov_vf(psp->adev))
-		return 0;
-
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
-	if (!cmd)
-		return -ENOMEM;
-
-	psp_prep_xgmi_ta_invoke_cmd_buf(cmd, ta_cmd_id,
-					psp->xgmi_context.session_id);
-
-	ret = psp_cmd_submit_buf(psp, NULL, cmd,
-				 psp->fence_buf_mc_addr);
-
-	kfree(cmd);
-
-        return ret;
 }
 
 static int psp_xgmi_terminate(struct psp_context *psp)
@@ -614,20 +606,6 @@ static int psp_xgmi_initialize(struct psp_context *psp)
 }
 
 // ras begin
-static void psp_prep_ras_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-		uint64_t ras_ta_mc, uint64_t ras_mc_shared,
-		uint32_t ras_ta_size, uint32_t shared_size)
-{
-	cmd->cmd_id = GFX_CMD_ID_LOAD_TA;
-	cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(ras_ta_mc);
-	cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(ras_ta_mc);
-	cmd->cmd.cmd_load_ta.app_len = ras_ta_size;
-
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(ras_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(ras_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
-}
-
 static int psp_ras_init_shared_buf(struct psp_context *psp)
 {
 	int ret;
@@ -663,15 +641,17 @@ static int psp_ras_load(struct psp_context *psp)
 	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
 	memcpy(psp->fw_pri_buf, psp->ta_ras_start_addr, psp->ta_ras_ucode_size);
 
-	psp_prep_ras_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr,
+	psp_prep_ta_load_cmd_buf(cmd,
-			psp->ras.ras_shared_mc_addr,
+				 psp->fw_pri_mc_addr,
-			psp->ta_ras_ucode_size, PSP_RAS_SHARED_MEM_SIZE);
+				 psp->ta_ras_ucode_size,
+				 psp->ras.ras_shared_mc_addr,
+				 PSP_RAS_SHARED_MEM_SIZE);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 			psp->fence_buf_mc_addr);
 
 	if (!ret) {
-		psp->ras.ras_initialized = 1;
+		psp->ras.ras_initialized = true;
 		psp->ras.session_id = cmd->resp.session_id;
 	}
 
@@ -680,13 +660,6 @@ static int psp_ras_load(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_ras_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-						uint32_t ras_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA;
-	cmd->cmd.cmd_unload_ta.session_id = ras_session_id;
-}
-
 static int psp_ras_unload(struct psp_context *psp)
 {
 	int ret;
@@ -702,7 +675,7 @@ static int psp_ras_unload(struct psp_context *psp)
 	if (!cmd)
 		return -ENOMEM;
 
-	psp_prep_ras_ta_unload_cmd_buf(cmd, psp->ras.session_id);
+	psp_prep_ta_unload_cmd_buf(cmd, psp->ras.session_id);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 			psp->fence_buf_mc_addr);
@@ -712,40 +685,15 @@ static int psp_ras_unload(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_ras_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-		uint32_t ta_cmd_id,
-		uint32_t ras_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD;
-	cmd->cmd.cmd_invoke_cmd.session_id = ras_session_id;
-	cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id;
-	/* Note: cmd_invoke_cmd.buf is not used for now */
-}
-
 int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
 {
-	int ret;
-	struct psp_gfx_cmd_resp *cmd;
-
 	/*
 	 * TODO: bypass the loading in sriov for now
 	 */
 	if (amdgpu_sriov_vf(psp->adev))
 		return 0;
 
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	return psp_ta_invoke(psp, ta_cmd_id, psp->ras.session_id);
-	if (!cmd)
-		return -ENOMEM;
-
-	psp_prep_ras_ta_invoke_cmd_buf(cmd, ta_cmd_id,
-			psp->ras.session_id);
-
-	ret = psp_cmd_submit_buf(psp, NULL, cmd,
-			psp->fence_buf_mc_addr);
-
-	kfree(cmd);
-
-	return ret;
 }
 
 int psp_ras_enable_features(struct psp_context *psp,
@@ -791,7 +739,7 @@ static int psp_ras_terminate(struct psp_context *psp)
 	if (ret)
 		return ret;
 
-	psp->ras.ras_initialized = 0;
+	psp->ras.ras_initialized = false;
 
 	/* free ras shared memory */
 	amdgpu_bo_free_kernel(&psp->ras.ras_shared_bo,
@@ -832,24 +780,6 @@ static int psp_ras_initialize(struct psp_context *psp)
 // ras end
 
 // HDCP start
-static void psp_prep_hdcp_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					  uint64_t hdcp_ta_mc,
-					  uint64_t hdcp_mc_shared,
-					  uint32_t hdcp_ta_size,
-					  uint32_t shared_size)
-{
-	cmd->cmd_id = GFX_CMD_ID_LOAD_TA;
-	cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(hdcp_ta_mc);
-	cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(hdcp_ta_mc);
-	cmd->cmd.cmd_load_ta.app_len = hdcp_ta_size;
-
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo =
-		lower_32_bits(hdcp_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi =
-		upper_32_bits(hdcp_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
-}
-
 static int psp_hdcp_init_shared_buf(struct psp_context *psp)
 {
 	int ret;
@@ -886,15 +816,16 @@ static int psp_hdcp_load(struct psp_context *psp)
 	memcpy(psp->fw_pri_buf, psp->ta_hdcp_start_addr,
 	       psp->ta_hdcp_ucode_size);
 
-	psp_prep_hdcp_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr,
+	psp_prep_ta_load_cmd_buf(cmd,
-				      psp->hdcp_context.hdcp_shared_mc_addr,
+				 psp->fw_pri_mc_addr,
-				      psp->ta_hdcp_ucode_size,
+				 psp->ta_hdcp_ucode_size,
-				      PSP_HDCP_SHARED_MEM_SIZE);
+				 psp->hdcp_context.hdcp_shared_mc_addr,
+				 PSP_HDCP_SHARED_MEM_SIZE);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
 
 	if (!ret) {
-		psp->hdcp_context.hdcp_initialized = 1;
+		psp->hdcp_context.hdcp_initialized = true;
 		psp->hdcp_context.session_id = cmd->resp.session_id;
 	}
 
@@ -930,12 +861,6 @@ static int psp_hdcp_initialize(struct psp_context *psp)
 
 	return 0;
 }
-static void psp_prep_hdcp_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					    uint32_t hdcp_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA;
-	cmd->cmd.cmd_unload_ta.session_id = hdcp_session_id;
-}
 
 static int psp_hdcp_unload(struct psp_context *psp)
 {
@@ -952,7 +877,7 @@ static int psp_hdcp_unload(struct psp_context *psp)
 	if (!cmd)
 		return -ENOMEM;
 
-	psp_prep_hdcp_ta_unload_cmd_buf(cmd, psp->hdcp_context.session_id);
+	psp_prep_ta_unload_cmd_buf(cmd, psp->hdcp_context.session_id);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
 
@@ -961,39 +886,15 @@ static int psp_hdcp_unload(struct psp_context *psp)
 	return ret;
 }
 
-static void psp_prep_hdcp_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					    uint32_t ta_cmd_id,
-					    uint32_t hdcp_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD;
-	cmd->cmd.cmd_invoke_cmd.session_id = hdcp_session_id;
-	cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id;
-	/* Note: cmd_invoke_cmd.buf is not used for now */
-}
-
 int psp_hdcp_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
 {
-	int ret;
-	struct psp_gfx_cmd_resp *cmd;
-
 	/*
 	 * TODO: bypass the loading in sriov for now
 	 */
 	if (amdgpu_sriov_vf(psp->adev))
 		return 0;
 
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	return psp_ta_invoke(psp, ta_cmd_id, psp->hdcp_context.session_id);
-	if (!cmd)
-		return -ENOMEM;
-
-	psp_prep_hdcp_ta_invoke_cmd_buf(cmd, ta_cmd_id,
-					psp->hdcp_context.session_id);
-
-	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
-
-	kfree(cmd);
-
-	return ret;
 }
 
 static int psp_hdcp_terminate(struct psp_context *psp)
@@ -1013,7 +914,7 @@ static int psp_hdcp_terminate(struct psp_context *psp)
 	if (ret)
 		return ret;
 
-	psp->hdcp_context.hdcp_initialized = 0;
+	psp->hdcp_context.hdcp_initialized = false;
 
 	/* free hdcp shared memory */
 	amdgpu_bo_free_kernel(&psp->hdcp_context.hdcp_shared_bo,
@@ -1025,22 +926,6 @@ static int psp_hdcp_terminate(struct psp_context *psp)
 // HDCP end
 
 // DTM start
-static void psp_prep_dtm_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					 uint64_t dtm_ta_mc,
-					 uint64_t dtm_mc_shared,
-					 uint32_t dtm_ta_size,
-					 uint32_t shared_size)
-{
-	cmd->cmd_id = GFX_CMD_ID_LOAD_TA;
-	cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(dtm_ta_mc);
-	cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(dtm_ta_mc);
-	cmd->cmd.cmd_load_ta.app_len = dtm_ta_size;
-
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(dtm_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(dtm_mc_shared);
-	cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
-}
-
 static int psp_dtm_init_shared_buf(struct psp_context *psp)
 {
 	int ret;
@@ -1076,15 +961,16 @@ static int psp_dtm_load(struct psp_context *psp)
 	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
 	memcpy(psp->fw_pri_buf, psp->ta_dtm_start_addr, psp->ta_dtm_ucode_size);
 
-	psp_prep_dtm_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr,
+	psp_prep_ta_load_cmd_buf(cmd,
-				     psp->dtm_context.dtm_shared_mc_addr,
+				 psp->fw_pri_mc_addr,
-				     psp->ta_dtm_ucode_size,
+				 psp->ta_dtm_ucode_size,
-				     PSP_DTM_SHARED_MEM_SIZE);
+				 psp->dtm_context.dtm_shared_mc_addr,
+				 PSP_DTM_SHARED_MEM_SIZE);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
 
 	if (!ret) {
-		psp->dtm_context.dtm_initialized = 1;
+		psp->dtm_context.dtm_initialized = true;
 		psp->dtm_context.session_id = cmd->resp.session_id;
 	}
 
@@ -1122,39 +1008,15 @@ static int psp_dtm_initialize(struct psp_context *psp)
 	return 0;
 }
 
-static void psp_prep_dtm_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd,
-					   uint32_t ta_cmd_id,
-					   uint32_t dtm_session_id)
-{
-	cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD;
-	cmd->cmd.cmd_invoke_cmd.session_id = dtm_session_id;
-	cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id;
-	/* Note: cmd_invoke_cmd.buf is not used for now */
-}
-
 int psp_dtm_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
 {
-	int ret;
-	struct psp_gfx_cmd_resp *cmd;
-
 	/*
 	 * TODO: bypass the loading in sriov for now
 	 */
 	if (amdgpu_sriov_vf(psp->adev))
 		return 0;
 
-	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	return psp_ta_invoke(psp, ta_cmd_id, psp->dtm_context.session_id);
-	if (!cmd)
-		return -ENOMEM;
-
-	psp_prep_dtm_ta_invoke_cmd_buf(cmd, ta_cmd_id,
-				       psp->dtm_context.session_id);
-
-	ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
-
-	kfree(cmd);
-
-	return ret;
 }
 
 static int psp_dtm_terminate(struct psp_context *psp)
@@ -1174,7 +1036,7 @@ static int psp_dtm_terminate(struct psp_context *psp)
 	if (ret)
 		return ret;
 
-	psp->dtm_context.dtm_initialized = 0;
+	psp->dtm_context.dtm_initialized = false;
 
 	/* free hdcp shared memory */
 	amdgpu_bo_free_kernel(&psp->dtm_context.dtm_shared_bo,
@@ -1310,6 +1172,9 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
 	case AMDGPU_UCODE_ID_VCN:
 		*type = GFX_FW_TYPE_VCN;
 		break;
+	case AMDGPU_UCODE_ID_VCN1:
+		*type = GFX_FW_TYPE_VCN1;
+		break;
 	case AMDGPU_UCODE_ID_DMCU_ERAM:
 		*type = GFX_FW_TYPE_DMCU_ERAM;
 		break;
@@ -1454,7 +1319,8 @@ out:
                     || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_G
 	            || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL
 	            || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM
-	            || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM))
+	            || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM
+	            || ucode->ucode_id == AMDGPU_UCODE_ID_SMC))
 			/*skip ucode loading in SRIOV VF */
 			continue;
 
@@ -1472,7 +1338,7 @@ out:
 
 		/* Start rlc autoload after psp recieved all the gfx firmware */
 		if (psp->autoload_supported && ucode->ucode_id == (amdgpu_sriov_vf(adev) ?
-		    AMDGPU_UCODE_ID_CP_MEC2 : AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM)) {
+		    AMDGPU_UCODE_ID_CP_MEC2 : AMDGPU_UCODE_ID_RLC_G)) {
 			ret = psp_rlc_autoload(psp);
 			if (ret) {
 				DRM_ERROR("Failed to start rlc autoload\n");
@@ -1503,16 +1369,13 @@ static int psp_load_fw(struct amdgpu_device *adev)
 	if (!psp->cmd)
 		return -ENOMEM;
 
-	/* this fw pri bo is not used under SRIOV */
+	ret = amdgpu_bo_create_kernel(adev, PSP_1_MEG, PSP_1_MEG,
-	if (!amdgpu_sriov_vf(psp->adev)) {
+					AMDGPU_GEM_DOMAIN_GTT,
-		ret = amdgpu_bo_create_kernel(adev, PSP_1_MEG, PSP_1_MEG,
+					&psp->fw_pri_bo,
-					      AMDGPU_GEM_DOMAIN_GTT,
+					&psp->fw_pri_mc_addr,
-					      &psp->fw_pri_bo,
+					&psp->fw_pri_buf);
-					      &psp->fw_pri_mc_addr,
+	if (ret)
-					      &psp->fw_pri_buf);
+		goto failed;
-		if (ret)
-			goto failed;
-	}
 
 	ret = amdgpu_bo_create_kernel(adev, PSP_FENCE_BUFFER_SIZE, PAGE_SIZE,
 					AMDGPU_GEM_DOMAIN_VRAM,

drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h

@@ -202,7 +202,6 @@ struct psp_memory_training_context {
 
 	/*vram offset of the p2c training data*/
 	u64 p2c_train_data_offset;
-	struct amdgpu_bo *p2c_bo;
 
 	/*vram offset of the c2p training data*/
 	u64 c2p_train_data_offset;

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c

@@ -315,7 +315,7 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user *
 	default:
 		ret = -EINVAL;
 		break;
-	};
+	}
 
 	if (ret)
 		return -EINVAL;
@@ -1311,6 +1311,7 @@ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev,
 	data = con->eh_data;
 	if (!data || data->count == 0) {
 		*bps = NULL;
+		ret = -EINVAL;
 		goto out;
 	}
 
@@ -1870,7 +1871,7 @@ void amdgpu_ras_resume(struct amdgpu_device *adev)
 		 * See feature_enable_on_boot
 		 */
 		amdgpu_ras_disable_all_features(adev, 1);
-		amdgpu_ras_reset_gpu(adev, 0);
+		amdgpu_ras_reset_gpu(adev);
 	}
 }
 
@@ -1933,6 +1934,6 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
 	if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) {
 		DRM_WARN("RAS event of type ERREVENT_ATHUB_INTERRUPT detected!\n");
 
-		amdgpu_ras_reset_gpu(adev, false);
+		amdgpu_ras_reset_gpu(adev);
 	}
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h

@@ -494,8 +494,7 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
 
 int amdgpu_ras_reserve_bad_pages(struct amdgpu_device *adev);
 
-static inline int amdgpu_ras_reset_gpu(struct amdgpu_device *adev,
+static inline int amdgpu_ras_reset_gpu(struct amdgpu_device *adev)
-		bool is_baco)
 {
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 

drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c

@@ -160,7 +160,7 @@ int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
 		struct amdgpu_iv_entry *entry)
 {
 	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
-	amdgpu_ras_reset_gpu(adev, 0);
+	amdgpu_ras_reset_gpu(adev);
 
 	return AMDGPU_RAS_SUCCESS;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h

@@ -52,6 +52,8 @@ struct amdgpu_sdma_instance {
 
 struct amdgpu_sdma {
 	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
+	struct drm_gpu_scheduler    *sdma_sched[AMDGPU_MAX_SDMA_INSTANCES];
+	uint32_t		    num_sdma_sched;
 	struct amdgpu_irq_src	trap_irq;
 	struct amdgpu_irq_src	illegal_inst_irq;
 	struct amdgpu_irq_src	ecc_irq;

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c

@@ -1714,12 +1714,17 @@ static int amdgpu_ttm_training_reserve_vram_fini(struct amdgpu_device *adev)
 	amdgpu_bo_free_kernel(&ctx->c2p_bo, NULL, NULL);
 	ctx->c2p_bo = NULL;
 
-	amdgpu_bo_free_kernel(&ctx->p2c_bo, NULL, NULL);
-	ctx->p2c_bo = NULL;
-
 	return 0;
 }
 
+static u64 amdgpu_ttm_training_get_c2p_offset(u64 vram_size)
+{
+       if ((vram_size & (SZ_1M - 1)) < (SZ_4K + 1) )
+               vram_size -= SZ_1M;
+
+       return ALIGN(vram_size, SZ_1M);
+}
+
 /**
  * amdgpu_ttm_training_reserve_vram_init - create bo vram reservation from memory training
  *
@@ -1738,7 +1743,7 @@ static int amdgpu_ttm_training_reserve_vram_init(struct amdgpu_device *adev)
 		return 0;
 	}
 
-	ctx->c2p_train_data_offset = adev->fw_vram_usage.mem_train_fb_loc;
+	ctx->c2p_train_data_offset = amdgpu_ttm_training_get_c2p_offset(adev->gmc.mc_vram_size);
 	ctx->p2c_train_data_offset = (adev->gmc.mc_vram_size - GDDR6_MEM_TRAINING_OFFSET);
 	ctx->train_data_size = GDDR6_MEM_TRAINING_DATA_SIZE_IN_BYTES;
 
@@ -1747,17 +1752,6 @@ static int amdgpu_ttm_training_reserve_vram_init(struct amdgpu_device *adev)
 		  ctx->p2c_train_data_offset,
 		  ctx->c2p_train_data_offset);
 
-	ret = amdgpu_bo_create_kernel_at(adev,
-					 ctx->p2c_train_data_offset,
-					 ctx->train_data_size,
-					 AMDGPU_GEM_DOMAIN_VRAM,
-					 &ctx->p2c_bo,
-					 NULL);
-	if (ret) {
-		DRM_ERROR("alloc p2c_bo failed(%d)!\n", ret);
-		goto Err_out;
-	}
-
 	ret = amdgpu_bo_create_kernel_at(adev,
 					 ctx->c2p_train_data_offset,
 					 ctx->train_data_size,
@@ -1766,15 +1760,12 @@ static int amdgpu_ttm_training_reserve_vram_init(struct amdgpu_device *adev)
 					 NULL);
 	if (ret) {
 		DRM_ERROR("alloc c2p_bo failed(%d)!\n", ret);
-		goto Err_out;
+		amdgpu_ttm_training_reserve_vram_fini(adev);
+		return ret;
 	}
 
 	ctx->init = PSP_MEM_TRAIN_RESERVE_SUCCESS;
 	return 0;
-
-Err_out:
-	amdgpu_ttm_training_reserve_vram_fini(adev);
-	return ret;
 }
 
 /**
@@ -1987,11 +1978,13 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 
 	if (enable) {
 		struct amdgpu_ring *ring;
-		struct drm_sched_rq *rq;
+		struct drm_gpu_scheduler *sched;
 
 		ring = adev->mman.buffer_funcs_ring;
-		rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_KERNEL];
+		sched = &ring->sched;
-		r = drm_sched_entity_init(&adev->mman.entity, &rq, 1, NULL);
+		r = drm_sched_entity_init(&adev->mman.entity,
+				          DRM_SCHED_PRIORITY_KERNEL, &sched,
+					  1, NULL);
 		if (r) {
 			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
 				  r);

drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h

@@ -300,10 +300,10 @@ enum AMDGPU_UCODE_ID {
 	AMDGPU_UCODE_ID_CP_MEC2_JT,
 	AMDGPU_UCODE_ID_CP_MES,
 	AMDGPU_UCODE_ID_CP_MES_DATA,
-	AMDGPU_UCODE_ID_RLC_G,
 	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL,
 	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM,
 	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM,
+	AMDGPU_UCODE_ID_RLC_G,
 	AMDGPU_UCODE_ID_STORAGE,
 	AMDGPU_UCODE_ID_SMC,
 	AMDGPU_UCODE_ID_UVD,

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c

@@ -95,13 +95,6 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 {
 	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
 
-	/* When “Full RAS” is enabled, the per-IP interrupt sources should
-	 * be disabled and the driver should only look for the aggregated
-	 * interrupt via sync flood
-	 */
-	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
-		return AMDGPU_RAS_SUCCESS;
-
 	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
 	if (adev->umc.funcs &&
 	    adev->umc.funcs->query_ras_error_count)
@@ -113,6 +106,7 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 		err_data->err_addr =
 			kcalloc(adev->umc.max_ras_err_cnt_per_query,
 				sizeof(struct eeprom_table_record), GFP_KERNEL);
+
 		/* still call query_ras_error_address to clear error status
 		 * even NOMEM error is encountered
 		 */
@@ -132,7 +126,7 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 						err_data->err_addr_cnt))
 			DRM_WARN("Failed to add ras bad page!\n");
 
-		amdgpu_ras_reset_gpu(adev, 0);
+		amdgpu_ras_reset_gpu(adev);
 	}
 
 	kfree(err_data->err_addr);

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h

@@ -21,38 +21,6 @@
 #ifndef __AMDGPU_UMC_H__
 #define __AMDGPU_UMC_H__
 
-/* implement 64 bits REG operations via 32 bits interface */
-#define RREG64_UMC(reg)	(RREG32(reg) | \
-				((uint64_t)RREG32((reg) + 1) << 32))
-#define WREG64_UMC(reg, v)	\
-	do {	\
-		WREG32((reg), lower_32_bits(v));	\
-		WREG32((reg) + 1, upper_32_bits(v));	\
-	} while (0)
-
-/*
- * void (*func)(struct amdgpu_device *adev, struct ras_err_data *err_data,
- *				uint32_t umc_reg_offset, uint32_t channel_index)
- */
-#define amdgpu_umc_for_each_channel(func)	\
-	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;	\
-	uint32_t umc_inst, channel_inst, umc_reg_offset, channel_index;	\
-	for (umc_inst = 0; umc_inst < adev->umc.umc_inst_num; umc_inst++) {	\
-		/* enable the index mode to query eror count per channel */	\
-		adev->umc.funcs->enable_umc_index_mode(adev, umc_inst);	\
-		for (channel_inst = 0;	\
-			channel_inst < adev->umc.channel_inst_num;	\
-			channel_inst++) {	\
-			/* calc the register offset according to channel instance */	\
-			umc_reg_offset = adev->umc.channel_offs * channel_inst;	\
-			/* get channel index of interleaved memory */	\
-			channel_index = adev->umc.channel_idx_tbl[	\
-				umc_inst * adev->umc.channel_inst_num + channel_inst];	\
-			(func)(adev, err_data, umc_reg_offset, channel_index);	\
-		}	\
-	}	\
-	adev->umc.funcs->disable_umc_index_mode(adev);
-
 struct amdgpu_umc_funcs {
 	void (*err_cnt_init)(struct amdgpu_device *adev);
 	int (*ras_late_init)(struct amdgpu_device *adev);
@@ -60,9 +28,6 @@ struct amdgpu_umc_funcs {
 					void *ras_error_status);
 	void (*query_ras_error_address)(struct amdgpu_device *adev,
 					void *ras_error_status);
-	void (*enable_umc_index_mode)(struct amdgpu_device *adev,
-					uint32_t umc_instance);
-	void (*disable_umc_index_mode)(struct amdgpu_device *adev);
 	void (*init_registers)(struct amdgpu_device *adev);
 };
 

drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c

@@ -330,12 +330,13 @@ int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
 int amdgpu_uvd_entity_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	struct drm_sched_rq *rq;
+	struct drm_gpu_scheduler *sched;
 	int r;
 
 	ring = &adev->uvd.inst[0].ring;
-	rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL];
+	sched = &ring->sched;
-	r = drm_sched_entity_init(&adev->uvd.entity, &rq, 1, NULL);
+	r = drm_sched_entity_init(&adev->uvd.entity, DRM_SCHED_PRIORITY_NORMAL,
+				  &sched, 1, NULL);
 	if (r) {
 		DRM_ERROR("Failed setting up UVD kernel entity.\n");
 		return r;

drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c

@@ -240,12 +240,13 @@ int amdgpu_vce_sw_fini(struct amdgpu_device *adev)
 int amdgpu_vce_entity_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	struct drm_sched_rq *rq;
+	struct drm_gpu_scheduler *sched;
 	int r;
 
 	ring = &adev->vce.ring[0];
-	rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL];
+	sched = &ring->sched;
-	r = drm_sched_entity_init(&adev->vce.entity, &rq, 1, NULL);
+	r = drm_sched_entity_init(&adev->vce.entity, DRM_SCHED_PRIORITY_NORMAL,
+				  &sched, 1, NULL);
 	if (r != 0) {
 		DRM_ERROR("Failed setting up VCE run queue.\n");
 		return r;

drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c

@@ -28,19 +28,10 @@
 #include <linux/module.h>
 #include <linux/pci.h>
 
-#include <drm/drm.h>
-
 #include "amdgpu.h"
 #include "amdgpu_pm.h"
 #include "amdgpu_vcn.h"
 #include "soc15d.h"
-#include "soc15_common.h"
-
-#include "vcn/vcn_1_0_offset.h"
-#include "vcn/vcn_1_0_sh_mask.h"
-
-/* 1 second timeout */
-#define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
 /* Firmware Names */
 #define FIRMWARE_RAVEN		"amdgpu/raven_vcn.bin"
@@ -294,6 +285,7 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
 		if (adev->vcn.harvest_config & (1 << j))
 			continue;
+
 		for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
 			fence[j] += amdgpu_fence_count_emitted(&adev->vcn.inst[j].ring_enc[i]);
 		}
@@ -306,26 +298,17 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 			else
 				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
 
-			if (amdgpu_fence_count_emitted(&adev->jpeg.inst[j].ring_dec))
-				new_state.jpeg = VCN_DPG_STATE__PAUSE;
-			else
-				new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
-
 			adev->vcn.pause_dpg_mode(adev, &new_state);
 		}
 
-		fence[j] += amdgpu_fence_count_emitted(&adev->jpeg.inst[j].ring_dec);
 		fence[j] += amdgpu_fence_count_emitted(&adev->vcn.inst[j].ring_dec);
 		fences += fence[j];
 	}
 
 	if (fences == 0) {
 		amdgpu_gfx_off_ctrl(adev, true);
-		if (adev->asic_type < CHIP_ARCTURUS && adev->pm.dpm_enabled)
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-			amdgpu_dpm_enable_uvd(adev, false);
+		       AMD_PG_STATE_GATE);
-		else
-			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-							       AMD_PG_STATE_GATE);
 	} else {
 		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
 	}
@@ -338,11 +321,8 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 
 	if (set_clocks) {
 		amdgpu_gfx_off_ctrl(adev, false);
-		if (adev->asic_type < CHIP_ARCTURUS && adev->pm.dpm_enabled)
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-			amdgpu_dpm_enable_uvd(adev, true);
+		       AMD_PG_STATE_UNGATE);
-		else
-			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-							       AMD_PG_STATE_UNGATE);
 	}
 
 	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)	{
@@ -358,15 +338,8 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 		else
 			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
 
-		if (amdgpu_fence_count_emitted(&adev->jpeg.inst[ring->me].ring_dec))
-			new_state.jpeg = VCN_DPG_STATE__PAUSE;
-		else
-			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
-
 		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC)
 			new_state.fw_based = VCN_DPG_STATE__PAUSE;
-		else if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
-			new_state.jpeg = VCN_DPG_STATE__PAUSE;
 
 		adev->vcn.pause_dpg_mode(adev, &new_state);
 	}
@@ -518,9 +491,14 @@ static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 
 int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
+	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence;
 	long r;
 
+	/* temporarily disable ib test for sriov */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	r = amdgpu_vcn_dec_get_create_msg(ring, 1, NULL);
 	if (r)
 		goto error;
@@ -676,10 +654,15 @@ err:
 
 int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
+	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence = NULL;
 	struct amdgpu_bo *bo = NULL;
 	long r;
 
+	/* temporarily disable ib test for sriov */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	r = amdgpu_bo_create_reserved(ring->adev, 128 * 1024, PAGE_SIZE,
 				      AMDGPU_GEM_DOMAIN_VRAM,
 				      &bo, NULL, NULL);

drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h

@@ -31,6 +31,7 @@
 #define AMDGPU_VCN_MAX_ENC_RINGS	3
 
 #define AMDGPU_MAX_VCN_INSTANCES	2
+#define AMDGPU_MAX_VCN_ENC_RINGS  AMDGPU_VCN_MAX_ENC_RINGS * AMDGPU_MAX_VCN_INSTANCES
 
 #define AMDGPU_VCN_HARVEST_VCN0 (1 << 0)
 #define AMDGPU_VCN_HARVEST_VCN1 (1 << 1)
@@ -56,6 +57,9 @@
 #define VCN_VID_IP_ADDRESS_2_0		0x0
 #define VCN_AON_IP_ADDRESS_2_0		0x30000
 
+/* 1 second timeout */
+#define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
 #define RREG32_SOC15_DPG_MODE(ip, inst, reg, mask, sram_sel) 				\
 	({	WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_MASK, mask); 			\
 		WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_CTL, 				\
@@ -186,8 +190,12 @@ struct amdgpu_vcn {
 	uint32_t		*dpg_sram_curr_addr;
 
 	uint8_t	num_vcn_inst;
-	struct amdgpu_vcn_inst	inst[AMDGPU_MAX_VCN_INSTANCES];
+	struct amdgpu_vcn_inst	 inst[AMDGPU_MAX_VCN_INSTANCES];
-	struct amdgpu_vcn_reg	internal;
+	struct amdgpu_vcn_reg	 internal;
+	struct drm_gpu_scheduler *vcn_enc_sched[AMDGPU_MAX_VCN_ENC_RINGS];
+	struct drm_gpu_scheduler *vcn_dec_sched[AMDGPU_MAX_VCN_INSTANCES];
+	uint32_t		 num_vcn_enc_sched;
+	uint32_t		 num_vcn_dec_sched;
 
 	unsigned	harvest_config;
 	int (*pause_dpg_mode)(struct amdgpu_device *adev,

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -2753,14 +2753,17 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	spin_lock_init(&vm->invalidated_lock);
 	INIT_LIST_HEAD(&vm->freed);
 
+
 	/* create scheduler entities for page table updates */
-	r = drm_sched_entity_init(&vm->direct, adev->vm_manager.vm_pte_rqs,
+	r = drm_sched_entity_init(&vm->direct, DRM_SCHED_PRIORITY_NORMAL,
-				  adev->vm_manager.vm_pte_num_rqs, NULL);
+				  adev->vm_manager.vm_pte_scheds,
+				  adev->vm_manager.vm_pte_num_scheds, NULL);
 	if (r)
 		return r;
 
-	r = drm_sched_entity_init(&vm->delayed, adev->vm_manager.vm_pte_rqs,
+	r = drm_sched_entity_init(&vm->delayed, DRM_SCHED_PRIORITY_NORMAL,
-				  adev->vm_manager.vm_pte_num_rqs, NULL);
+				  adev->vm_manager.vm_pte_scheds,
+				  adev->vm_manager.vm_pte_num_scheds, NULL);
 	if (r)
 		goto error_free_direct;
 

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h

@@ -327,8 +327,8 @@ struct amdgpu_vm_manager {
 	u64					vram_base_offset;
 	/* vm pte handling */
 	const struct amdgpu_vm_pte_funcs	*vm_pte_funcs;
-	struct drm_sched_rq			*vm_pte_rqs[AMDGPU_MAX_RINGS];
+	struct drm_gpu_scheduler		*vm_pte_scheds[AMDGPU_MAX_RINGS];
-	unsigned				vm_pte_num_rqs;
+	unsigned				vm_pte_num_scheds;
 	struct amdgpu_ring			*page_fault;
 
 	/* partial resident texture handling */

drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c

@@ -261,6 +261,7 @@ struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev, int lo
 	INIT_LIST_HEAD(&tmp->device_list);
 	mutex_init(&tmp->hive_lock);
 	mutex_init(&tmp->reset_lock);
+	task_barrier_init(&tmp->tb);
 
 	if (lock)
 		mutex_lock(&tmp->hive_lock);
@@ -408,6 +409,8 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
 	top_info->num_nodes = count;
 	hive->number_devices = count;
 
+	task_barrier_add_task(&hive->tb);
+
 	if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP)) {
 		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
 			/* update node list for other device in the hive */
@@ -470,6 +473,7 @@ void amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
 		mutex_destroy(&hive->hive_lock);
 		mutex_destroy(&hive->reset_lock);
 	} else {
+		task_barrier_rem_task(&hive->tb);
 		amdgpu_xgmi_sysfs_rem_dev_info(adev, hive);
 		mutex_unlock(&hive->hive_lock);
 	}

drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h

@@ -22,6 +22,7 @@
 #ifndef __AMDGPU_XGMI_H__
 #define __AMDGPU_XGMI_H__
 
+#include <drm/task_barrier.h>
 #include "amdgpu_psp.h"
 
 struct amdgpu_hive_info {
@@ -33,6 +34,7 @@ struct amdgpu_hive_info {
 	struct device_attribute dev_attr;
 	struct amdgpu_device *adev;
 	int pstate; /*0 -- low , 1 -- high , -1 unknown*/
+	struct task_barrier tb;
 };
 
 struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev, int lock);

drivers/gpu/drm/amd/amdgpu/cik_sdma.c

@@ -228,7 +228,7 @@ static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
 	u32 extra_bits = vmid & 0xf;
 
 	/* IB packet must end on a 8 DW boundary */
-	cik_sdma_ring_insert_nop(ring, (12 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	cik_sdma_ring_insert_nop(ring, (4 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_INDIRECT_BUFFER, 0, extra_bits));
 	amdgpu_ring_write(ring, ib->gpu_addr & 0xffffffe0); /* base must be 32 byte aligned */
@@ -811,7 +811,7 @@ static void cik_sdma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1372,16 +1372,14 @@ static const struct amdgpu_vm_pte_funcs cik_sdma_vm_pte_funcs = {
 
 static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
-	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
 	adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] =
-		adev->vm_manager.vm_pte_rqs[i] =
+			&adev->sdma.instance[i].ring.sched;
-			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
-	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version cik_sdma_ip_block =

drivers/gpu/drm/amd/amdgpu/df_v3_6.c

@@ -183,6 +183,61 @@ static void df_v3_6_perfmon_wreg(struct amdgpu_device *adev, uint32_t lo_addr,
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 }
 
+/* same as perfmon_wreg but return status on write value check */
+static int df_v3_6_perfmon_arm_with_status(struct amdgpu_device *adev,
+					  uint32_t lo_addr, uint32_t lo_val,
+					  uint32_t hi_addr, uint32_t  hi_val)
+{
+	unsigned long flags, address, data;
+	uint32_t lo_val_rb, hi_val_rb;
+
+	address = adev->nbio.funcs->get_pcie_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, lo_addr);
+	WREG32(data, lo_val);
+	WREG32(address, hi_addr);
+	WREG32(data, hi_val);
+
+	WREG32(address, lo_addr);
+	lo_val_rb = RREG32(data);
+	WREG32(address, hi_addr);
+	hi_val_rb = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+
+	if (!(lo_val == lo_val_rb && hi_val == hi_val_rb))
+		return -EBUSY;
+
+	return 0;
+}
+
+
+/*
+ * retry arming counters every 100 usecs within 1 millisecond interval.
+ * if retry fails after time out, return error.
+ */
+#define ARM_RETRY_USEC_TIMEOUT	1000
+#define ARM_RETRY_USEC_INTERVAL	100
+static int df_v3_6_perfmon_arm_with_retry(struct amdgpu_device *adev,
+					  uint32_t lo_addr, uint32_t lo_val,
+					  uint32_t hi_addr, uint32_t  hi_val)
+{
+	int countdown = ARM_RETRY_USEC_TIMEOUT;
+
+	while (countdown) {
+
+		if (!df_v3_6_perfmon_arm_with_status(adev, lo_addr, lo_val,
+						     hi_addr, hi_val))
+			break;
+
+		countdown -= ARM_RETRY_USEC_INTERVAL;
+		udelay(ARM_RETRY_USEC_INTERVAL);
+	}
+
+	return countdown > 0 ? 0 : -ETIME;
+}
+
 /* get the number of df counters available */
 static ssize_t df_v3_6_get_df_cntr_avail(struct device *dev,
 		struct device_attribute *attr,
@@ -334,20 +389,20 @@ static void df_v3_6_pmc_get_addr(struct amdgpu_device *adev,
 	switch (target_cntr) {
 
 	case 0:
-		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo0 : smnPerfMonCtrLo0;
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo4 : smnPerfMonCtrLo4;
-		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi0 : smnPerfMonCtrHi0;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi4 : smnPerfMonCtrHi4;
 		break;
 	case 1:
-		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo1 : smnPerfMonCtrLo1;
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo5 : smnPerfMonCtrLo5;
-		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi1 : smnPerfMonCtrHi1;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi5 : smnPerfMonCtrHi5;
 		break;
 	case 2:
-		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo2 : smnPerfMonCtrLo2;
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo6 : smnPerfMonCtrLo6;
-		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi2 : smnPerfMonCtrHi2;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi6 : smnPerfMonCtrHi6;
 		break;
 	case 3:
-		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo3 : smnPerfMonCtrLo3;
+		*lo_base_addr = is_ctrl ? smnPerfMonCtlLo7 : smnPerfMonCtrLo7;
-		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi3 : smnPerfMonCtrHi3;
+		*hi_base_addr = is_ctrl ? smnPerfMonCtlHi7 : smnPerfMonCtrHi7;
 		break;
 
 	}
@@ -422,6 +477,44 @@ static int df_v3_6_pmc_add_cntr(struct amdgpu_device *adev,
 	return -ENOSPC;
 }
 
+#define DEFERRED_ARM_MASK	(1 << 31)
+static int df_v3_6_pmc_set_deferred(struct amdgpu_device *adev,
+				    uint64_t config, bool is_deferred)
+{
+	int target_cntr;
+
+	target_cntr = df_v3_6_pmc_config_2_cntr(adev, config);
+
+	if (target_cntr < 0)
+		return -EINVAL;
+
+	if (is_deferred)
+		adev->df_perfmon_config_assign_mask[target_cntr] |=
+							DEFERRED_ARM_MASK;
+	else
+		adev->df_perfmon_config_assign_mask[target_cntr] &=
+							~DEFERRED_ARM_MASK;
+
+	return 0;
+}
+
+static bool df_v3_6_pmc_is_deferred(struct amdgpu_device *adev,
+				    uint64_t config)
+{
+	int target_cntr;
+
+	target_cntr = df_v3_6_pmc_config_2_cntr(adev, config);
+
+	/*
+	 * we never get target_cntr < 0 since this funciton is only called in
+	 * pmc_count for now but we should check anyways.
+	 */
+	return (target_cntr >= 0 &&
+			(adev->df_perfmon_config_assign_mask[target_cntr]
+			& DEFERRED_ARM_MASK));
+
+}
+
 /* release performance counter */
 static void df_v3_6_pmc_release_cntr(struct amdgpu_device *adev,
 				     uint64_t config)
@@ -451,29 +544,33 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
 			     int is_enable)
 {
 	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
-	int ret = 0;
+	int err = 0, ret = 0;
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA20:
+		if (is_enable)
+			return df_v3_6_pmc_add_cntr(adev, config);
 
 		df_v3_6_reset_perfmon_cntr(adev, config);
 
-		if (is_enable) {
+		ret = df_v3_6_pmc_get_ctrl_settings(adev,
-			ret = df_v3_6_pmc_add_cntr(adev, config);
-		} else {
-			ret = df_v3_6_pmc_get_ctrl_settings(adev,
 					config,
 					&lo_base_addr,
 					&hi_base_addr,
 					&lo_val,
 					&hi_val);
 
-			if (ret)
+		if (ret)
-				return ret;
+			return ret;
 
-			df_v3_6_perfmon_wreg(adev, lo_base_addr, lo_val,
+		err = df_v3_6_perfmon_arm_with_retry(adev,
-					hi_base_addr, hi_val);
+						     lo_base_addr,
-		}
+						     lo_val,
+						     hi_base_addr,
+						     hi_val);
+
+		if (err)
+			ret = df_v3_6_pmc_set_deferred(adev, config, true);
 
 		break;
 	default:
@@ -501,7 +598,7 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config,
 		if (ret)
 			return ret;
 
-		df_v3_6_perfmon_wreg(adev, lo_base_addr, 0, hi_base_addr, 0);
+		df_v3_6_reset_perfmon_cntr(adev, config);
 
 		if (is_disable)
 			df_v3_6_pmc_release_cntr(adev, config);
@@ -518,18 +615,29 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev,
 				  uint64_t config,
 				  uint64_t *count)
 {
-	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
+	uint32_t lo_base_addr, hi_base_addr, lo_val = 0, hi_val = 0;
 	*count = 0;
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA20:
-
 		df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr,
 				      &hi_base_addr);
 
 		if ((lo_base_addr == 0) || (hi_base_addr == 0))
 			return;
 
+		/* rearm the counter or throw away count value on failure */
+		if (df_v3_6_pmc_is_deferred(adev, config)) {
+			int rearm_err = df_v3_6_perfmon_arm_with_status(adev,
+							lo_base_addr, lo_val,
+							hi_base_addr, hi_val);
+
+			if (rearm_err)
+				return;
+
+			df_v3_6_pmc_set_deferred(adev, config, false);
+		}
+
 		df_v3_6_perfmon_rreg(adev, lo_base_addr, &lo_val,
 				hi_base_addr, &hi_val);
 
@@ -542,7 +650,6 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev,
 			 config, lo_base_addr, hi_base_addr, lo_val, hi_val);
 
 		break;
-
 	default:
 		break;
 	}

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c

@@ -471,18 +471,10 @@ static int gfx_v10_0_ring_test_ring(struct amdgpu_ring *ring)
 		else
 			udelay(1);
 	}
-	if (i < adev->usec_timeout) {
+
-		if (amdgpu_emu_mode == 1)
+	if (i >= adev->usec_timeout)
-			DRM_INFO("ring test on %d succeeded in %d msecs\n",
+		r = -ETIMEDOUT;
-				 ring->idx, i);
+
-		else
-			DRM_INFO("ring test on %d succeeded in %d usecs\n",
-				 ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
-			  ring->idx, scratch, tmp);
-		r = -EINVAL;
-	}
 	amdgpu_gfx_scratch_free(adev, scratch);
 
 	return r;
@@ -532,14 +524,10 @@ static int gfx_v10_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	}
 
 	tmp = RREG32(scratch);
-	if (tmp == 0xDEADBEEF) {
+	if (tmp == 0xDEADBEEF)
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
 		r = 0;
-	} else {
+	else
-		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
-			  scratch, tmp);
 		r = -EINVAL;
-	}
 err2:
 	amdgpu_ib_free(adev, &ib, NULL);
 	dma_fence_put(f);
@@ -588,8 +576,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
 	}
 
 	if (adev->gfx.cp_fw_write_wait == false)
-		DRM_WARN_ONCE("Warning: check cp_fw_version and update it to realize \
+		DRM_WARN_ONCE("CP firmware version too old, please update!");
-			      GRBM requires 1-cycle delay in cp firmware\n");
 }
 
 
@@ -1963,7 +1950,7 @@ static int gfx_v10_0_parse_rlc_toc(struct amdgpu_device *adev)
 		rlc_autoload_info[rlc_toc->id].size = rlc_toc->size * 4;
 
 		rlc_toc++;
-	};
+	}
 
 	return 0;
 }
@@ -3606,23 +3593,16 @@ static int gfx_v10_0_cp_resume(struct amdgpu_device *adev)
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
 		ring = &adev->gfx.gfx_ring[i];
-		DRM_INFO("gfx %d ring me %d pipe %d q %d\n",
+		r = amdgpu_ring_test_helper(ring);
-			 i, ring->me, ring->pipe, ring->queue);
+		if (r)
-		r = amdgpu_ring_test_ring(ring);
-		if (r) {
-			ring->sched.ready = false;
 			return r;
-		}
 	}
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		ring = &adev->gfx.compute_ring[i];
-		ring->sched.ready = true;
+		r = amdgpu_ring_test_helper(ring);
-		DRM_INFO("compute ring %d mec %d pipe %d q %d\n",
-			 i, ring->me, ring->pipe, ring->queue);
-		r = amdgpu_ring_test_ring(ring);
 		if (r)
-			ring->sched.ready = false;
+			return r;
 	}
 
 	return 0;

drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c

@@ -48,8 +48,15 @@
 
 #include "amdgpu_ras.h"
 
-#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma0/sdma0_4_2_offset.h"
-#include "sdma1/sdma1_4_0_offset.h"
+#include "sdma1/sdma1_4_2_offset.h"
+#include "sdma2/sdma2_4_2_2_offset.h"
+#include "sdma3/sdma3_4_2_2_offset.h"
+#include "sdma4/sdma4_4_2_2_offset.h"
+#include "sdma5/sdma5_4_2_2_offset.h"
+#include "sdma6/sdma6_4_2_2_offset.h"
+#include "sdma7/sdma7_4_2_2_offset.h"
+
 #define GFX9_NUM_GFX_RINGS     1
 #define GFX9_MEC_HPD_SIZE 4096
 #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L
@@ -981,8 +988,7 @@ static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev)
 	    (adev->gfx.mec_feature_version < 46) ||
 	    (adev->gfx.pfp_fw_version < 0x000000b7) ||
 	    (adev->gfx.pfp_feature_version < 46))
-		DRM_WARN_ONCE("Warning: check cp_fw_version and update it to realize \
+		DRM_WARN_ONCE("CP firmware version too old, please update!");
-			      GRBM requires 1-cycle delay in cp firmware\n");
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
@@ -1042,17 +1048,10 @@ static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev)
 	case CHIP_VEGA20:
 		break;
 	case CHIP_RAVEN:
-		/* Disable GFXOFF on original raven.  There are combinations
+		if (!(adev->rev_id >= 0x8 ||
-		 * of sbios and platforms that are not stable.
+		      adev->pdev->device == 0x15d8) &&
-		 */
+		    (adev->pm.fw_version < 0x41e2b || /* not raven1 fresh */
-		if (!(adev->rev_id >= 0x8 || adev->pdev->device == 0x15d8))
+		     !adev->gfx.rlc.is_rlc_v2_1)) /* without rlc save restore ucodes */
-			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
-		else if (!(adev->rev_id >= 0x8 || adev->pdev->device == 0x15d8)
-			 &&((adev->gfx.rlc_fw_version != 106 &&
-			     adev->gfx.rlc_fw_version < 531) ||
-			    (adev->gfx.rlc_fw_version == 53815) ||
-			    (adev->gfx.rlc_feature_version < 1) ||
-			    !adev->gfx.rlc.is_rlc_v2_1))
 			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
 
 		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
@@ -3933,43 +3932,58 @@ static const u32 sgpr_init_compute_shader[] =
 	0xbe800080, 0xbf810000,
 };
 
+/* When below register arrays changed, please update gpr_reg_size,
+  and sec_ded_counter_reg_size in function gfx_v9_0_do_edc_gpr_workarounds,
+  to cover all gfx9 ASICs */
 static const struct soc15_reg_entry vgpr_init_regs[] = {
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 4 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x3f },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x400000 },  /* 64KB LDS */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
 };
 
 static const struct soc15_reg_entry sgpr1_init_regs[] = {
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x000000ff },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x000000ff },
+};
+
+static const struct soc15_reg_entry sgpr2_init_regs[] = {
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
-};
-
-static const struct soc15_reg_entry sgpr2_init_regs[] = {
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
 };
 
 static const struct soc15_reg_entry sec_ded_counter_registers[] = {
@@ -4006,9 +4020,15 @@ static const struct soc15_reg_entry sec_ded_counter_registers[] = {
    { SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16},
    { SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2},
    { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6},
+   { SOC15_REG_ENTRY(HDP, 0, mmHDP_EDC_CNT), 0, 1, 1},
    { SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER), 0, 1, 1},
    { SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_EDC_COUNTER), 0, 1, 1},
-   { SOC15_REG_ENTRY(HDP, 0, mmHDP_EDC_CNT), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA2, 0, mmSDMA2_EDC_COUNTER), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA3, 0, mmSDMA3_EDC_COUNTER), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA4, 0, mmSDMA4_EDC_COUNTER), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA5, 0, mmSDMA5_EDC_COUNTER), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA6, 0, mmSDMA6_EDC_COUNTER), 0, 1, 1},
+   { SOC15_REG_ENTRY(SDMA7, 0, mmSDMA7_EDC_COUNTER), 0, 1, 1},
 };
 
 static int gfx_v9_0_do_edc_gds_workarounds(struct amdgpu_device *adev)
@@ -4067,6 +4087,13 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 	unsigned total_size, vgpr_offset, sgpr_offset;
 	u64 gpu_addr;
 
+	int compute_dim_x = adev->gfx.config.max_shader_engines *
+						adev->gfx.config.max_cu_per_sh *
+						adev->gfx.config.max_sh_per_se;
+	int sgpr_work_group_size = 5;
+	int gpr_reg_size = compute_dim_x / 16 + 6;
+	int sec_ded_counter_reg_size = adev->sdma.num_instances + 34;
+
 	/* only support when RAS is enabled */
 	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
 		return 0;
@@ -4076,11 +4103,11 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 		return 0;
 
 	total_size =
-		((ARRAY_SIZE(vgpr_init_regs) * 3) + 4 + 5 + 2) * 4;
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* VGPRS */
 	total_size +=
-		((ARRAY_SIZE(sgpr1_init_regs) * 3) + 4 + 5 + 2) * 4;
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS1 */
 	total_size +=
-		((ARRAY_SIZE(sgpr2_init_regs) * 3) + 4 + 5 + 2) * 4;
+		(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS2 */
 	total_size = ALIGN(total_size, 256);
 	vgpr_offset = total_size;
 	total_size += ALIGN(sizeof(vgpr_init_compute_shader), 256);
@@ -4107,7 +4134,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* VGPR */
 	/* write the register state for the compute dispatch */
-	for (i = 0; i < ARRAY_SIZE(vgpr_init_regs); i++) {
+	for (i = 0; i < gpr_reg_size; i++) {
 		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
 		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(vgpr_init_regs[i])
 								- PACKET3_SET_SH_REG_START;
@@ -4123,7 +4150,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* write dispatch packet */
 	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
-	ib.ptr[ib.length_dw++] = 0x40*2; /* x */
+	ib.ptr[ib.length_dw++] = compute_dim_x; /* x */
 	ib.ptr[ib.length_dw++] = 1; /* y */
 	ib.ptr[ib.length_dw++] = 1; /* z */
 	ib.ptr[ib.length_dw++] =
@@ -4135,7 +4162,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* SGPR1 */
 	/* write the register state for the compute dispatch */
-	for (i = 0; i < ARRAY_SIZE(sgpr1_init_regs); i++) {
+	for (i = 0; i < gpr_reg_size; i++) {
 		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
 		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr1_init_regs[i])
 								- PACKET3_SET_SH_REG_START;
@@ -4151,7 +4178,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* write dispatch packet */
 	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
-	ib.ptr[ib.length_dw++] = 0xA0*2; /* x */
+	ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */
 	ib.ptr[ib.length_dw++] = 1; /* y */
 	ib.ptr[ib.length_dw++] = 1; /* z */
 	ib.ptr[ib.length_dw++] =
@@ -4163,7 +4190,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* SGPR2 */
 	/* write the register state for the compute dispatch */
-	for (i = 0; i < ARRAY_SIZE(sgpr2_init_regs); i++) {
+	for (i = 0; i < gpr_reg_size; i++) {
 		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
 		ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr2_init_regs[i])
 								- PACKET3_SET_SH_REG_START;
@@ -4179,7 +4206,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* write dispatch packet */
 	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
-	ib.ptr[ib.length_dw++] = 0xA0*2; /* x */
+	ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */
 	ib.ptr[ib.length_dw++] = 1; /* y */
 	ib.ptr[ib.length_dw++] = 1; /* z */
 	ib.ptr[ib.length_dw++] =
@@ -4205,7 +4232,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 
 	/* read back registers to clear the counters */
 	mutex_lock(&adev->grbm_idx_mutex);
-	for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) {
+	for (i = 0; i < sec_ded_counter_reg_size; i++) {
 		for (j = 0; j < sec_ded_counter_registers[i].se_num; j++) {
 			for (k = 0; k < sec_ded_counter_registers[i].instance; k++) {
 				gfx_v9_0_select_se_sh(adev, j, 0x0, k);

drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c

@@ -75,40 +75,45 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 	WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
 	WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
 
-	/* Program the system aperture low logical page number. */
+	if (!amdgpu_sriov_vf(adev) || adev->asic_type <= CHIP_VEGA10) {
-	WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		/* Program the system aperture low logical page number. */
-		     min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
 
-	if (adev->asic_type == CHIP_RAVEN && adev->rev_id >= 0x8)
+		if (adev->asic_type == CHIP_RAVEN && adev->rev_id >= 0x8)
-		/*
+			/*
-		 * Raven2 has a HW issue that it is unable to use the vram which
+			* Raven2 has a HW issue that it is unable to use the
-		 * is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR. So here is the
+			* vram which is out of MC_VM_SYSTEM_APERTURE_HIGH_ADDR.
-		 * workaround that increase system aperture high address (add 1)
+			* So here is the workaround that increase system
-		 * to get rid of the VM fault and hardware hang.
+			* aperture high address (add 1) to get rid of the VM
-		 */
+			* fault and hardware hang.
-		WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+			*/
-			     max((adev->gmc.fb_end >> 18) + 0x1,
+			WREG32_SOC15_RLC(GC, 0,
-				 adev->gmc.agp_end >> 18));
+					 mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-	else
+					 max((adev->gmc.fb_end >> 18) + 0x1,
-		WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+					     adev->gmc.agp_end >> 18));
-			     max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+		else
+			WREG32_SOC15_RLC(
+				GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+				max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
 
-	/* Set default page address. */
+		/* Set default page address. */
-	value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start
+		value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start +
-		+ adev->vm_manager.vram_base_offset;
+			adev->vm_manager.vram_base_offset;
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
-		     (u32)(value >> 12));
+			     (u32)(value >> 12));
-	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
-		     (u32)(value >> 44));
+			     (u32)(value >> 44));
 
-	/* Program "protection fault". */
+		/* Program "protection fault". */
-	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
-		     (u32)(adev->dummy_page_addr >> 12));
+			     (u32)(adev->dummy_page_addr >> 12));
-	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
-		     (u32)((u64)adev->dummy_page_addr >> 44));
+			     (u32)((u64)adev->dummy_page_addr >> 44));
 
-	WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2,
+		WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2,
-		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+			       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	}
 }
 
 static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
@@ -264,7 +269,7 @@ static void gfxhub_v1_0_program_invalidation(struct amdgpu_device *adev)
 
 int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev)
 {
-	if (amdgpu_sriov_vf(adev)) {
+	if (amdgpu_sriov_vf(adev) && adev->asic_type != CHIP_ARCTURUS) {
 		/*
 		 * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
 		 * VF copy registers so vbios post doesn't program them, for
@@ -280,10 +285,12 @@ int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev)
 	gfxhub_v1_0_init_gart_aperture_regs(adev);
 	gfxhub_v1_0_init_system_aperture_regs(adev);
 	gfxhub_v1_0_init_tlb_regs(adev);
-	gfxhub_v1_0_init_cache_regs(adev);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_0_init_cache_regs(adev);
 
 	gfxhub_v1_0_enable_system_domain(adev);
-	gfxhub_v1_0_disable_identity_aperture(adev);
+	if (!amdgpu_sriov_vf(adev))
+		gfxhub_v1_0_disable_identity_aperture(adev);
 	gfxhub_v1_0_setup_vmid_config(adev);
 	gfxhub_v1_0_program_invalidation(adev);
 

drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c

@@ -564,22 +564,11 @@ static int gmc_v10_0_early_init(void *handle)
 static int gmc_v10_0_late_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	unsigned vm_inv_eng[AMDGPU_MAX_VMHUBS] = { 4, 4 };
+	int r;
-	unsigned i;
 
-	for(i = 0; i < adev->num_rings; ++i) {
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
-		struct amdgpu_ring *ring = adev->rings[i];
+	if (r)
-		unsigned vmhub = ring->funcs->vmhub;
+		return r;
-
-		ring->vm_inv_eng = vm_inv_eng[vmhub]++;
-		dev_info(adev->dev, "ring %u(%s) uses VM inv eng %u on hub %u\n",
-			 ring->idx, ring->name, ring->vm_inv_eng,
-			 ring->funcs->vmhub);
-	}
-
-	/* Engine 17 is used for GART flushes */
-	for(i = 0; i < AMDGPU_MAX_VMHUBS; ++i)
-		BUG_ON(vm_inv_eng[i] > 17);
 
 	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
 }

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -207,6 +207,11 @@ static int gmc_v9_0_ecc_interrupt_state(struct amdgpu_device *adev,
 {
 	u32 bits, i, tmp, reg;
 
+	/* Devices newer then VEGA10/12 shall have these programming
+	     sequences performed by PSP BL */
+	if (adev->asic_type >= CHIP_VEGA20)
+		return 0;
+
 	bits = 0x7f;
 
 	switch (state) {
@@ -393,8 +398,10 @@ static void gmc_v9_0_set_irq_funcs(struct amdgpu_device *adev)
 	adev->gmc.vm_fault.num_types = 1;
 	adev->gmc.vm_fault.funcs = &gmc_v9_0_irq_funcs;
 
-	adev->gmc.ecc_irq.num_types = 1;
+	if (!amdgpu_sriov_vf(adev)) {
-	adev->gmc.ecc_irq.funcs = &gmc_v9_0_ecc_funcs;
+		adev->gmc.ecc_irq.num_types = 1;
+		adev->gmc.ecc_irq.funcs = &gmc_v9_0_ecc_funcs;
+	}
 }
 
 static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid,
@@ -790,36 +797,6 @@ static bool gmc_v9_0_keep_stolen_memory(struct amdgpu_device *adev)
 	}
 }
 
-static int gmc_v9_0_allocate_vm_inv_eng(struct amdgpu_device *adev)
-{
-	struct amdgpu_ring *ring;
-	unsigned vm_inv_engs[AMDGPU_MAX_VMHUBS] =
-		{GFXHUB_FREE_VM_INV_ENGS_BITMAP, MMHUB_FREE_VM_INV_ENGS_BITMAP,
-		GFXHUB_FREE_VM_INV_ENGS_BITMAP};
-	unsigned i;
-	unsigned vmhub, inv_eng;
-
-	for (i = 0; i < adev->num_rings; ++i) {
-		ring = adev->rings[i];
-		vmhub = ring->funcs->vmhub;
-
-		inv_eng = ffs(vm_inv_engs[vmhub]);
-		if (!inv_eng) {
-			dev_err(adev->dev, "no VM inv eng for ring %s\n",
-				ring->name);
-			return -EINVAL;
-		}
-
-		ring->vm_inv_eng = inv_eng - 1;
-		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);
-	}
-
-	return 0;
-}
-
 static int gmc_v9_0_late_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -828,7 +805,7 @@ static int gmc_v9_0_late_init(void *handle)
 	if (!gmc_v9_0_keep_stolen_memory(adev))
 		amdgpu_bo_late_init(adev);
 
-	r = gmc_v9_0_allocate_vm_inv_eng(adev);
+	r = amdgpu_gmc_allocate_vm_inv_eng(adev);
 	if (r)
 		return r;
 	/* Check if ecc is available */
@@ -1112,11 +1089,13 @@ static int gmc_v9_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	/* interrupt sent to DF. */
+	if (!amdgpu_sriov_vf(adev)) {
-	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DF, 0,
+		/* interrupt sent to DF. */
-			&adev->gmc.ecc_irq);
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_DF, 0,
-	if (r)
+				      &adev->gmc.ecc_irq);
-		return r;
+		if (r)
+			return r;
+	}
 
 	/* Set the internal MC address mask
 	 * This is the max address of the GPU's
@@ -1302,12 +1281,13 @@ static int gmc_v9_0_hw_init(void *handle)
 	else
 		value = true;
 
-	gfxhub_v1_0_set_fault_enable_default(adev, value);
+	if (!amdgpu_sriov_vf(adev)) {
-	if (adev->asic_type == CHIP_ARCTURUS)
+		gfxhub_v1_0_set_fault_enable_default(adev, value);
-		mmhub_v9_4_set_fault_enable_default(adev, value);
+		if (adev->asic_type == CHIP_ARCTURUS)
-	else
+			mmhub_v9_4_set_fault_enable_default(adev, value);
-		mmhub_v1_0_set_fault_enable_default(adev, value);
+		else
-
+			mmhub_v1_0_set_fault_enable_default(adev, value);
+	}
 	for (i = 0; i < adev->num_vmhubs; ++i)
 		gmc_v9_0_flush_gpu_tlb(adev, 0, i, 0);
 

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.h

@@ -24,16 +24,6 @@
 #ifndef __GMC_V9_0_H__
 #define __GMC_V9_0_H__
 
-	/*
-	 * The latest engine allocation on gfx9 is:
-	 * Engine 2, 3: firmware
-	 * Engine 0, 1, 4~16: amdgpu ring,
-	 *                    subject to change when ring number changes
-	 * Engine 17: Gart flushes
-	 */
-#define GFXHUB_FREE_VM_INV_ENGS_BITMAP		0x1FFF3
-#define MMHUB_FREE_VM_INV_ENGS_BITMAP		0x1FFF3
-
 extern const struct amd_ip_funcs gmc_v9_0_ip_funcs;
 extern const struct amdgpu_ip_block_version gmc_v9_0_ip_block;
 #endif

drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c

@@ -25,6 +25,7 @@
 #include "amdgpu_jpeg.h"
 #include "soc15.h"
 #include "soc15d.h"
+#include "vcn_v1_0.h"
 
 #include "vcn/vcn_1_0_offset.h"
 #include "vcn/vcn_1_0_sh_mask.h"
@@ -561,7 +562,7 @@ static const struct amdgpu_ring_funcs jpeg_v1_0_decode_ring_vm_funcs = {
 	.insert_start = jpeg_v1_0_decode_ring_insert_start,
 	.insert_end = jpeg_v1_0_decode_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
+	.begin_use = vcn_v1_0_ring_begin_use,
 	.end_use = amdgpu_vcn_ring_end_use,
 	.emit_wreg = jpeg_v1_0_decode_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v1_0_decode_ring_emit_reg_wait,

drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c

@@ -128,45 +128,53 @@ static void mmhub_v9_4_init_system_aperture_regs(struct amdgpu_device *adev,
 			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
 			    adev->gmc.agp_start >> 24);
 
-	/* Program the system aperture low logical page number. */
+	if (!amdgpu_sriov_vf(adev)) {
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+		/* Program the system aperture low logical page number. */
-			    mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		WREG32_SOC15_OFFSET(
-			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			MMHUB, 0, mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_LOW_ADDR,
-			    min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
-			    mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		WREG32_SOC15_OFFSET(
-			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			MMHUB, 0, mmVMSHAREDVC0_MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
-			    max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			max(adev->gmc.fb_end, adev->gmc.agp_end) >> 18);
 
-	/* Set default page address. */
+		/* Set default page address. */
-	value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start +
+		value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start +
-		adev->vm_manager.vram_base_offset;
+			adev->vm_manager.vram_base_offset;
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
 			mmVMSHAREDPF0_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
 			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
 			(u32)(value >> 12));
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+		WREG32_SOC15_OFFSET(
+			MMHUB, 0,
 			mmVMSHAREDPF0_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
 			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
 			(u32)(value >> 44));
 
-	/* Program "protection fault". */
+		/* Program "protection fault". */
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+		WREG32_SOC15_OFFSET(
-			    mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+			MMHUB, 0,
-			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
-			    (u32)(adev->dummy_page_addr >> 12));
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
-	WREG32_SOC15_OFFSET(MMHUB, 0,
+			(u32)(adev->dummy_page_addr >> 12));
-			    mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		WREG32_SOC15_OFFSET(
-			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			MMHUB, 0,
-			    (u32)((u64)adev->dummy_page_addr >> 44));
+			mmVML2PF0_VM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+			(u32)((u64)adev->dummy_page_addr >> 44));
 
-	tmp = RREG32_SOC15_OFFSET(MMHUB, 0,
+		tmp = RREG32_SOC15_OFFSET(
-				  mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+			MMHUB, 0, mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
-				  hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
+			hubid * MMHUB_INSTANCE_REGISTER_OFFSET);
-	tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
-			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+				    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
-	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+		WREG32_SOC15_OFFSET(MMHUB, 0,
-			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
+				    mmVML2PF0_VM_L2_PROTECTION_FAULT_CNTL2,
+				    hubid * MMHUB_INSTANCE_REGISTER_OFFSET,
+				    tmp);
+	}
 }
 
 static void mmhub_v9_4_init_tlb_regs(struct amdgpu_device *adev, int hubid)
@@ -368,30 +376,16 @@ int mmhub_v9_4_gart_enable(struct amdgpu_device *adev)
 	int i;
 
 	for (i = 0; i < MMHUB_NUM_INSTANCES; i++) {
-		if (amdgpu_sriov_vf(adev)) {
-			/*
-			 * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase
-			 * they are VF copy registers so vbios post doesn't
-			 * program them, for SRIOV driver need to program them
-			 */
-			WREG32_SOC15_OFFSET(MMHUB, 0,
-				     mmVMSHAREDVC0_MC_VM_FB_LOCATION_BASE,
-				     i * MMHUB_INSTANCE_REGISTER_OFFSET,
-				     adev->gmc.vram_start >> 24);
-			WREG32_SOC15_OFFSET(MMHUB, 0,
-				     mmVMSHAREDVC0_MC_VM_FB_LOCATION_TOP,
-				     i * MMHUB_INSTANCE_REGISTER_OFFSET,
-				     adev->gmc.vram_end >> 24);
-		}
-
 		/* GART Enable. */
 		mmhub_v9_4_init_gart_aperture_regs(adev, i);
 		mmhub_v9_4_init_system_aperture_regs(adev, i);
 		mmhub_v9_4_init_tlb_regs(adev, i);
-		mmhub_v9_4_init_cache_regs(adev, i);
+		if (!amdgpu_sriov_vf(adev))
+			mmhub_v9_4_init_cache_regs(adev, i);
 
 		mmhub_v9_4_enable_system_domain(adev, i);
-		mmhub_v9_4_disable_identity_aperture(adev, i);
+		if (!amdgpu_sriov_vf(adev))
+			mmhub_v9_4_disable_identity_aperture(adev, i);
 		mmhub_v9_4_setup_vmid_config(adev, i);
 		mmhub_v9_4_program_invalidation(adev, i);
 	}

drivers/gpu/drm/amd/amdgpu/mmsch_v1_0.h

@@ -47,6 +47,18 @@ struct mmsch_v1_0_init_header {
 	uint32_t uvd_table_size;
 };
 
+struct mmsch_vf_eng_init_header {
+	uint32_t init_status;
+	uint32_t table_offset;
+	uint32_t table_size;
+};
+
+struct mmsch_v1_1_init_header {
+	uint32_t version;
+	uint32_t total_size;
+	struct mmsch_vf_eng_init_header eng[2];
+};
+
 struct mmsch_v1_0_cmd_direct_reg_header {
 	uint32_t reg_offset   : 28;
 	uint32_t command_type : 4;

drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c

@@ -250,7 +250,7 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
 	 */
 	locked = mutex_trylock(&adev->lock_reset);
 	if (locked)
-		adev->in_gpu_reset = 1;
+		adev->in_gpu_reset = true;
 
 	do {
 		if (xgpu_ai_mailbox_peek_msg(adev) == IDH_FLR_NOTIFICATION_CMPL)
@@ -262,7 +262,7 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
 
 flr_done:
 	if (locked) {
-		adev->in_gpu_reset = 0;
+		adev->in_gpu_reset = false;
 		mutex_unlock(&adev->lock_reset);
 	}
 

drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c

@@ -252,7 +252,7 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct *work)
 	 */
 	locked = mutex_trylock(&adev->lock_reset);
 	if (locked)
-		adev->in_gpu_reset = 1;
+		adev->in_gpu_reset = true;
 
 	do {
 		if (xgpu_nv_mailbox_peek_msg(adev) == IDH_FLR_NOTIFICATION_CMPL)
@@ -264,12 +264,16 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct *work)
 
 flr_done:
 	if (locked) {
-		adev->in_gpu_reset = 0;
+		adev->in_gpu_reset = false;
 		mutex_unlock(&adev->lock_reset);
 	}
 
 	/* Trigger recovery for world switch failure if no TDR */
-	if (amdgpu_device_should_recover_gpu(adev))
+	if (amdgpu_device_should_recover_gpu(adev)
+		&& (adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->gfx_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->compute_timeout == MAX_SCHEDULE_TIMEOUT ||
+		adev->video_timeout == MAX_SCHEDULE_TIMEOUT))
 		amdgpu_device_gpu_recover(adev, NULL);
 }
 

drivers/gpu/drm/amd/amdgpu/navi10_ih.c

@@ -110,7 +110,6 @@ static uint32_t navi10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl
 static int navi10_ih_irq_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ih_ring *ih = &adev->irq.ih;
-	int ret = 0;
 	u32 ih_rb_cntl, ih_doorbell_rtpr, ih_chicken;
 	u32 tmp;
 
@@ -179,7 +178,7 @@ static int navi10_ih_irq_init(struct amdgpu_device *adev)
 	/* enable interrupts */
 	navi10_ih_enable_interrupts(adev);
 
-	return ret;
+	return 0;
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c

@@ -339,7 +339,7 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device
 		/* ras_controller_int is dedicated for nbif ras error,
 		 * not the global interrupt for sync flood
 		 */
-		amdgpu_ras_reset_gpu(adev, true);
+		amdgpu_ras_reset_gpu(adev);
 	}
 }
 
@@ -456,10 +456,8 @@ static int nbio_v7_4_init_ras_controller_interrupt (struct amdgpu_device *adev)
 	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
 			      NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT,
 			      &adev->nbio.ras_controller_irq);
-	if (r)
-		return r;
 
-	return 0;
+	return r;
 }
 
 static int nbio_v7_4_init_ras_err_event_athub_interrupt (struct amdgpu_device *adev)
@@ -476,10 +474,8 @@ static int nbio_v7_4_init_ras_err_event_athub_interrupt (struct amdgpu_device *a
 	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
 			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
 			      &adev->nbio.ras_err_event_athub_irq);
-	if (r)
-		return r;
 
-	return 0;
+	return r;
 }
 
 #define smnPARITY_ERROR_STATUS_UNCORR_GRP2	0x13a20030

drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h

@@ -242,6 +242,7 @@ enum psp_gfx_fw_type {
 	GFX_FW_TYPE_SDMA5                           = 55,   /* SDMA5                    MI      */
 	GFX_FW_TYPE_SDMA6                           = 56,   /* SDMA6                    MI      */
 	GFX_FW_TYPE_SDMA7                           = 57,   /* SDMA7                    MI      */
+	GFX_FW_TYPE_VCN1                            = 58,   /* VCN1                     MI      */
 	GFX_FW_TYPE_MAX
 };
 

drivers/gpu/drm/amd/amdgpu/psp_v11_0.c

@@ -233,6 +233,29 @@ out:
 	return err;
 }
 
+int psp_v11_0_wait_for_bootloader(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+
+	int ret;
+	int retry_loop;
+
+	for (retry_loop = 0; retry_loop < 10; retry_loop++) {
+		/* Wait for bootloader to signify that is
+		    ready having bit 31 of C2PMSG_35 set to 1 */
+		ret = psp_wait_for(psp,
+				   SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+				   0x80000000,
+				   0x80000000,
+				   false);
+
+		if (ret == 0)
+			return 0;
+	}
+
+	return ret;
+}
+
 static bool psp_v11_0_is_sos_alive(struct psp_context *psp)
 {
 	struct amdgpu_device *adev = psp->adev;
@@ -258,9 +281,7 @@ static int psp_v11_0_bootloader_load_kdb(struct psp_context *psp)
 		return 0;
 	}
 
-	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_v11_0_wait_for_bootloader(psp);
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
 	if (ret)
 		return ret;
 
@@ -276,9 +297,7 @@ static int psp_v11_0_bootloader_load_kdb(struct psp_context *psp)
 	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
 	       psp_gfxdrv_command_reg);
 
-	/* Wait for bootloader to signify that is ready having  bit 31 of C2PMSG_35 set to 1*/
+	ret = psp_v11_0_wait_for_bootloader(psp);
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
 
 	return ret;
 }
@@ -298,9 +317,7 @@ static int psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp)
 		return 0;
 	}
 
-	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_v11_0_wait_for_bootloader(psp);
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
 	if (ret)
 		return ret;
 
@@ -319,8 +336,7 @@ static int psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp)
 	/* there might be handshake issue with hardware which needs delay */
 	mdelay(20);
 
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+	ret = psp_v11_0_wait_for_bootloader(psp);
-			   0x80000000, 0x80000000, false);
 
 	return ret;
 }
@@ -337,9 +353,7 @@ static int psp_v11_0_bootloader_load_sos(struct psp_context *psp)
 	if (psp_v11_0_is_sos_alive(psp))
 		return 0;
 
-	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_v11_0_wait_for_bootloader(psp);
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			   0x80000000, 0x80000000, false);
 	if (ret)
 		return ret;
 

drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c

@@ -255,7 +255,7 @@ static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v2_4_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v2_4_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
 			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
@@ -750,7 +750,7 @@ static void sdma_v2_4_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1260,16 +1260,14 @@ static const struct amdgpu_vm_pte_funcs sdma_v2_4_vm_pte_funcs = {
 
 static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
-	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
 	adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] =
-		adev->vm_manager.vm_pte_rqs[i] =
+			&adev->sdma.instance[i].ring.sched;
-			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
-	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v2_4_ip_block =

drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c

@@ -429,7 +429,7 @@ static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v3_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v3_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
 			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
@@ -1021,7 +1021,7 @@ static void sdma_v3_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1698,16 +1698,14 @@ static const struct amdgpu_vm_pte_funcs sdma_v3_0_vm_pte_funcs = {
 
 static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
-	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
 	adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] =
-		adev->vm_manager.vm_pte_rqs[i] =
+			 &adev->sdma.instance[i].ring.sched;
-			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
-	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v3_0_ip_block =

drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c

@@ -698,7 +698,7 @@ static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 
 	/* IB packet must end on a 8 DW boundary */
-	sdma_v4_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	sdma_v4_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
 			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
@@ -1579,7 +1579,7 @@ static void sdma_v4_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -2409,10 +2409,9 @@ static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 			sched = &adev->sdma.instance[i].page.sched;
 		else
 			sched = &adev->sdma.instance[i].ring.sched;
-		adev->vm_manager.vm_pte_rqs[i] =
+		adev->vm_manager.vm_pte_scheds[i] = sched;
-			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
-	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version sdma_v4_0_ip_block = {

drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c

@@ -382,8 +382,15 @@ static void sdma_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	uint64_t csa_mc_addr = amdgpu_sdma_get_csa_mc_addr(ring, vmid);
 
-	/* IB packet must end on a 8 DW boundary */
+	/* An IB packet must end on a 8 DW boundary--the next dword
-	sdma_v5_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+	 * must be on a 8-dword boundary. Our IB packet below is 6
+	 * dwords long, thus add x number of NOPs, such that, in
+	 * modular arithmetic,
+	 * wptr + 6 + x = 8k, k >= 0, which in C is,
+	 * (wptr + 6 + x) % 8 = 0.
+	 * The expression below, is a solution of x.
+	 */
+	sdma_v5_0_ring_insert_nop(ring, (2 - lower_32_bits(ring->wptr)) & 7);
 
 	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
 			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
@@ -907,16 +914,9 @@ static int sdma_v5_0_ring_test_ring(struct amdgpu_ring *ring)
 			udelay(1);
 	}
 
-	if (i < adev->usec_timeout) {
+	if (i >= adev->usec_timeout)
-		if (amdgpu_emu_mode == 1)
+		r = -ETIMEDOUT;
-			DRM_INFO("ring test on %d succeeded in %d msecs\n", ring->idx, i);
+
-		else
-			DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
-	} else {
-		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
-			  ring->idx, tmp);
-		r = -EINVAL;
-	}
 	amdgpu_device_wb_free(adev, index);
 
 	return r;
@@ -981,13 +981,10 @@ static int sdma_v5_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 		goto err1;
 	}
 	tmp = le32_to_cpu(adev->wb.wb[index]);
-	if (tmp == 0xDEADBEEF) {
+	if (tmp == 0xDEADBEEF)
-		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
 		r = 0;
-	} else {
+	else
-		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
 		r = -EINVAL;
-	}
 
 err1:
 	amdgpu_ib_free(adev, &ib, NULL);
@@ -1086,10 +1083,10 @@ static void sdma_v5_0_vm_set_pte_pde(struct amdgpu_ib *ib,
 }
 
 /**
- * sdma_v5_0_ring_pad_ib - pad the IB to the required number of dw
+ * sdma_v5_0_ring_pad_ib - pad the IB
- *
  * @ib: indirect buffer to fill with padding
  *
+ * Pad the IB with NOPs to a boundary multiple of 8.
  */
 static void sdma_v5_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
@@ -1097,7 +1094,7 @@ static void sdma_v5_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib
 	u32 pad_count;
 	int i;
 
-	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	pad_count = (-ib->length_dw) & 0x7;
 	for (i = 0; i < pad_count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
 			ib->ptr[ib->length_dw++] =
@@ -1721,17 +1718,15 @@ static const struct amdgpu_vm_pte_funcs sdma_v5_0_vm_pte_funcs = {
 
 static void sdma_v5_0_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
-	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
 	if (adev->vm_manager.vm_pte_funcs == NULL) {
 		adev->vm_manager.vm_pte_funcs = &sdma_v5_0_vm_pte_funcs;
 		for (i = 0; i < adev->sdma.num_instances; i++) {
-			sched = &adev->sdma.instance[i].ring.sched;
+			adev->vm_manager.vm_pte_scheds[i] =
-			adev->vm_manager.vm_pte_rqs[i] =
+				&adev->sdma.instance[i].ring.sched;
-				&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 		}
-		adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+		adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 	}
 }
 

drivers/gpu/drm/amd/amdgpu/si_dma.c

@@ -834,16 +834,14 @@ static const struct amdgpu_vm_pte_funcs si_dma_vm_pte_funcs = {
 
 static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev)
 {
-	struct drm_gpu_scheduler *sched;
 	unsigned i;
 
 	adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		sched = &adev->sdma.instance[i].ring.sched;
+		adev->vm_manager.vm_pte_scheds[i] =
-		adev->vm_manager.vm_pte_rqs[i] =
+			&adev->sdma.instance[i].ring.sched;
-			&sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL];
 	}
-	adev->vm_manager.vm_pte_num_rqs = adev->sdma.num_instances;
+	adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
 }
 
 const struct amdgpu_ip_block_version si_dma_ip_block =

drivers/gpu/drm/amd/amdgpu/soc15.c

@@ -613,6 +613,7 @@ static bool soc15_supports_baco(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
 	case CHIP_VEGA12:
+	case CHIP_ARCTURUS:
 		soc15_asic_get_baco_capability(adev, &baco_support);
 		break;
 	case CHIP_VEGA20:
@@ -827,11 +828,15 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev)
 			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
 		amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
 		amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
-		if (!amdgpu_sriov_vf(adev))
+		amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
-			amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
 
-		if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT))
+		if (amdgpu_sriov_vf(adev)) {
-			amdgpu_device_ip_block_add(adev, &vcn_v2_5_ip_block);
+			if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP))
+				amdgpu_device_ip_block_add(adev, &vcn_v2_5_ip_block);
+		} else {
+			if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT))
+				amdgpu_device_ip_block_add(adev, &vcn_v2_5_ip_block);
+		}
 		if (!amdgpu_sriov_vf(adev))
 			amdgpu_device_ip_block_add(adev, &jpeg_v2_5_ip_block);
 		break;

drivers/gpu/drm/amd/amdgpu/umc_v6_1.c

@@ -28,27 +28,26 @@
 #include "rsmu/rsmu_0_0_2_sh_mask.h"
 #include "umc/umc_6_1_1_offset.h"
 #include "umc/umc_6_1_1_sh_mask.h"
+#include "umc/umc_6_1_2_offset.h"
 
 #define smnMCA_UMC0_MCUMC_ADDRT0	0x50f10
 
-/* UMC 6_1_2 register offsets */
+#define UMC_6_INST_DIST			0x40000
-#define mmUMCCH0_0_EccErrCntSel_ARCT                 0x0360
-#define mmUMCCH0_0_EccErrCntSel_ARCT_BASE_IDX        1
-#define mmUMCCH0_0_EccErrCnt_ARCT                    0x0361
-#define mmUMCCH0_0_EccErrCnt_ARCT_BASE_IDX           1
-#define mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT           0x03c2
-#define mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT_BASE_IDX  1
 
 /*
  * (addr / 256) * 8192, the higher 26 bits in ErrorAddr
  * is the index of 8KB block
  */
-#define ADDR_OF_8KB_BLOCK(addr)		(((addr) & ~0xffULL) << 5)
+#define ADDR_OF_8KB_BLOCK(addr)			(((addr) & ~0xffULL) << 5)
 /* channel index is the index of 256B block */
 #define ADDR_OF_256B_BLOCK(channel_index)	((channel_index) << 8)
 /* offset in 256B block */
 #define OFFSET_IN_256B_BLOCK(addr)		((addr) & 0xffULL)
 
+#define LOOP_UMC_INST(umc_inst) for ((umc_inst) = 0; (umc_inst) < adev->umc.umc_inst_num; (umc_inst)++)
+#define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++)
+#define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst))
+
 const uint32_t
 	umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM] = {
 		{2, 18, 11, 27},	{4, 20, 13, 29},
@@ -57,41 +56,17 @@ const uint32_t
 		{9, 25, 0, 16},		{15, 31, 6, 22}
 };
 
-static void umc_v6_1_enable_umc_index_mode(struct amdgpu_device *adev,
-					   uint32_t umc_instance)
-{
-	uint32_t rsmu_umc_index;
-
-	rsmu_umc_index = RREG32_SOC15(RSMU, 0,
-			mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU);
-	rsmu_umc_index = REG_SET_FIELD(rsmu_umc_index,
-			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
-			RSMU_UMC_INDEX_MODE_EN, 1);
-	rsmu_umc_index = REG_SET_FIELD(rsmu_umc_index,
-			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
-			RSMU_UMC_INDEX_INSTANCE, umc_instance);
-	rsmu_umc_index = REG_SET_FIELD(rsmu_umc_index,
-			RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
-			RSMU_UMC_INDEX_WREN, 1 << umc_instance);
-	WREG32_SOC15(RSMU, 0, mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
-				rsmu_umc_index);
-}
-
 static void umc_v6_1_disable_umc_index_mode(struct amdgpu_device *adev)
 {
 	WREG32_FIELD15(RSMU, 0, RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
 			RSMU_UMC_INDEX_MODE_EN, 0);
 }
 
-static uint32_t umc_v6_1_get_umc_inst(struct amdgpu_device *adev)
+static inline uint32_t get_umc_6_reg_offset(struct amdgpu_device *adev,
+					    uint32_t umc_inst,
+					    uint32_t ch_inst)
 {
-	uint32_t rsmu_umc_index;
+	return adev->umc.channel_offs*ch_inst + UMC_6_INST_DIST*umc_inst;
-
-	rsmu_umc_index = RREG32_SOC15(RSMU, 0,
-				mmRSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU);
-	return REG_GET_FIELD(rsmu_umc_index,
-				RSMU_UMC_INDEX_REGISTER_NBIF_VG20_GPU,
-				RSMU_UMC_INDEX_INSTANCE);
 }
 
 static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
@@ -105,7 +80,6 @@ static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
 
 	if (adev->asic_type == CHIP_ARCTURUS) {
 		/* UMC 6_1_2 registers */
-
 		ecc_err_cnt_sel_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel_ARCT);
 		ecc_err_cnt_addr =
@@ -114,7 +88,6 @@ static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
 			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
 	} else {
 		/* UMC 6_1_1 registers */
-
 		ecc_err_cnt_sel_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel);
 		ecc_err_cnt_addr =
@@ -124,31 +97,31 @@ static void umc_v6_1_query_correctable_error_count(struct amdgpu_device *adev,
 	}
 
 	/* select the lower chip and check the error count */
-	ecc_err_cnt_sel = RREG32(ecc_err_cnt_sel_addr + umc_reg_offset);
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
 	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
 					EccErrCntCsSel, 0);
-	WREG32(ecc_err_cnt_sel_addr + umc_reg_offset, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
-	ecc_err_cnt = RREG32(ecc_err_cnt_addr + umc_reg_offset);
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
 	*error_count +=
 		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
 		 UMC_V6_1_CE_CNT_INIT);
 	/* clear the lower chip err count */
-	WREG32(ecc_err_cnt_addr + umc_reg_offset, UMC_V6_1_CE_CNT_INIT);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
 
 	/* select the higher chip and check the err counter */
 	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
 					EccErrCntCsSel, 1);
-	WREG32(ecc_err_cnt_sel_addr + umc_reg_offset, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
-	ecc_err_cnt = RREG32(ecc_err_cnt_addr + umc_reg_offset);
+	ecc_err_cnt = RREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4);
 	*error_count +=
 		(REG_GET_FIELD(ecc_err_cnt, UMCCH0_0_EccErrCnt, EccErrCnt) -
 		 UMC_V6_1_CE_CNT_INIT);
 	/* clear the higher chip err count */
-	WREG32(ecc_err_cnt_addr + umc_reg_offset, UMC_V6_1_CE_CNT_INIT);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
 
 	/* check for SRAM correctable error
 	  MCUMC_STATUS is a 64 bit register */
-	mc_umc_status = RREG64_UMC(mc_umc_status_addr + umc_reg_offset);
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
 	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, ErrorCodeExt) == 6 &&
 	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
 	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, CECC) == 1)
@@ -164,18 +137,16 @@ static void umc_v6_1_querry_uncorrectable_error_count(struct amdgpu_device *adev
 
 	if (adev->asic_type == CHIP_ARCTURUS) {
 		/* UMC 6_1_2 registers */
-
 		mc_umc_status_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
 	} else {
 		/* UMC 6_1_1 registers */
-
 		mc_umc_status_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
 	}
 
 	/* check the MCUMC_STATUS */
-	mc_umc_status = RREG64_UMC(mc_umc_status_addr + umc_reg_offset);
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
 	if ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
 	    (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1 ||
 	    REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UECC) == 1 ||
@@ -185,38 +156,46 @@ static void umc_v6_1_querry_uncorrectable_error_count(struct amdgpu_device *adev
 		*error_count += 1;
 }
 
-static void umc_v6_1_query_error_count(struct amdgpu_device *adev,
-					   struct ras_err_data *err_data, uint32_t umc_reg_offset,
-					   uint32_t channel_index)
-{
-	umc_v6_1_query_correctable_error_count(adev, umc_reg_offset,
-						   &(err_data->ce_count));
-	umc_v6_1_querry_uncorrectable_error_count(adev, umc_reg_offset,
-						  &(err_data->ue_count));
-}
-
 static void umc_v6_1_query_ras_error_count(struct amdgpu_device *adev,
 					   void *ras_error_status)
 {
-	amdgpu_umc_for_each_channel(umc_v6_1_query_error_count);
+	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;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_query_correctable_error_count(adev,
+						       umc_reg_offset,
+						       &(err_data->ce_count));
+		umc_v6_1_querry_uncorrectable_error_count(adev,
+							  umc_reg_offset,
+							  &(err_data->ue_count));
+	}
 }
 
 static void umc_v6_1_query_error_address(struct amdgpu_device *adev,
 					 struct ras_err_data *err_data,
-					 uint32_t umc_reg_offset, uint32_t channel_index)
+					 uint32_t umc_reg_offset,
+					 uint32_t ch_inst,
+					 uint32_t umc_inst)
 {
 	uint32_t lsb, mc_umc_status_addr;
 	uint64_t mc_umc_status, err_addr, retired_page;
 	struct eeprom_table_record *err_rec;
+	uint32_t channel_index = adev->umc.channel_idx_tbl[umc_inst * adev->umc.channel_inst_num + ch_inst];
 
 	if (adev->asic_type == CHIP_ARCTURUS) {
 		/* UMC 6_1_2 registers */
-
 		mc_umc_status_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0_ARCT);
 	} else {
 		/* UMC 6_1_1 registers */
-
 		mc_umc_status_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmMCA_UMC_UMC0_MCUMC_STATUST0);
 	}
@@ -224,12 +203,12 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev,
 	/* skip error address process if -ENOMEM */
 	if (!err_data->err_addr) {
 		/* clear umc status */
-		WREG64_UMC(mc_umc_status_addr + umc_reg_offset, 0x0ULL);
+		WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
 		return;
 	}
 
 	err_rec = &err_data->err_addr[err_data->err_addr_cnt];
-	mc_umc_status = RREG64_UMC(mc_umc_status_addr + umc_reg_offset);
+	mc_umc_status = RREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4);
 
 	/* calculate error address if ue/ce error is detected */
 	if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
@@ -257,39 +236,53 @@ static void umc_v6_1_query_error_address(struct amdgpu_device *adev,
 			err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE;
 			err_rec->cu = 0;
 			err_rec->mem_channel = channel_index;
-			err_rec->mcumc_id = umc_v6_1_get_umc_inst(adev);
+			err_rec->mcumc_id = umc_inst;
 
 			err_data->err_addr_cnt++;
 		}
 	}
 
 	/* clear umc status */
-	WREG64_UMC(mc_umc_status_addr + umc_reg_offset, 0x0ULL);
+	WREG64_PCIE((mc_umc_status_addr + umc_reg_offset) * 4, 0x0ULL);
 }
 
 static void umc_v6_1_query_ras_error_address(struct amdgpu_device *adev,
 					     void *ras_error_status)
 {
-	amdgpu_umc_for_each_channel(umc_v6_1_query_error_address);
+	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;
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_query_error_address(adev,
+					     err_data,
+					     umc_reg_offset,
+					     ch_inst,
+					     umc_inst);
+	}
+
 }
 
 static void umc_v6_1_err_cnt_init_per_channel(struct amdgpu_device *adev,
-					 struct ras_err_data *err_data,
+					      uint32_t umc_reg_offset)
-					 uint32_t umc_reg_offset, uint32_t channel_index)
 {
 	uint32_t ecc_err_cnt_sel, ecc_err_cnt_sel_addr;
 	uint32_t ecc_err_cnt_addr;
 
 	if (adev->asic_type == CHIP_ARCTURUS) {
 		/* UMC 6_1_2 registers */
-
 		ecc_err_cnt_sel_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel_ARCT);
 		ecc_err_cnt_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCnt_ARCT);
 	} else {
 		/* UMC 6_1_1 registers */
-
 		ecc_err_cnt_sel_addr =
 			SOC15_REG_OFFSET(UMC, 0, mmUMCCH0_0_EccErrCntSel);
 		ecc_err_cnt_addr =
@@ -297,28 +290,38 @@ static void umc_v6_1_err_cnt_init_per_channel(struct amdgpu_device *adev,
 	}
 
 	/* select the lower chip and check the error count */
-	ecc_err_cnt_sel = RREG32(ecc_err_cnt_sel_addr + umc_reg_offset);
+	ecc_err_cnt_sel = RREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4);
 	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
 					EccErrCntCsSel, 0);
 	/* set ce error interrupt type to APIC based interrupt */
 	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
 					EccErrInt, 0x1);
-	WREG32(ecc_err_cnt_sel_addr + umc_reg_offset, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
 	/* set error count to initial value */
-	WREG32(ecc_err_cnt_addr + umc_reg_offset, UMC_V6_1_CE_CNT_INIT);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
 
 	/* select the higher chip and check the err counter */
 	ecc_err_cnt_sel = REG_SET_FIELD(ecc_err_cnt_sel, UMCCH0_0_EccErrCntSel,
 					EccErrCntCsSel, 1);
-	WREG32(ecc_err_cnt_sel_addr + umc_reg_offset, ecc_err_cnt_sel);
+	WREG32_PCIE((ecc_err_cnt_sel_addr + umc_reg_offset) * 4, ecc_err_cnt_sel);
-	WREG32(ecc_err_cnt_addr + umc_reg_offset, UMC_V6_1_CE_CNT_INIT);
+	WREG32_PCIE((ecc_err_cnt_addr + umc_reg_offset) * 4, UMC_V6_1_CE_CNT_INIT);
 }
 
 static void umc_v6_1_err_cnt_init(struct amdgpu_device *adev)
 {
-	void *ras_error_status = NULL;
+	uint32_t umc_inst        = 0;
+	uint32_t ch_inst         = 0;
+	uint32_t umc_reg_offset  = 0;
 
-	amdgpu_umc_for_each_channel(umc_v6_1_err_cnt_init_per_channel);
+	umc_v6_1_disable_umc_index_mode(adev);
+
+	LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
+		umc_reg_offset = get_umc_6_reg_offset(adev,
+						      umc_inst,
+						      ch_inst);
+
+		umc_v6_1_err_cnt_init_per_channel(adev, umc_reg_offset);
+	}
 }
 
 const struct amdgpu_umc_funcs umc_v6_1_funcs = {
@@ -326,6 +329,4 @@ const struct amdgpu_umc_funcs umc_v6_1_funcs = {
 	.ras_late_init = amdgpu_umc_ras_late_init,
 	.query_ras_error_count = umc_v6_1_query_ras_error_count,
 	.query_ras_error_address = umc_v6_1_query_ras_error_address,
-	.enable_umc_index_mode = umc_v6_1_enable_umc_index_mode,
-	.disable_umc_index_mode = umc_v6_1_disable_umc_index_mode,
 };

drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c

@@ -25,6 +25,7 @@
 
 #include "amdgpu.h"
 #include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
 #include "soc15.h"
 #include "soc15d.h"
 #include "soc15_common.h"
@@ -51,6 +52,8 @@ static int vcn_v1_0_set_powergating_state(void *handle, enum amd_powergating_sta
 static int vcn_v1_0_pause_dpg_mode(struct amdgpu_device *adev,
 				struct dpg_pause_state *new_state);
 
+static void vcn_v1_0_idle_work_handler(struct work_struct *work);
+
 /**
  * vcn_v1_0_early_init - set function pointers
  *
@@ -105,6 +108,9 @@ static int vcn_v1_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	/* Override the work func */
+	adev->vcn.idle_work.work.func = vcn_v1_0_idle_work_handler;
+
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
 		const struct common_firmware_header *hdr;
 		hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
@@ -1758,6 +1764,86 @@ static int vcn_v1_0_set_powergating_state(void *handle,
 	return ret;
 }
 
+static void vcn_v1_0_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, vcn.idle_work.work);
+	unsigned int fences = 0, i;
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i)
+		fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_enc[i]);
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		struct dpg_pause_state new_state;
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(&adev->jpeg.inst->ring_dec))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		adev->vcn.pause_dpg_mode(adev, &new_state);
+	}
+
+	fences += amdgpu_fence_count_emitted(&adev->jpeg.inst->ring_dec);
+	fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_dec);
+
+	if (fences == 0) {
+		amdgpu_gfx_off_ctrl(adev, true);
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_uvd(adev, false);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+			       AMD_PG_STATE_GATE);
+	} else {
+		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+	}
+}
+
+void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
+
+	if (set_clocks) {
+		amdgpu_gfx_off_ctrl(adev, false);
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_uvd(adev, true);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+			       AMD_PG_STATE_UNGATE);
+	}
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+		struct dpg_pause_state new_state;
+		unsigned int fences = 0, i;
+
+		for (i = 0; i < adev->vcn.num_enc_rings; ++i)
+			fences += amdgpu_fence_count_emitted(&adev->vcn.inst->ring_enc[i]);
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(&adev->jpeg.inst->ring_dec))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+
+		adev->vcn.pause_dpg_mode(adev, &new_state);
+	}
+}
+
 static const struct amd_ip_funcs vcn_v1_0_ip_funcs = {
 	.name = "vcn_v1_0",
 	.early_init = vcn_v1_0_early_init,
@@ -1804,7 +1890,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
 	.insert_start = vcn_v1_0_dec_ring_insert_start,
 	.insert_end = vcn_v1_0_dec_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
+	.begin_use = vcn_v1_0_ring_begin_use,
 	.end_use = amdgpu_vcn_ring_end_use,
 	.emit_wreg = vcn_v1_0_dec_ring_emit_wreg,
 	.emit_reg_wait = vcn_v1_0_dec_ring_emit_reg_wait,
@@ -1836,7 +1922,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
 	.insert_nop = amdgpu_ring_insert_nop,
 	.insert_end = vcn_v1_0_enc_ring_insert_end,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
-	.begin_use = amdgpu_vcn_ring_begin_use,
+	.begin_use = vcn_v1_0_ring_begin_use,
 	.end_use = amdgpu_vcn_ring_end_use,
 	.emit_wreg = vcn_v1_0_enc_ring_emit_wreg,
 	.emit_reg_wait = vcn_v1_0_enc_ring_emit_reg_wait,

drivers/gpu/drm/amd/amdgpu/vcn_v1_0.h

@@ -24,6 +24,8 @@
 #ifndef __VCN_V1_0_H__
 #define __VCN_V1_0_H__
 
+void vcn_v1_0_ring_begin_use(struct amdgpu_ring *ring);
+
 extern const struct amdgpu_ip_block_version vcn_v1_0_ip_block;
 
 #endif

drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c

@@ -29,6 +29,7 @@
 #include "soc15.h"
 #include "soc15d.h"
 #include "vcn_v2_0.h"
+#include "mmsch_v1_0.h"
 
 #include "vcn/vcn_2_5_offset.h"
 #include "vcn/vcn_2_5_sh_mask.h"
@@ -54,6 +55,7 @@ static void vcn_v2_5_set_enc_ring_funcs(struct amdgpu_device *adev);
 static void vcn_v2_5_set_irq_funcs(struct amdgpu_device *adev);
 static int vcn_v2_5_set_powergating_state(void *handle,
 				enum amd_powergating_state state);
+static int vcn_v2_5_sriov_start(struct amdgpu_device *adev);
 
 static int amdgpu_ih_clientid_vcns[] = {
 	SOC15_IH_CLIENTID_VCN,
@@ -88,7 +90,13 @@ static int vcn_v2_5_early_init(void *handle)
 	} else
 		adev->vcn.num_vcn_inst = 1;
 
-	adev->vcn.num_enc_rings = 2;
+	if (amdgpu_sriov_vf(adev)) {
+		adev->vcn.num_vcn_inst = 2;
+		adev->vcn.harvest_config = 0;
+		adev->vcn.num_enc_rings = 1;
+	} else {
+		adev->vcn.num_enc_rings = 2;
+	}
 
 	vcn_v2_5_set_dec_ring_funcs(adev);
 	vcn_v2_5_set_enc_ring_funcs(adev);
@@ -176,7 +184,9 @@ static int vcn_v2_5_sw_init(void *handle)
 
 		ring = &adev->vcn.inst[j].ring_dec;
 		ring->use_doorbell = true;
-		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8*j;
+
+		ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+				(amdgpu_sriov_vf(adev) ? 2*j : 8*j);
 		sprintf(ring->name, "vcn_dec_%d", j);
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq, 0);
 		if (r)
@@ -185,7 +195,10 @@ static int vcn_v2_5_sw_init(void *handle)
 		for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
 			ring = &adev->vcn.inst[j].ring_enc[i];
 			ring->use_doorbell = true;
-			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + i + 8*j;
+
+			ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					(amdgpu_sriov_vf(adev) ? (1 + i + 2*j) : (2 + i + 8*j));
+
 			sprintf(ring->name, "vcn_enc_%d.%d", j, i);
 			r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[j].irq, 0);
 			if (r)
@@ -193,6 +206,12 @@ static int vcn_v2_5_sw_init(void *handle)
 		}
 	}
 
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_alloc_mm_table(adev);
+		if (r)
+			return r;
+	}
+
 	return 0;
 }
 
@@ -208,6 +227,9 @@ static int vcn_v2_5_sw_fini(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_free_mm_table(adev);
+
 	r = amdgpu_vcn_suspend(adev);
 	if (r)
 		return r;
@@ -228,25 +250,37 @@ static int vcn_v2_5_hw_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	struct amdgpu_ring *ring;
-	int i, j, r;
+	int i, j, r = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		r = vcn_v2_5_sriov_start(adev);
 
 	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
 		if (adev->vcn.harvest_config & (1 << j))
 			continue;
-		ring = &adev->vcn.inst[j].ring_dec;
 
-		adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+		if (amdgpu_sriov_vf(adev)) {
+			adev->vcn.inst[j].ring_enc[0].sched.ready = true;
+			adev->vcn.inst[j].ring_enc[1].sched.ready = false;
+			adev->vcn.inst[j].ring_enc[2].sched.ready = false;
+			adev->vcn.inst[j].ring_dec.sched.ready = true;
+		} else {
+
+			ring = &adev->vcn.inst[j].ring_dec;
+
+			adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
 						     ring->doorbell_index, j);
 
-		r = amdgpu_ring_test_helper(ring);
-		if (r)
-			goto done;
-
-		for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
-			ring = &adev->vcn.inst[j].ring_enc[i];
 			r = amdgpu_ring_test_helper(ring);
 			if (r)
 				goto done;
+
+			for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+				ring = &adev->vcn.inst[j].ring_enc[i];
+				r = amdgpu_ring_test_helper(ring);
+				if (r)
+					goto done;
+			}
 		}
 	}
 
@@ -741,6 +775,204 @@ static int vcn_v2_5_start(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int vcn_v2_5_mmsch_start(struct amdgpu_device *adev,
+				struct amdgpu_mm_table *table)
+{
+	uint32_t data = 0, loop = 0, size = 0;
+	uint64_t addr = table->gpu_addr;
+	struct mmsch_v1_1_init_header *header = NULL;;
+
+	header = (struct mmsch_v1_1_init_header *)table->cpu_addr;
+	size = header->total_size;
+
+	/*
+	 * 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr of
+	 *  memory descriptor location
+	 */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_LO, lower_32_bits(addr));
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_HI, upper_32_bits(addr));
+
+	/* 2, update vmid of descriptor */
+	data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID);
+	data &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	/* use domain0 for MM scheduler */
+	data |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID, data);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP, 0);
+
+	/*
+	 * 5, kick off the initialization and wait until
+	 * VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero
+	 */
+	WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_HOST, 0x10000001);
+
+	data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+	loop = 10;
+	while ((data & 0x10000002) != 0x10000002) {
+		udelay(100);
+		data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+		loop--;
+		if (!loop)
+			break;
+	}
+
+	if (!loop) {
+		dev_err(adev->dev,
+			"failed to init MMSCH, mmMMSCH_VF_MAILBOX_RESP = %x\n",
+			data);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int vcn_v2_5_sriov_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint32_t offset, size, tmp, i, rb_bufsz;
+	uint32_t table_size = 0;
+	struct mmsch_v1_0_cmd_direct_write direct_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_direct_read_modify_write direct_rd_mod_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_direct_polling direct_poll = { { 0 } };
+	struct mmsch_v1_0_cmd_end end = { { 0 } };
+	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+	struct mmsch_v1_1_init_header *header = (struct mmsch_v1_1_init_header *)init_table;
+
+	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	direct_poll.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_POLLING;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	header->version = MMSCH_VERSION;
+	header->total_size = sizeof(struct mmsch_v1_1_init_header) >> 2;
+	init_table += header->total_size;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		header->eng[i].table_offset = header->total_size;
+		header->eng[i].init_status = 0;
+		header->eng[i].table_size = 0;
+
+		table_size = 0;
+
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+			~UVD_STATUS__UVD_BUSY, UVD_STATUS__UVD_BUSY);
+
+		size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
+		/* mc resume*/
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_lo);
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + i].tmr_mc_addr_hi);
+			offset = 0;
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0), 0);
+		} else {
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+				lower_32_bits(adev->vcn.inst[i].gpu_addr));
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i,
+					mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+				upper_32_bits(adev->vcn.inst[i].gpu_addr));
+			offset = size;
+			MMSCH_V1_0_INSERT_DIRECT_WT(
+				SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0),
+				AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		}
+
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0),
+			size);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[i].gpu_addr + offset));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1),
+			0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1),
+			AMDGPU_VCN_STACK_SIZE);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+			lower_32_bits(adev->vcn.inst[i].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+			upper_32_bits(adev->vcn.inst[i].gpu_addr + offset +
+				AMDGPU_VCN_STACK_SIZE));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2),
+			0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2),
+			AMDGPU_VCN_CONTEXT_SIZE);
+
+		ring = &adev->vcn.inst[i].ring_enc[0];
+		ring->wptr = 0;
+
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO),
+			lower_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI),
+			upper_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE),
+			ring->ring_size / 4);
+
+		ring = &adev->vcn.inst[i].ring_dec;
+		ring->wptr = 0;
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
+			lower_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i,
+				mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
+			upper_32_bits(ring->gpu_addr));
+
+		/* force RBC into idle state */
+		rb_bufsz = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		MMSCH_V1_0_INSERT_DIRECT_WT(
+			SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp);
+
+		/* add end packet */
+		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
+		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+		init_table += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+
+		/* refine header */
+		header->eng[i].table_size = table_size;
+		header->total_size += table_size;
+	}
+
+	return vcn_v2_5_mmsch_start(adev, &adev->virt.mm_table);
+}
+
 static int vcn_v2_5_stop(struct amdgpu_device *adev)
 {
 	uint32_t tmp;
@@ -1048,6 +1280,9 @@ static int vcn_v2_5_set_clockgating_state(void *handle,
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
 
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	if (enable) {
 		if (vcn_v2_5_is_idle(handle))
 			return -EBUSY;
@@ -1065,6 +1300,9 @@ static int vcn_v2_5_set_powergating_state(void *handle,
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	int ret;
 
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	if(state == adev->vcn.cur_state)
 		return 0;
 

drivers/gpu/drm/amd/amdgpu/vega10_ih.c

@@ -234,16 +234,9 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev)
 	WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (ih->gpu_addr >> 40) & 0xff);
 
 	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
-	ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
 	ih_rb_cntl = vega10_ih_rb_cntl(ih, ih_rb_cntl);
-	if (adev->irq.ih.use_bus_addr) {
-		ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
-	} else {
-		ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, MC_SPACE_FBPA_ENABLE, 1);
-	}
 	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM,
 				   !!adev->irq.msi_enabled);
-
 	if (amdgpu_sriov_vf(adev)) {
 		if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL, ih_rb_cntl)) {
 			DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
@@ -253,10 +246,19 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev)
 		WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
 	}
 
-	if ((adev->asic_type == CHIP_ARCTURUS
+	if ((adev->asic_type == CHIP_ARCTURUS &&
-		&& adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)
+	     adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
-		|| adev->asic_type == CHIP_RENOIR)
+	    adev->asic_type == CHIP_RENOIR) {
+		ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
+		if (adev->irq.ih.use_bus_addr) {
+			ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN,
+						   MC_SPACE_GPA_ENABLE, 1);
+		} else {
+			ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN,
+						   MC_SPACE_FBPA_ENABLE, 1);
+		}
 		WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
+	}
 
 	/* set the writeback address whether it's enabled or not */
 	WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO,

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

@@ -42,6 +42,7 @@
 
 static long kfd_ioctl(struct file *, unsigned int, unsigned long);
 static int kfd_open(struct inode *, struct file *);
+static int kfd_release(struct inode *, struct file *);
 static int kfd_mmap(struct file *, struct vm_area_struct *);
 
 static const char kfd_dev_name[] = "kfd";
@@ -51,6 +52,7 @@ static const struct file_operations kfd_fops = {
 	.unlocked_ioctl = kfd_ioctl,
 	.compat_ioctl = compat_ptr_ioctl,
 	.open = kfd_open,
+	.release = kfd_release,
 	.mmap = kfd_mmap,
 };
 
@@ -124,8 +126,13 @@ static int kfd_open(struct inode *inode, struct file *filep)
 	if (IS_ERR(process))
 		return PTR_ERR(process);
 
-	if (kfd_is_locked())
+	if (kfd_is_locked()) {
+		kfd_unref_process(process);
 		return -EAGAIN;
+	}
+
+	/* filep now owns the reference returned by kfd_create_process */
+	filep->private_data = process;
 
 	dev_dbg(kfd_device, "process %d opened, compat mode (32 bit) - %d\n",
 		process->pasid, process->is_32bit_user_mode);
@@ -133,6 +140,16 @@ static int kfd_open(struct inode *inode, struct file *filep)
 	return 0;
 }
 
+static int kfd_release(struct inode *inode, struct file *filep)
+{
+	struct kfd_process *process = filep->private_data;
+
+	if (process)
+		kfd_unref_process(process);
+
+	return 0;
+}
+
 static int kfd_ioctl_get_version(struct file *filep, struct kfd_process *p,
 					void *data)
 {
@@ -1801,9 +1818,14 @@ static long kfd_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
 
 	dev_dbg(kfd_device, "ioctl cmd 0x%x (#0x%x), arg 0x%lx\n", cmd, nr, arg);
 
-	process = kfd_get_process(current);
+	/* Get the process struct from the filep. Only the process
-	if (IS_ERR(process)) {
+	 * that opened /dev/kfd can use the file descriptor. Child
-		dev_dbg(kfd_device, "no process\n");
+	 * processes need to create their own KFD device context.
+	 */
+	process = filep->private_data;
+	if (process->lead_thread != current->group_leader) {
+		dev_dbg(kfd_device, "Using KFD FD in wrong process\n");
+		retcode = -EBADF;
 		goto err_i1;
 	}
 

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

@@ -93,7 +93,7 @@ void kfd_debugfs_init(void)
 			    kfd_debugfs_hqds_by_device, &kfd_debugfs_fops);
 	debugfs_create_file("rls", S_IFREG | 0444, debugfs_root,
 			    kfd_debugfs_rls_by_device, &kfd_debugfs_fops);
-	debugfs_create_file("hang_hws", S_IFREG | 0644, debugfs_root,
+	debugfs_create_file("hang_hws", S_IFREG | 0200, debugfs_root,
 			    NULL, &kfd_debugfs_hang_hws_fops);
 }
 

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

@@ -728,6 +728,9 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd)
 {
 	if (!kfd->init_complete)
 		return 0;
+
+	kfd->dqm->ops.pre_reset(kfd->dqm);
+
 	kgd2kfd_suspend(kfd);
 
 	kfd_signal_reset_event(kfd);
@@ -822,6 +825,21 @@ dqm_start_error:
 	return err;
 }
 
+static inline void kfd_queue_work(struct workqueue_struct *wq,
+				  struct work_struct *work)
+{
+	int cpu, new_cpu;
+
+	cpu = new_cpu = smp_processor_id();
+	do {
+		new_cpu = cpumask_next(new_cpu, cpu_online_mask) % nr_cpu_ids;
+		if (cpu_to_node(new_cpu) == numa_node_id())
+			break;
+	} while (cpu != new_cpu);
+
+	queue_work_on(new_cpu, wq, work);
+}
+
 /* This is called directly from KGD at ISR. */
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 {
@@ -844,7 +862,7 @@ void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 				   patched_ihre, &is_patched)
 	    && enqueue_ih_ring_entry(kfd,
 				     is_patched ? patched_ihre : ih_ring_entry))
-		queue_work(kfd->ih_wq, &kfd->interrupt_work);
+		kfd_queue_work(kfd->ih_wq, &kfd->interrupt_work);
 
 	spin_unlock_irqrestore(&kfd->interrupt_lock, flags);
 }

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

@@ -930,7 +930,6 @@ static void uninitialize(struct device_queue_manager *dqm)
 	for (i = 0 ; i < KFD_MQD_TYPE_MAX ; i++)
 		kfree(dqm->mqd_mgrs[i]);
 	mutex_destroy(&dqm->lock_hidden);
-	kfd_gtt_sa_free(dqm->dev, dqm->pipeline_mem);
 }
 
 static int start_nocpsch(struct device_queue_manager *dqm)
@@ -947,12 +946,19 @@ static int start_nocpsch(struct device_queue_manager *dqm)
 static int stop_nocpsch(struct device_queue_manager *dqm)
 {
 	if (dqm->dev->device_info->asic_family == CHIP_HAWAII)
-		pm_uninit(&dqm->packets);
+		pm_uninit(&dqm->packets, false);
 	dqm->sched_running = false;
 
 	return 0;
 }
 
+static void pre_reset(struct device_queue_manager *dqm)
+{
+	dqm_lock(dqm);
+	dqm->is_resetting = true;
+	dqm_unlock(dqm);
+}
+
 static int allocate_sdma_queue(struct device_queue_manager *dqm,
 				struct queue *q)
 {
@@ -1100,6 +1106,7 @@ static int start_cpsch(struct device_queue_manager *dqm)
 	dqm_lock(dqm);
 	/* clear hang status when driver try to start the hw scheduler */
 	dqm->is_hws_hang = false;
+	dqm->is_resetting = false;
 	dqm->sched_running = true;
 	execute_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0);
 	dqm_unlock(dqm);
@@ -1107,20 +1114,24 @@ static int start_cpsch(struct device_queue_manager *dqm)
 	return 0;
 fail_allocate_vidmem:
 fail_set_sched_resources:
-	pm_uninit(&dqm->packets);
+	pm_uninit(&dqm->packets, false);
 fail_packet_manager_init:
 	return retval;
 }
 
 static int stop_cpsch(struct device_queue_manager *dqm)
 {
+	bool hanging;
+
 	dqm_lock(dqm);
-	unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0);
+	if (!dqm->is_hws_hang)
+		unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0);
+	hanging = dqm->is_hws_hang || dqm->is_resetting;
 	dqm->sched_running = false;
 	dqm_unlock(dqm);
 
 	kfd_gtt_sa_free(dqm->dev, dqm->fence_mem);
-	pm_uninit(&dqm->packets);
+	pm_uninit(&dqm->packets, hanging);
 
 	return 0;
 }
@@ -1352,8 +1363,17 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm,
 	/* should be timed out */
 	retval = amdkfd_fence_wait_timeout(dqm->fence_addr, KFD_FENCE_COMPLETED,
 				queue_preemption_timeout_ms);
-	if (retval)
+	if (retval) {
+		pr_err("The cp might be in an unrecoverable state due to an unsuccessful queues preemption\n");
+		dqm->is_hws_hang = true;
+		/* It's possible we're detecting a HWS hang in the
+		 * middle of a GPU reset. No need to schedule another
+		 * reset in this case.
+		 */
+		if (!dqm->is_resetting)
+			schedule_work(&dqm->hw_exception_work);
 		return retval;
+	}
 
 	pm_release_ib(&dqm->packets);
 	dqm->active_runlist = false;
@@ -1371,12 +1391,8 @@ static int execute_queues_cpsch(struct device_queue_manager *dqm,
 	if (dqm->is_hws_hang)
 		return -EIO;
 	retval = unmap_queues_cpsch(dqm, filter, filter_param);
-	if (retval) {
+	if (retval)
-		pr_err("The cp might be in an unrecoverable state due to an unsuccessful queues preemption\n");
-		dqm->is_hws_hang = true;
-		schedule_work(&dqm->hw_exception_work);
 		return retval;
-	}
 
 	return map_queues_cpsch(dqm);
 }
@@ -1770,6 +1786,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.initialize = initialize_cpsch;
 		dqm->ops.start = start_cpsch;
 		dqm->ops.stop = stop_cpsch;
+		dqm->ops.pre_reset = pre_reset;
 		dqm->ops.destroy_queue = destroy_queue_cpsch;
 		dqm->ops.update_queue = update_queue;
 		dqm->ops.register_process = register_process;
@@ -1788,6 +1805,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		/* initialize dqm for no cp scheduling */
 		dqm->ops.start = start_nocpsch;
 		dqm->ops.stop = stop_nocpsch;
+		dqm->ops.pre_reset = pre_reset;
 		dqm->ops.create_queue = create_queue_nocpsch;
 		dqm->ops.destroy_queue = destroy_queue_nocpsch;
 		dqm->ops.update_queue = update_queue;

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

@@ -104,6 +104,7 @@ struct device_queue_manager_ops {
 	int	(*initialize)(struct device_queue_manager *dqm);
 	int	(*start)(struct device_queue_manager *dqm);
 	int	(*stop)(struct device_queue_manager *dqm);
+	void	(*pre_reset)(struct device_queue_manager *dqm);
 	void	(*uninitialize)(struct device_queue_manager *dqm);
 	int	(*create_kernel_queue)(struct device_queue_manager *dqm,
 					struct kernel_queue *kq,
@@ -190,7 +191,6 @@ struct device_queue_manager {
 	/* the pasid mapping for each kfd vmid */
 	uint16_t		vmid_pasid[VMID_NUM];
 	uint64_t		pipelines_addr;
-	struct kfd_mem_obj	*pipeline_mem;
 	uint64_t		fence_gpu_addr;
 	unsigned int		*fence_addr;
 	struct kfd_mem_obj	*fence_mem;
@@ -199,6 +199,7 @@ struct device_queue_manager {
 
 	/* hw exception  */
 	bool			is_hws_hang;
+	bool			is_resetting;
 	struct work_struct	hw_exception_work;
 	struct kfd_mem_obj	hiq_sdma_mqd;
 	bool			sched_running;

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

@@ -195,9 +195,9 @@ err_get_kernel_doorbell:
 }
 
 /* Uninitialize a kernel queue and free all its memory usages. */
-static void kq_uninitialize(struct kernel_queue *kq)
+static void kq_uninitialize(struct kernel_queue *kq, bool hanging)
 {
-	if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ)
+	if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ && !hanging)
 		kq->mqd_mgr->destroy_mqd(kq->mqd_mgr,
 					kq->queue->mqd,
 					KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
 	return NULL;
 }
 
-void kernel_queue_uninit(struct kernel_queue *kq)
+void kernel_queue_uninit(struct kernel_queue *kq, bool hanging)
 {
-	kq_uninitialize(kq);
+	kq_uninitialize(kq, hanging);
 	kfree(kq);
 }
 

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

@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)
 	return 0;
 }
 
-void pm_uninit(struct packet_manager *pm)
+void pm_uninit(struct packet_manager *pm, bool hanging)
 {
 	mutex_destroy(&pm->lock);
-	kernel_queue_uninit(pm->priv_queue);
+	kernel_queue_uninit(pm->priv_queue, hanging);
 }
 
 int pm_send_set_resources(struct packet_manager *pm,

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

@@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);
 void device_queue_manager_uninit(struct device_queue_manager *dqm);
 struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
 					enum kfd_queue_type type);
-void kernel_queue_uninit(struct kernel_queue *kq);
+void kernel_queue_uninit(struct kernel_queue *kq, bool hanging);
 int kfd_process_vm_fault(struct device_queue_manager *dqm, unsigned int pasid);
 
 /* Process Queue Manager */
@@ -972,7 +972,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs;
 extern const struct packet_manager_funcs kfd_v9_pm_funcs;
 
 int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm);
-void pm_uninit(struct packet_manager *pm);
+void pm_uninit(struct packet_manager *pm, bool hanging);
 int pm_send_set_resources(struct packet_manager *pm,
 				struct scheduling_resources *res);
 int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues);

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

@@ -324,6 +324,8 @@ struct kfd_process *kfd_create_process(struct file *filep)
 					(int)process->lead_thread->pid);
 	}
 out:
+	if (!IS_ERR(process))
+		kref_get(&process->ref);
 	mutex_unlock(&kfd_processes_mutex);
 
 	return process;

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

@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)
 		/* destroy kernel queue (DIQ) */
 		dqm = pqn->kq->dev->dqm;
 		dqm->ops.destroy_kernel_queue(dqm, pqn->kq, &pdd->qpd);
-		kernel_queue_uninit(pqn->kq);
+		kernel_queue_uninit(pqn->kq, false);
 	}
 
 	if (pqn->q) {

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

@@ -486,6 +486,10 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 			dev->node_props.num_sdma_engines);
 	sysfs_show_32bit_prop(buffer, "num_sdma_xgmi_engines",
 			dev->node_props.num_sdma_xgmi_engines);
+	sysfs_show_32bit_prop(buffer, "num_sdma_queues_per_engine",
+			dev->node_props.num_sdma_queues_per_engine);
+	sysfs_show_32bit_prop(buffer, "num_cp_queues",
+			dev->node_props.num_cp_queues);
 
 	if (dev->gpu) {
 		log_max_watch_addr =
@@ -1309,9 +1313,12 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 	dev->node_props.num_sdma_engines = gpu->device_info->num_sdma_engines;
 	dev->node_props.num_sdma_xgmi_engines =
 				gpu->device_info->num_xgmi_sdma_engines;
+	dev->node_props.num_sdma_queues_per_engine =
+				gpu->device_info->num_sdma_queues_per_engine;
 	dev->node_props.num_gws = (hws_gws_support &&
 		dev->gpu->dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS) ?
 		amdgpu_amdkfd_get_num_gws(dev->gpu->kgd) : 0;
+	dev->node_props.num_cp_queues = get_queues_num(dev->gpu->dqm);
 
 	kfd_fill_mem_clk_max_info(dev);
 	kfd_fill_iolink_non_crat_info(dev);

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

@@ -81,6 +81,8 @@ struct kfd_node_properties {
 	int32_t  drm_render_minor;
 	uint32_t num_sdma_engines;
 	uint32_t num_sdma_xgmi_engines;
+	uint32_t num_sdma_queues_per_engine;
+	uint32_t num_cp_queues;
 	char name[KFD_TOPOLOGY_PUBLIC_NAME_SIZE];
 };
 

drivers/gpu/drm/amd/display/Kconfig

@@ -6,7 +6,7 @@ config DRM_AMD_DC
 	bool "AMD DC - Enable new display engine"
 	default y
 	select SND_HDA_COMPONENT if SND_HDA_CORE
-	select DRM_AMD_DC_DCN if X86 && !(KCOV_INSTRUMENT_ALL && KCOV_ENABLE_COMPARISONS)
+	select DRM_AMD_DC_DCN if (X86 || PPC64) && !(KCOV_INSTRUMENT_ALL && KCOV_ENABLE_COMPARISONS)
 	help
 	  Choose this option if you want to use the new display engine
 	  support for AMDGPU. This adds required support for Vega and

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -98,6 +98,12 @@ MODULE_FIRMWARE(FIRMWARE_RENOIR_DMUB);
 #define FIRMWARE_RAVEN_DMCU		"amdgpu/raven_dmcu.bin"
 MODULE_FIRMWARE(FIRMWARE_RAVEN_DMCU);
 
+/* Number of bytes in PSP header for firmware. */
+#define PSP_HEADER_BYTES 0x100
+
+/* Number of bytes in PSP footer for firmware. */
+#define PSP_FOOTER_BYTES 0x100
+
 /**
  * DOC: overview
  *
@@ -741,28 +747,27 @@ void amdgpu_dm_audio_eld_notify(struct amdgpu_device *adev, int pin)
 
 static int dm_dmub_hw_init(struct amdgpu_device *adev)
 {
-	const unsigned int psp_header_bytes = 0x100;
-	const unsigned int psp_footer_bytes = 0x100;
 	const struct dmcub_firmware_header_v1_0 *hdr;
 	struct dmub_srv *dmub_srv = adev->dm.dmub_srv;
+	struct dmub_srv_fb_info *fb_info = adev->dm.dmub_fb_info;
 	const struct firmware *dmub_fw = adev->dm.dmub_fw;
 	struct dmcu *dmcu = adev->dm.dc->res_pool->dmcu;
 	struct abm *abm = adev->dm.dc->res_pool->abm;
-	struct dmub_srv_region_params region_params;
-	struct dmub_srv_region_info region_info;
-	struct dmub_srv_fb_params fb_params;
-	struct dmub_srv_fb_info fb_info;
 	struct dmub_srv_hw_params hw_params;
 	enum dmub_status status;
 	const unsigned char *fw_inst_const, *fw_bss_data;
-	uint32_t i;
+	uint32_t i, fw_inst_const_size, fw_bss_data_size;
-	int r;
 	bool has_hw_support;
 
 	if (!dmub_srv)
 		/* DMUB isn't supported on the ASIC. */
 		return 0;
 
+	if (!fb_info) {
+		DRM_ERROR("No framebuffer info for DMUB service.\n");
+		return -EINVAL;
+	}
+
 	if (!dmub_fw) {
 		/* Firmware required for DMUB support. */
 		DRM_ERROR("No firmware provided for DMUB.\n");
@@ -782,60 +787,36 @@ static int dm_dmub_hw_init(struct amdgpu_device *adev)
 
 	hdr = (const struct dmcub_firmware_header_v1_0 *)dmub_fw->data;
 
-	/* Calculate the size of all the regions for the DMUB service. */
-	memset(&region_params, 0, sizeof(region_params));
-
-	region_params.inst_const_size = le32_to_cpu(hdr->inst_const_bytes) -
-					psp_header_bytes - psp_footer_bytes;
-	region_params.bss_data_size = le32_to_cpu(hdr->bss_data_bytes);
-	region_params.vbios_size = adev->dm.dc->ctx->dc_bios->bios_size;
-
-	status = dmub_srv_calc_region_info(dmub_srv, &region_params,
-					   &region_info);
-
-	if (status != DMUB_STATUS_OK) {
-		DRM_ERROR("Error calculating DMUB region info: %d\n", status);
-		return -EINVAL;
-	}
-
-	/*
-	 * Allocate a framebuffer based on the total size of all the regions.
-	 * TODO: Move this into GART.
-	 */
-	r = amdgpu_bo_create_kernel(adev, region_info.fb_size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_VRAM, &adev->dm.dmub_bo,
-				    &adev->dm.dmub_bo_gpu_addr,
-				    &adev->dm.dmub_bo_cpu_addr);
-	if (r)
-		return r;
-
-	/* Rebase the regions on the framebuffer address. */
-	memset(&fb_params, 0, sizeof(fb_params));
-	fb_params.cpu_addr = adev->dm.dmub_bo_cpu_addr;
-	fb_params.gpu_addr = adev->dm.dmub_bo_gpu_addr;
-	fb_params.region_info = &region_info;
-
-	status = dmub_srv_calc_fb_info(dmub_srv, &fb_params, &fb_info);
-	if (status != DMUB_STATUS_OK) {
-		DRM_ERROR("Error calculating DMUB FB info: %d\n", status);
-		return -EINVAL;
-	}
-
 	fw_inst_const = dmub_fw->data +
 			le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
-			psp_header_bytes;
+			PSP_HEADER_BYTES;
 
 	fw_bss_data = dmub_fw->data +
 		      le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
 		      le32_to_cpu(hdr->inst_const_bytes);
 
 	/* Copy firmware and bios info into FB memory. */
-	memcpy(fb_info.fb[DMUB_WINDOW_0_INST_CONST].cpu_addr, fw_inst_const,
+	fw_inst_const_size = le32_to_cpu(hdr->inst_const_bytes) -
-	       region_params.inst_const_size);
+			     PSP_HEADER_BYTES - PSP_FOOTER_BYTES;
-	memcpy(fb_info.fb[DMUB_WINDOW_2_BSS_DATA].cpu_addr, fw_bss_data,
+
-	       region_params.bss_data_size);
+	fw_bss_data_size = le32_to_cpu(hdr->bss_data_bytes);
-	memcpy(fb_info.fb[DMUB_WINDOW_3_VBIOS].cpu_addr,
+
-	       adev->dm.dc->ctx->dc_bios->bios, region_params.vbios_size);
+	memcpy(fb_info->fb[DMUB_WINDOW_0_INST_CONST].cpu_addr, fw_inst_const,
+	       fw_inst_const_size);
+	memcpy(fb_info->fb[DMUB_WINDOW_2_BSS_DATA].cpu_addr, fw_bss_data,
+	       fw_bss_data_size);
+	memcpy(fb_info->fb[DMUB_WINDOW_3_VBIOS].cpu_addr, adev->bios,
+	       adev->bios_size);
+
+	/* Reset regions that need to be reset. */
+	memset(fb_info->fb[DMUB_WINDOW_4_MAILBOX].cpu_addr, 0,
+	fb_info->fb[DMUB_WINDOW_4_MAILBOX].size);
+
+	memset(fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].cpu_addr, 0,
+	       fb_info->fb[DMUB_WINDOW_5_TRACEBUFF].size);
+
+	memset(fb_info->fb[DMUB_WINDOW_6_FW_STATE].cpu_addr, 0,
+	       fb_info->fb[DMUB_WINDOW_6_FW_STATE].size);
 
 	/* Initialize hardware. */
 	memset(&hw_params, 0, sizeof(hw_params));
@@ -845,8 +826,8 @@ static int dm_dmub_hw_init(struct amdgpu_device *adev)
 	if (dmcu)
 		hw_params.psp_version = dmcu->psp_version;
 
-	for (i = 0; i < fb_info.num_fb; ++i)
+	for (i = 0; i < fb_info->num_fb; ++i)
-		hw_params.fb[i] = &fb_info.fb[i];
+		hw_params.fb[i] = &fb_info->fb[i];
 
 	status = dmub_srv_hw_init(dmub_srv, &hw_params);
 	if (status != DMUB_STATUS_OK) {
@@ -1174,6 +1155,11 @@ static void amdgpu_dm_dmub_reg_write(void *ctx, uint32_t address,
 static int dm_dmub_sw_init(struct amdgpu_device *adev)
 {
 	struct dmub_srv_create_params create_params;
+	struct dmub_srv_region_params region_params;
+	struct dmub_srv_region_info region_info;
+	struct dmub_srv_fb_params fb_params;
+	struct dmub_srv_fb_info *fb_info;
+	struct dmub_srv *dmub_srv;
 	const struct dmcub_firmware_header_v1_0 *hdr;
 	const char *fw_name_dmub;
 	enum dmub_asic dmub_asic;
@@ -1191,24 +1177,6 @@ static int dm_dmub_sw_init(struct amdgpu_device *adev)
 		return 0;
 	}
 
-	adev->dm.dmub_srv = kzalloc(sizeof(*adev->dm.dmub_srv), GFP_KERNEL);
-	if (!adev->dm.dmub_srv) {
-		DRM_ERROR("Failed to allocate DMUB service!\n");
-		return -ENOMEM;
-	}
-
-	memset(&create_params, 0, sizeof(create_params));
-	create_params.user_ctx = adev;
-	create_params.funcs.reg_read = amdgpu_dm_dmub_reg_read;
-	create_params.funcs.reg_write = amdgpu_dm_dmub_reg_write;
-	create_params.asic = dmub_asic;
-
-	status = dmub_srv_create(adev->dm.dmub_srv, &create_params);
-	if (status != DMUB_STATUS_OK) {
-		DRM_ERROR("Error creating DMUB service: %d\n", status);
-		return -EINVAL;
-	}
-
 	r = request_firmware_direct(&adev->dm.dmub_fw, fw_name_dmub, adev->dev);
 	if (r) {
 		DRM_ERROR("DMUB firmware loading failed: %d\n", r);
@@ -1238,6 +1206,80 @@ static int dm_dmub_sw_init(struct amdgpu_device *adev)
 	DRM_INFO("Loading DMUB firmware via PSP: version=0x%08X\n",
 		 adev->dm.dmcub_fw_version);
 
+	adev->dm.dmub_srv = kzalloc(sizeof(*adev->dm.dmub_srv), GFP_KERNEL);
+	dmub_srv = adev->dm.dmub_srv;
+
+	if (!dmub_srv) {
+		DRM_ERROR("Failed to allocate DMUB service!\n");
+		return -ENOMEM;
+	}
+
+	memset(&create_params, 0, sizeof(create_params));
+	create_params.user_ctx = adev;
+	create_params.funcs.reg_read = amdgpu_dm_dmub_reg_read;
+	create_params.funcs.reg_write = amdgpu_dm_dmub_reg_write;
+	create_params.asic = dmub_asic;
+
+	/* Create the DMUB service. */
+	status = dmub_srv_create(dmub_srv, &create_params);
+	if (status != DMUB_STATUS_OK) {
+		DRM_ERROR("Error creating DMUB service: %d\n", status);
+		return -EINVAL;
+	}
+
+	/* Calculate the size of all the regions for the DMUB service. */
+	memset(&region_params, 0, sizeof(region_params));
+
+	region_params.inst_const_size = le32_to_cpu(hdr->inst_const_bytes) -
+					PSP_HEADER_BYTES - PSP_FOOTER_BYTES;
+	region_params.bss_data_size = le32_to_cpu(hdr->bss_data_bytes);
+	region_params.vbios_size = adev->bios_size;
+	region_params.fw_bss_data =
+		adev->dm.dmub_fw->data +
+		le32_to_cpu(hdr->header.ucode_array_offset_bytes) +
+		le32_to_cpu(hdr->inst_const_bytes);
+
+	status = dmub_srv_calc_region_info(dmub_srv, &region_params,
+					   &region_info);
+
+	if (status != DMUB_STATUS_OK) {
+		DRM_ERROR("Error calculating DMUB region info: %d\n", status);
+		return -EINVAL;
+	}
+
+	/*
+	 * Allocate a framebuffer based on the total size of all the regions.
+	 * TODO: Move this into GART.
+	 */
+	r = amdgpu_bo_create_kernel(adev, region_info.fb_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM, &adev->dm.dmub_bo,
+				    &adev->dm.dmub_bo_gpu_addr,
+				    &adev->dm.dmub_bo_cpu_addr);
+	if (r)
+		return r;
+
+	/* Rebase the regions on the framebuffer address. */
+	memset(&fb_params, 0, sizeof(fb_params));
+	fb_params.cpu_addr = adev->dm.dmub_bo_cpu_addr;
+	fb_params.gpu_addr = adev->dm.dmub_bo_gpu_addr;
+	fb_params.region_info = &region_info;
+
+	adev->dm.dmub_fb_info =
+		kzalloc(sizeof(*adev->dm.dmub_fb_info), GFP_KERNEL);
+	fb_info = adev->dm.dmub_fb_info;
+
+	if (!fb_info) {
+		DRM_ERROR(
+			"Failed to allocate framebuffer info for DMUB service!\n");
+		return -ENOMEM;
+	}
+
+	status = dmub_srv_calc_fb_info(dmub_srv, &fb_params, fb_info);
+	if (status != DMUB_STATUS_OK) {
+		DRM_ERROR("Error calculating DMUB FB info: %d\n", status);
+		return -EINVAL;
+	}
+
 	return 0;
 }
 
@@ -1257,6 +1299,9 @@ static int dm_sw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
+	kfree(adev->dm.dmub_fb_info);
+	adev->dm.dmub_fb_info = NULL;
+
 	if (adev->dm.dmub_srv) {
 		dmub_srv_destroy(adev->dm.dmub_srv);
 		adev->dm.dmub_srv = NULL;
@@ -1559,7 +1604,7 @@ static int dm_resume(void *handle)
 	struct dm_plane_state *dm_new_plane_state;
 	struct dm_atomic_state *dm_state = to_dm_atomic_state(dm->atomic_obj.state);
 	enum dc_connection_type new_connection_type = dc_connection_none;
-	int i;
+	int i, r;
 
 	/* Recreate dc_state - DC invalidates it when setting power state to S3. */
 	dc_release_state(dm_state->context);
@@ -1567,6 +1612,11 @@ static int dm_resume(void *handle)
 	/* TODO: Remove dc_state->dccg, use dc->dccg directly. */
 	dc_resource_state_construct(dm->dc, dm_state->context);
 
+	/* Before powering on DC we need to re-initialize DMUB. */
+	r = dm_dmub_hw_init(adev);
+	if (r)
+		DRM_ERROR("DMUB interface failed to initialize: status=%d\n", r);
+
 	/* power on hardware */
 	dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D0);
 
@@ -3654,27 +3704,21 @@ get_output_color_space(const struct dc_crtc_timing *dc_crtc_timing)
 	return color_space;
 }
 
-static void reduce_mode_colour_depth(struct dc_crtc_timing *timing_out)
+static bool adjust_colour_depth_from_display_info(
-{
+	struct dc_crtc_timing *timing_out,
-	if (timing_out->display_color_depth <= COLOR_DEPTH_888)
+	const struct drm_display_info *info)
-		return;
-
-	timing_out->display_color_depth--;
-}
-
-static void adjust_colour_depth_from_display_info(struct dc_crtc_timing *timing_out,
-						const struct drm_display_info *info)
 {
+	enum dc_color_depth depth = timing_out->display_color_depth;
 	int normalized_clk;
-	if (timing_out->display_color_depth <= COLOR_DEPTH_888)
-		return;
 	do {
 		normalized_clk = timing_out->pix_clk_100hz / 10;
 		/* YCbCr 4:2:0 requires additional adjustment of 1/2 */
 		if (timing_out->pixel_encoding == PIXEL_ENCODING_YCBCR420)
 			normalized_clk /= 2;
 		/* Adjusting pix clock following on HDMI spec based on colour depth */
-		switch (timing_out->display_color_depth) {
+		switch (depth) {
+		case COLOR_DEPTH_888:
+			break;
 		case COLOR_DEPTH_101010:
 			normalized_clk = (normalized_clk * 30) / 24;
 			break;
@@ -3685,14 +3729,15 @@ static void adjust_colour_depth_from_display_info(struct dc_crtc_timing *timing_
 			normalized_clk = (normalized_clk * 48) / 24;
 			break;
 		default:
-			return;
+			/* The above depths are the only ones valid for HDMI. */
+			return false;
 		}
-		if (normalized_clk <= info->max_tmds_clock)
+		if (normalized_clk <= info->max_tmds_clock) {
-			return;
+			timing_out->display_color_depth = depth;
-		reduce_mode_colour_depth(timing_out);
+			return true;
-
+		}
-	} while (timing_out->display_color_depth > COLOR_DEPTH_888);
+	} while (--depth > COLOR_DEPTH_666);
-
+	return false;
 }
 
 static void fill_stream_properties_from_drm_display_mode(
@@ -3773,8 +3818,14 @@ static void fill_stream_properties_from_drm_display_mode(
 
 	stream->out_transfer_func->type = TF_TYPE_PREDEFINED;
 	stream->out_transfer_func->tf = TRANSFER_FUNCTION_SRGB;
-	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A)
+	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A) {
-		adjust_colour_depth_from_display_info(timing_out, info);
+		if (!adjust_colour_depth_from_display_info(timing_out, info) &&
+		    drm_mode_is_420_also(info, mode_in) &&
+		    timing_out->pixel_encoding != PIXEL_ENCODING_YCBCR420) {
+			timing_out->pixel_encoding = PIXEL_ENCODING_YCBCR420;
+			adjust_colour_depth_from_display_info(timing_out, info);
+		}
+	}
 }
 
 static void fill_audio_info(struct audio_info *audio_info,
@@ -4025,7 +4076,8 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector,
 
 	if (aconnector->dc_link && sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT) {
 #if defined(CONFIG_DRM_AMD_DC_DCN)
-		dc_dsc_parse_dsc_dpcd(aconnector->dc_link->dpcd_caps.dsc_caps.dsc_basic_caps.raw,
+		dc_dsc_parse_dsc_dpcd(aconnector->dc_link->ctx->dc,
+				      aconnector->dc_link->dpcd_caps.dsc_caps.dsc_basic_caps.raw,
 				      aconnector->dc_link->dpcd_caps.dsc_caps.dsc_ext_caps.raw,
 				      &dsc_caps);
 #endif
@@ -5561,9 +5613,9 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm,
 
 	drm_connector_attach_max_bpc_property(&aconnector->base, 8, 16);
 
-	/* This defaults to the max in the range, but we want 8bpc. */
+	/* This defaults to the max in the range, but we want 8bpc for non-edp. */
-	aconnector->base.state->max_bpc = 8;
+	aconnector->base.state->max_bpc = (connector_type == DRM_MODE_CONNECTOR_eDP) ? 16 : 8;
-	aconnector->base.state->max_requested_bpc = 8;
+	aconnector->base.state->max_requested_bpc = aconnector->base.state->max_bpc;
 
 	if (connector_type == DRM_MODE_CONNECTOR_eDP &&
 	    dc_is_dmcu_initialized(adev->dm.dc)) {

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h

@@ -132,6 +132,13 @@ struct amdgpu_display_manager {
 	 */
 	struct dmub_srv *dmub_srv;
 
+	/**
+	 * @dmub_fb_info:
+	 *
+	 * Framebuffer regions for the DMUB.
+	 */
+	struct dmub_srv_fb_info *dmub_fb_info;
+
 	/**
 	 * @dmub_fw:
 	 *

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c

@@ -111,17 +111,12 @@ static void init_handler_common_data(struct amdgpu_dm_irq_handler_data *hcd,
  */
 static void dm_irq_work_func(struct work_struct *work)
 {
-	struct list_head *entry;
 	struct irq_list_head *irq_list_head =
 		container_of(work, struct irq_list_head, work);
 	struct list_head *handler_list = &irq_list_head->head;
 	struct amdgpu_dm_irq_handler_data *handler_data;
 
-	list_for_each(entry, handler_list) {
+	list_for_each_entry(handler_data, handler_list, list) {
-		handler_data = list_entry(entry,
-					  struct amdgpu_dm_irq_handler_data,
-					  list);
-
 		DRM_DEBUG_KMS("DM_IRQ: work_func: for dal_src=%d\n",
 				handler_data->irq_source);
 
@@ -528,19 +523,13 @@ static void amdgpu_dm_irq_immediate_work(struct amdgpu_device *adev,
 					 enum dc_irq_source irq_source)
 {
 	struct amdgpu_dm_irq_handler_data *handler_data;
-	struct list_head *entry;
 	unsigned long irq_table_flags;
 
 	DM_IRQ_TABLE_LOCK(adev, irq_table_flags);
 
-	list_for_each(
+	list_for_each_entry(handler_data,
-		entry,
+			    &adev->dm.irq_handler_list_high_tab[irq_source],
-		&adev->dm.irq_handler_list_high_tab[irq_source]) {
+			    list) {
-
-		handler_data = list_entry(entry,
-					  struct amdgpu_dm_irq_handler_data,
-					  list);
-
 		/* Call a subcomponent which registered for immediate
 		 * interrupt notification */
 		handler_data->handler(handler_data->handler_arg);

drivers/gpu/drm/amd/display/dc/calcs/Makefile

@@ -1,5 +1,6 @@
 #
 # Copyright 2017 Advanced Micro Devices, Inc.
+# Copyright 2019 Raptor Engineering, LLC
 #
 # Permission is hereby granted, free of charge, to any person obtaining a
 # copy of this software and associated documentation files (the "Software"),
@@ -24,7 +25,13 @@
 # It calculates Bandwidth and Watermarks values for HW programming
 #
 
+ifdef CONFIG_X86
 calcs_ccflags := -mhard-float -msse
+endif
+
+ifdef CONFIG_PPC64
+calcs_ccflags := -mhard-float -maltivec
+endif
 
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
@@ -32,6 +39,7 @@ IS_OLD_GCC = 1
 endif
 endif
 
+ifdef CONFIG_X86
 ifdef IS_OLD_GCC
 # Stack alignment mismatch, proceed with caution.
 # GCC < 7.1 cannot compile code using `double` and -mpreferred-stack-boundary=3
@@ -40,6 +48,7 @@ calcs_ccflags += -mpreferred-stack-boundary=4
 else
 calcs_ccflags += -msse2
 endif
+endif
 
 CFLAGS_$(AMDDALPATH)/dc/calcs/dcn_calcs.o := $(calcs_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/calcs/dcn_calc_auto.o := $(calcs_ccflags)

drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c

@@ -154,14 +154,14 @@ static void calculate_bandwidth(
 
 
 
-	if (data->d0_underlay_mode == bw_def_none) { d0_underlay_enable = 0; }
+	if (data->d0_underlay_mode == bw_def_none)
-	else {
+		d0_underlay_enable = false;
-		d0_underlay_enable = 1;
+	else
-	}
+		d0_underlay_enable = true;
-	if (data->d1_underlay_mode == bw_def_none) { d1_underlay_enable = 0; }
+	if (data->d1_underlay_mode == bw_def_none)
-	else {
+		d1_underlay_enable = false;
-		d1_underlay_enable = 1;
+	else
-	}
+		d1_underlay_enable = true;
 	data->number_of_underlay_surfaces = d0_underlay_enable + d1_underlay_enable;
 	switch (data->underlay_surface_type) {
 	case bw_def_420:
@@ -286,8 +286,8 @@ static void calculate_bandwidth(
 	data->cursor_width_pixels[2] = bw_int_to_fixed(0);
 	data->cursor_width_pixels[3] = bw_int_to_fixed(0);
 	/* graphics surface parameters from spreadsheet*/
-	fbc_enabled = 0;
+	fbc_enabled = false;
-	lpt_enabled = 0;
+	lpt_enabled = false;
 	for (i = 4; i <= maximum_number_of_surfaces - 3; i++) {
 		if (i < data->number_of_displays + 4) {
 			if (i == 4 && data->d0_underlay_mode == bw_def_underlay_only) {
@@ -338,9 +338,9 @@ static void calculate_bandwidth(
 			data->access_one_channel_only[i] = 0;
 		}
 		if (data->fbc_en[i] == 1) {
-			fbc_enabled = 1;
+			fbc_enabled = true;
 			if (data->lpt_en[i] == 1) {
-				lpt_enabled = 1;
+				lpt_enabled = true;
 			}
 		}
 		data->cursor_width_pixels[i] = bw_int_to_fixed(vbios->cursor_width);

drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c

@@ -1,5 +1,6 @@
 /*
  * Copyright 2017 Advanced Micro Devices, Inc.
+ * Copyright 2019 Raptor Engineering, LLC
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -622,7 +623,7 @@ static bool dcn_bw_apply_registry_override(struct dc *dc)
 {
 	bool updated = false;
 
-	kernel_fpu_begin();
+	DC_FP_START();
 	if ((int)(dc->dcn_soc->sr_exit_time * 1000) != dc->debug.sr_exit_time_ns
 			&& dc->debug.sr_exit_time_ns) {
 		updated = true;
@@ -658,7 +659,7 @@ static bool dcn_bw_apply_registry_override(struct dc *dc)
 		dc->dcn_soc->dram_clock_change_latency =
 				dc->debug.dram_clock_change_latency_ns / 1000.0;
 	}
-	kernel_fpu_end();
+	DC_FP_END();
 
 	return updated;
 }
@@ -738,7 +739,7 @@ bool dcn_validate_bandwidth(
 		dcn_bw_sync_calcs_and_dml(dc);
 
 	memset(v, 0, sizeof(*v));
-	kernel_fpu_begin();
+	DC_FP_START();
 
 	v->sr_exit_time = dc->dcn_soc->sr_exit_time;
 	v->sr_enter_plus_exit_time = dc->dcn_soc->sr_enter_plus_exit_time;
@@ -1271,7 +1272,7 @@ bool dcn_validate_bandwidth(
 	bw_limit = dc->dcn_soc->percent_disp_bw_limit * v->fabric_and_dram_bandwidth_vmax0p9;
 	bw_limit_pass = (v->total_data_read_bandwidth / 1000.0) < bw_limit;
 
-	kernel_fpu_end();
+	DC_FP_END();
 
 	PERFORMANCE_TRACE_END();
 	BW_VAL_TRACE_FINISH();
@@ -1439,7 +1440,7 @@ void dcn_bw_update_from_pplib(struct dc *dc)
 	res = dm_pp_get_clock_levels_by_type_with_voltage(
 			ctx, DM_PP_CLOCK_TYPE_FCLK, &fclks);
 
-	kernel_fpu_begin();
+	DC_FP_START();
 
 	if (res)
 		res = verify_clock_values(&fclks);
@@ -1459,12 +1460,12 @@ void dcn_bw_update_from_pplib(struct dc *dc)
 	} else
 		BREAK_TO_DEBUGGER();
 
-	kernel_fpu_end();
+	DC_FP_END();
 
 	res = dm_pp_get_clock_levels_by_type_with_voltage(
 			ctx, DM_PP_CLOCK_TYPE_DCFCLK, &dcfclks);
 
-	kernel_fpu_begin();
+	DC_FP_START();
 
 	if (res)
 		res = verify_clock_values(&dcfclks);
@@ -1477,7 +1478,7 @@ void dcn_bw_update_from_pplib(struct dc *dc)
 	} else
 		BREAK_TO_DEBUGGER();
 
-	kernel_fpu_end();
+	DC_FP_END();
 }
 
 void dcn_bw_notify_pplib_of_wm_ranges(struct dc *dc)
@@ -1492,11 +1493,11 @@ void dcn_bw_notify_pplib_of_wm_ranges(struct dc *dc)
 	if (!pp || !pp->set_wm_ranges)
 		return;
 
-	kernel_fpu_begin();
+	DC_FP_START();
 	min_fclk_khz = dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 * 1000000 / 32;
 	min_dcfclk_khz = dc->dcn_soc->dcfclkv_min0p65 * 1000;
 	socclk_khz = dc->dcn_soc->socclk * 1000;
-	kernel_fpu_end();
+	DC_FP_END();
 
 	/* Now notify PPLib/SMU about which Watermarks sets they should select
 	 * depending on DPM state they are in. And update BW MGR GFX Engine and
@@ -1547,7 +1548,7 @@ void dcn_bw_notify_pplib_of_wm_ranges(struct dc *dc)
 
 void dcn_bw_sync_calcs_and_dml(struct dc *dc)
 {
-	kernel_fpu_begin();
+	DC_FP_START();
 	DC_LOG_BANDWIDTH_CALCS("sr_exit_time: %f ns\n"
 			"sr_enter_plus_exit_time: %f ns\n"
 			"urgent_latency: %f ns\n"
@@ -1736,5 +1737,5 @@ void dcn_bw_sync_calcs_and_dml(struct dc *dc)
 	dc->dml.ip.bug_forcing_LC_req_same_size_fixed =
 		dc->dcn_ip->bug_forcing_luma_and_chroma_request_to_same_size_fixed == dcn_bw_yes;
 	dc->dml.ip.dcfclk_cstate_latency = dc->dcn_ip->dcfclk_cstate_latency;
-	kernel_fpu_end();
+	DC_FP_END();
 }

drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.c

@@ -27,6 +27,7 @@
 #include "clk_mgr_internal.h"
 
 #include "dce100/dce_clk_mgr.h"
+#include "dcn20_clk_mgr.h"
 #include "reg_helper.h"
 #include "core_types.h"
 #include "dm_helpers.h"
@@ -100,13 +101,13 @@ uint32_t dentist_get_did_from_divider(int divider)
 }
 
 void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
-		struct dc_state *context)
+		struct dc_state *context, bool safe_to_lower)
 {
 	int i;
 
 	clk_mgr->dccg->ref_dppclk = clk_mgr->base.clks.dppclk_khz;
 	for (i = 0; i < clk_mgr->base.ctx->dc->res_pool->pipe_count; i++) {
-		int dpp_inst, dppclk_khz;
+		int dpp_inst, dppclk_khz, prev_dppclk_khz;
 
 		/* Loop index will match dpp->inst if resource exists,
 		 * and we want to avoid dependency on dpp object
@@ -114,8 +115,12 @@ void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
 		dpp_inst = i;
 		dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
 
-		clk_mgr->dccg->funcs->update_dpp_dto(
+		prev_dppclk_khz = clk_mgr->base.ctx->dc->current_state->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
-				clk_mgr->dccg, dpp_inst, dppclk_khz);
+
+		if (safe_to_lower || prev_dppclk_khz < dppclk_khz) {
+			clk_mgr->dccg->funcs->update_dpp_dto(
+							clk_mgr->dccg, dpp_inst, dppclk_khz);
+		}
 	}
 }
 
@@ -161,6 +166,9 @@ void dcn2_update_clocks(struct clk_mgr *clk_mgr_base,
 		dc->debug.force_clock_mode & 0x1) {
 		//this is from resume or boot up, if forced_clock cfg option used, we bypass program dispclk and DPPCLK, but need set them for S3.
 		force_reset = true;
+
+		dcn2_read_clocks_from_hw_dentist(clk_mgr_base);
+
 		//force_clock_mode 0x1:  force reset the clock even it is the same clock as long as it is in Passive level.
 	}
 	display_count = clk_mgr_helper_get_active_display_cnt(dc, context);
@@ -240,7 +248,7 @@ void dcn2_update_clocks(struct clk_mgr *clk_mgr_base,
 	if (dc->config.forced_clocks == false || (force_reset && safe_to_lower)) {
 		if (dpp_clock_lowered) {
 			// if clock is being lowered, increase DTO before lowering refclk
-			dcn20_update_clocks_update_dpp_dto(clk_mgr, context);
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
 			dcn20_update_clocks_update_dentist(clk_mgr);
 		} else {
 			// if clock is being raised, increase refclk before lowering DTO
@@ -248,7 +256,7 @@ void dcn2_update_clocks(struct clk_mgr *clk_mgr_base,
 				dcn20_update_clocks_update_dentist(clk_mgr);
 			// always update dtos unless clock is lowered and not safe to lower
 			if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
-				dcn20_update_clocks_update_dpp_dto(clk_mgr, context);
+				dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
 		}
 	}
 
@@ -339,6 +347,32 @@ void dcn2_enable_pme_wa(struct clk_mgr *clk_mgr_base)
 	}
 }
 
+
+void dcn2_read_clocks_from_hw_dentist(struct clk_mgr *clk_mgr_base)
+{
+	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
+	uint32_t dispclk_wdivider;
+	uint32_t dppclk_wdivider;
+	int disp_divider;
+	int dpp_divider;
+
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, &dispclk_wdivider);
+	REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPPCLK_WDIVIDER, &dppclk_wdivider);
+
+	disp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+	dpp_divider = dentist_get_divider_from_did(dispclk_wdivider);
+
+	if (disp_divider && dpp_divider) {
+		/* Calculate the current DFS clock, in kHz.*/
+		clk_mgr_base->clks.dispclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+			* clk_mgr->base.dentist_vco_freq_khz) / disp_divider;
+
+		clk_mgr_base->clks.dppclk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR
+				* clk_mgr->base.dentist_vco_freq_khz) / dpp_divider;
+	}
+
+}
+
 void dcn2_get_clock(struct clk_mgr *clk_mgr,
 		struct dc_state *context,
 			enum dc_clock_type clock_type,

drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.h

@@ -34,7 +34,7 @@ void dcn2_update_clocks_fpga(struct clk_mgr *clk_mgr,
 			struct dc_state *context,
 			bool safe_to_lower);
 void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
-		struct dc_state *context);
+		struct dc_state *context, bool safe_to_lower);
 
 void dcn2_init_clocks(struct clk_mgr *clk_mgr);
 
@@ -51,4 +51,8 @@ void dcn2_get_clock(struct clk_mgr *clk_mgr,
 			struct dc_clock_config *clock_cfg);
 
 void dcn20_update_clocks_update_dentist(struct clk_mgr_internal *clk_mgr);
+
+void dcn2_read_clocks_from_hw_dentist(struct clk_mgr *clk_mgr_base);
+
+
 #endif //__DCN20_CLK_MGR_H__

drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c

@@ -164,16 +164,16 @@ void rn_update_clocks(struct clk_mgr *clk_mgr_base,
 	}
 
 	if (dpp_clock_lowered) {
-		// if clock is being lowered, increase DTO before lowering refclk
+		// increase per DPP DTO before lowering global dppclk
-		dcn20_update_clocks_update_dpp_dto(clk_mgr, context);
+		dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
 		rn_vbios_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
 	} else {
-		// if clock is being raised, increase refclk before lowering DTO
+		// increase global DPPCLK before lowering per DPP DTO
 		if (update_dppclk || update_dispclk)
 			rn_vbios_smu_set_dppclk(clk_mgr, clk_mgr_base->clks.dppclk_khz);
 		// always update dtos unless clock is lowered and not safe to lower
 		if (new_clocks->dppclk_khz >= dc->current_state->bw_ctx.bw.dcn.clk.dppclk_khz)
-			dcn20_update_clocks_update_dpp_dto(clk_mgr, context);
+			dcn20_update_clocks_update_dpp_dto(clk_mgr, context, safe_to_lower);
 	}
 
 	if (update_dispclk &&
@@ -409,7 +409,7 @@ void build_watermark_ranges(struct clk_bw_params *bw_params, struct pp_smu_wm_ra
 			continue;
 
 		ranges->reader_wm_sets[num_valid_sets].wm_inst = bw_params->wm_table.entries[i].wm_inst;
-		ranges->reader_wm_sets[num_valid_sets].wm_type = bw_params->wm_table.entries[i].wm_type;;
+		ranges->reader_wm_sets[num_valid_sets].wm_type = bw_params->wm_table.entries[i].wm_type;
 		/* We will not select WM based on dcfclk, so leave it as unconstrained */
 		ranges->reader_wm_sets[num_valid_sets].min_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MIN;
 		ranges->reader_wm_sets[num_valid_sets].max_drain_clk_mhz = PP_SMU_WM_SET_RANGE_CLK_UNCONSTRAINED_MAX;

drivers/gpu/drm/amd/display/dc/core/dc.c

@@ -66,6 +66,9 @@
 
 #include "dce/dce_i2c.h"
 
+#define CTX \
+	dc->ctx
+
 #define DC_LOGGER \
 	dc->ctx->logger
 
@@ -579,6 +582,40 @@ static void dc_destruct(struct dc *dc)
 
 }
 
+static bool dc_construct_ctx(struct dc *dc,
+		const struct dc_init_data *init_params)
+{
+	struct dc_context *dc_ctx;
+	enum dce_version dc_version = DCE_VERSION_UNKNOWN;
+
+	dc_ctx = kzalloc(sizeof(*dc_ctx), GFP_KERNEL);
+	if (!dc_ctx)
+		return false;
+
+	dc_ctx->cgs_device = init_params->cgs_device;
+	dc_ctx->driver_context = init_params->driver;
+	dc_ctx->dc = dc;
+	dc_ctx->asic_id = init_params->asic_id;
+	dc_ctx->dc_sink_id_count = 0;
+	dc_ctx->dc_stream_id_count = 0;
+	dc_ctx->dce_environment = init_params->dce_environment;
+
+	/* Create logger */
+
+	dc_version = resource_parse_asic_id(init_params->asic_id);
+	dc_ctx->dce_version = dc_version;
+
+	dc_ctx->perf_trace = dc_perf_trace_create();
+	if (!dc_ctx->perf_trace) {
+		ASSERT_CRITICAL(false);
+		return false;
+	}
+
+	dc->ctx = dc_ctx;
+
+	return true;
+}
+
 static bool dc_construct(struct dc *dc,
 		const struct dc_init_data *init_params)
 {
@@ -590,7 +627,6 @@ static bool dc_construct(struct dc *dc,
 	struct dcn_ip_params *dcn_ip;
 #endif
 
-	enum dce_version dc_version = DCE_VERSION_UNKNOWN;
 	dc->config = init_params->flags;
 
 	// Allocate memory for the vm_helper
@@ -636,26 +672,12 @@ static bool dc_construct(struct dc *dc,
 	dc->soc_bounding_box = init_params->soc_bounding_box;
 #endif
 
-	dc_ctx = kzalloc(sizeof(*dc_ctx), GFP_KERNEL);
+	if (!dc_construct_ctx(dc, init_params)) {
-	if (!dc_ctx) {
 		dm_error("%s: failed to create ctx\n", __func__);
 		goto fail;
 	}
 
-	dc_ctx->cgs_device = init_params->cgs_device;
+        dc_ctx = dc->ctx;
-	dc_ctx->driver_context = init_params->driver;
-	dc_ctx->dc = dc;
-	dc_ctx->asic_id = init_params->asic_id;
-	dc_ctx->dc_sink_id_count = 0;
-	dc_ctx->dc_stream_id_count = 0;
-	dc->ctx = dc_ctx;
-
-	/* Create logger */
-
-	dc_ctx->dce_environment = init_params->dce_environment;
-
-	dc_version = resource_parse_asic_id(init_params->asic_id);
-	dc_ctx->dce_version = dc_version;
 
 	/* Resource should construct all asic specific resources.
 	 * This should be the only place where we need to parse the asic id
@@ -670,7 +692,7 @@ static bool dc_construct(struct dc *dc,
 		bp_init_data.bios = init_params->asic_id.atombios_base_address;
 
 		dc_ctx->dc_bios = dal_bios_parser_create(
-				&bp_init_data, dc_version);
+				&bp_init_data, dc_ctx->dce_version);
 
 		if (!dc_ctx->dc_bios) {
 			ASSERT_CRITICAL(false);
@@ -678,17 +700,13 @@ static bool dc_construct(struct dc *dc,
 		}
 
 		dc_ctx->created_bios = true;
-		}
-
-	dc_ctx->perf_trace = dc_perf_trace_create();
-	if (!dc_ctx->perf_trace) {
-		ASSERT_CRITICAL(false);
-		goto fail;
 	}
 
+
+
 	/* Create GPIO service */
 	dc_ctx->gpio_service = dal_gpio_service_create(
-			dc_version,
+			dc_ctx->dce_version,
 			dc_ctx->dce_environment,
 			dc_ctx);
 
@@ -697,7 +715,7 @@ static bool dc_construct(struct dc *dc,
 		goto fail;
 	}
 
-	dc->res_pool = dc_create_resource_pool(dc, init_params, dc_version);
+	dc->res_pool = dc_create_resource_pool(dc, init_params, dc_ctx->dce_version);
 	if (!dc->res_pool)
 		goto fail;
 
@@ -728,8 +746,6 @@ static bool dc_construct(struct dc *dc,
 	return true;
 
 fail:
-
-	dc_destruct(dc);
 	return false;
 }
 
@@ -783,6 +799,33 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context)
 	dc_release_state(current_ctx);
 }
 
+static void wait_for_no_pipes_pending(struct dc *dc, struct dc_state *context)
+{
+	int i;
+	int count = 0;
+	struct pipe_ctx *pipe;
+	PERF_TRACE();
+	for (i = 0; i < MAX_PIPES; i++) {
+		pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->plane_state)
+			continue;
+
+		/* Timeout 100 ms */
+		while (count < 100000) {
+			/* Must set to false to start with, due to OR in update function */
+			pipe->plane_state->status.is_flip_pending = false;
+			dc->hwss.update_pending_status(pipe);
+			if (!pipe->plane_state->status.is_flip_pending)
+				break;
+			udelay(1);
+			count++;
+		}
+		ASSERT(!pipe->plane_state->status.is_flip_pending);
+	}
+	PERF_TRACE();
+}
+
 /*******************************************************************************
  * Public functions
  ******************************************************************************/
@@ -795,28 +838,38 @@ struct dc *dc_create(const struct dc_init_data *init_params)
 	if (NULL == dc)
 		goto alloc_fail;
 
-	if (false == dc_construct(dc, init_params))
+	if (init_params->dce_environment == DCE_ENV_VIRTUAL_HW) {
-		goto construct_fail;
+		if (false == dc_construct_ctx(dc, init_params)) {
+			dc_destruct(dc);
+			goto construct_fail;
+		}
+	} else {
+		if (false == dc_construct(dc, init_params)) {
+			dc_destruct(dc);
+			goto construct_fail;
+		}
 
-	full_pipe_count = dc->res_pool->pipe_count;
+		full_pipe_count = dc->res_pool->pipe_count;
-	if (dc->res_pool->underlay_pipe_index != NO_UNDERLAY_PIPE)
+		if (dc->res_pool->underlay_pipe_index != NO_UNDERLAY_PIPE)
-		full_pipe_count--;
+			full_pipe_count--;
-	dc->caps.max_streams = min(
+		dc->caps.max_streams = min(
-			full_pipe_count,
+				full_pipe_count,
-			dc->res_pool->stream_enc_count);
+				dc->res_pool->stream_enc_count);
 
-	dc->caps.max_links = dc->link_count;
+		dc->optimize_seamless_boot_streams = 0;
-	dc->caps.max_audios = dc->res_pool->audio_count;
+		dc->caps.max_links = dc->link_count;
-	dc->caps.linear_pitch_alignment = 64;
+		dc->caps.max_audios = dc->res_pool->audio_count;
+		dc->caps.linear_pitch_alignment = 64;
 
-	dc->caps.max_dp_protocol_version = DP_VERSION_1_4;
+		dc->caps.max_dp_protocol_version = DP_VERSION_1_4;
+
+		if (dc->res_pool->dmcu != NULL)
+			dc->versions.dmcu_version = dc->res_pool->dmcu->dmcu_version;
+	}
 
 	/* Populate versioning information */
 	dc->versions.dc_ver = DC_VER;
 
-	if (dc->res_pool->dmcu != NULL)
-		dc->versions.dmcu_version = dc->res_pool->dmcu->dmcu_version;
-
 	dc->build_id = DC_BUILD_ID;
 
 	DC_LOG_DC("Display Core initialized\n");
@@ -834,7 +887,8 @@ alloc_fail:
 
 void dc_hardware_init(struct dc *dc)
 {
-	dc->hwss.init_hw(dc);
+	if (dc->ctx->dce_environment != DCE_ENV_VIRTUAL_HW)
+		dc->hwss.init_hw(dc);
 }
 
 void dc_init_callbacks(struct dc *dc,
@@ -1148,10 +1202,10 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c
 
 	for (i = 0; i < context->stream_count; i++) {
 		if (context->streams[i]->apply_seamless_boot_optimization)
-			dc->optimize_seamless_boot = true;
+			dc->optimize_seamless_boot_streams++;
 	}
 
-	if (!dc->optimize_seamless_boot)
+	if (dc->optimize_seamless_boot_streams == 0)
 		dc->hwss.prepare_bandwidth(dc, context);
 
 	/* re-program planes for existing stream, in case we need to
@@ -1224,9 +1278,12 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c
 
 	dc_enable_stereo(dc, context, dc_streams, context->stream_count);
 
-	if (!dc->optimize_seamless_boot)
+	if (dc->optimize_seamless_boot_streams == 0) {
-			/* pplib is notified if disp_num changed */
+		/* Must wait for no flips to be pending before doing optimize bw */
-			dc->hwss.optimize_bandwidth(dc, context);
+		wait_for_no_pipes_pending(dc, context);
+		/* pplib is notified if disp_num changed */
+		dc->hwss.optimize_bandwidth(dc, context);
+	}
 
 	for (i = 0; i < context->stream_count; i++)
 		context->streams[i]->mode_changed = false;
@@ -1267,7 +1324,7 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc)
 	int i;
 	struct dc_state *context = dc->current_state;
 
-	if (!dc->optimized_required || dc->optimize_seamless_boot)
+	if (!dc->optimized_required || dc->optimize_seamless_boot_streams > 0)
 		return true;
 
 	post_surface_trace(dc);
@@ -1543,7 +1600,7 @@ static enum surface_update_type get_scaling_info_update_type(
 
 		update_flags->bits.scaling_change = 1;
 		if (u->scaling_info->src_rect.width > u->surface->src_rect.width
-				&& u->scaling_info->src_rect.height > u->surface->src_rect.height)
+				|| u->scaling_info->src_rect.height > u->surface->src_rect.height)
 			/* Making src rect bigger requires a bandwidth change */
 			update_flags->bits.clock_change = 1;
 	}
@@ -1557,11 +1614,11 @@ static enum surface_update_type get_scaling_info_update_type(
 		update_flags->bits.position_change = 1;
 
 	if (update_flags->bits.clock_change
-			|| update_flags->bits.bandwidth_change)
+			|| update_flags->bits.bandwidth_change
+			|| update_flags->bits.scaling_change)
 		return UPDATE_TYPE_FULL;
 
-	if (update_flags->bits.scaling_change
+	if (update_flags->bits.position_change)
-			|| update_flags->bits.position_change)
 		return UPDATE_TYPE_MED;
 
 	return UPDATE_TYPE_FAST;
@@ -2051,7 +2108,7 @@ static void commit_planes_do_stream_update(struct dc *dc,
 
 					dc->hwss.optimize_bandwidth(dc, dc->current_state);
 				} else {
-					if (!dc->optimize_seamless_boot)
+					if (dc->optimize_seamless_boot_streams == 0)
 						dc->hwss.prepare_bandwidth(dc, dc->current_state);
 
 					core_link_enable_stream(dc->current_state, pipe_ctx);
@@ -2092,7 +2149,7 @@ static void commit_planes_for_stream(struct dc *dc,
 	int i, j;
 	struct pipe_ctx *top_pipe_to_program = NULL;
 
-	if (dc->optimize_seamless_boot && surface_count > 0) {
+	if (dc->optimize_seamless_boot_streams > 0 && surface_count > 0) {
 		/* Optimize seamless boot flag keeps clocks and watermarks high until
 		 * first flip. After first flip, optimization is required to lower
 		 * bandwidth. Important to note that it is expected UEFI will
@@ -2101,12 +2158,14 @@ static void commit_planes_for_stream(struct dc *dc,
 		 */
 		if (stream->apply_seamless_boot_optimization) {
 			stream->apply_seamless_boot_optimization = false;
-			dc->optimize_seamless_boot = false;
+			dc->optimize_seamless_boot_streams--;
-			dc->optimized_required = true;
+
+			if (dc->optimize_seamless_boot_streams == 0)
+				dc->optimized_required = true;
 		}
 	}
 
-	if (update_type == UPDATE_TYPE_FULL && !dc->optimize_seamless_boot) {
+	if (update_type == UPDATE_TYPE_FULL && dc->optimize_seamless_boot_streams == 0) {
 		dc->hwss.prepare_bandwidth(dc, context);
 		context_clock_trace(dc, context);
 	}

drivers/gpu/drm/amd/display/dc/core/dc_link_ddc.c

@@ -590,7 +590,7 @@ bool dal_ddc_submit_aux_command(struct ddc_service *ddc,
 		struct aux_payload *payload)
 {
 	uint32_t retrieved = 0;
-	bool ret = 0;
+	bool ret = false;
 
 	if (!ddc)
 		return false;

drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c

@@ -2854,10 +2854,12 @@ bool dc_link_handle_hpd_rx_irq(struct dc_link *link, union hpd_irq_data *out_hpd
 	/* For now we only handle 'Downstream port status' case.
 	 * If we got sink count changed it means
 	 * Downstream port status changed,
-	 * then DM should call DC to do the detection. */
+	 * then DM should call DC to do the detection.
-	if (hpd_rx_irq_check_link_loss_status(
+	 * NOTE: Do not handle link loss on eDP since it is internal link*/
-		link,
+	if ((link->connector_signal != SIGNAL_TYPE_EDP) &&
-		&hpd_irq_dpcd_data)) {
+		hpd_rx_irq_check_link_loss_status(
+			link,
+			&hpd_irq_dpcd_data)) {
 		/* Connectivity log: link loss */
 		CONN_DATA_LINK_LOSS(link,
 					hpd_irq_dpcd_data.raw,

drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c

@@ -173,15 +173,20 @@ bool edp_receiver_ready_T9(struct dc_link *link)
 }
 bool edp_receiver_ready_T7(struct dc_link *link)
 {
-	unsigned int tries = 0;
 	unsigned char sinkstatus = 0;
 	unsigned char edpRev = 0;
 	enum dc_status result = DC_OK;
 
+	/* use absolute time stamp to constrain max T7*/
+	unsigned long long enter_timestamp = 0;
+	unsigned long long finish_timestamp = 0;
+	unsigned long long time_taken_in_ns = 0;
+
 	result = core_link_read_dpcd(link, DP_EDP_DPCD_REV, &edpRev, sizeof(edpRev));
 	if (result == DC_OK && edpRev < DP_EDP_12)
 		return true;
 	/* start from eDP version 1.2, SINK_STAUS indicate the sink is ready.*/
+	enter_timestamp = dm_get_timestamp(link->ctx);
 	do {
 		sinkstatus = 0;
 		result = core_link_read_dpcd(link, DP_SINK_STATUS, &sinkstatus, sizeof(sinkstatus));
@@ -189,8 +194,10 @@ bool edp_receiver_ready_T7(struct dc_link *link)
 			break;
 		if (result != DC_OK)
 			break;
-		udelay(25); //MAx T7 is 50ms
+		udelay(25);
-	} while (++tries < 300);
+		finish_timestamp = dm_get_timestamp(link->ctx);
+		time_taken_in_ns = dm_get_elapse_time_in_ns(link->ctx, finish_timestamp, enter_timestamp);
+	} while (time_taken_in_ns < 50 * 1000000); //MAx T7 is 50ms
 
 	if (link->local_sink->edid_caps.panel_patch.extra_t7_ms > 0)
 		udelay(link->local_sink->edid_caps.panel_patch.extra_t7_ms * 1000);

drivers/gpu/drm/amd/display/dc/core/dc_resource.c

@@ -940,30 +940,43 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx)
 
 }
 
-static void calculate_integer_scaling(struct pipe_ctx *pipe_ctx)
+/*
+ * When handling 270 rotation in mixed SLS mode, we have
+ * stream->timing.h_border_left that is non zero.  If we are doing
+ * pipe-splitting, this h_border_left value gets added to recout.x and when it
+ * calls calculate_inits_and_adj_vp() and
+ * adjust_vp_and_init_for_seamless_clip(), it can cause viewport.height for a
+ * pipe to be incorrect.
+ *
+ * To fix this, instead of using stream->timing.h_border_left, we can use
+ * stream->dst.x to represent the border instead.  So we will set h_border_left
+ * to 0 and shift the appropriate amount in stream->dst.x.  We will then
+ * perform all calculations in resource_build_scaling_params() based on this
+ * and then restore the h_border_left and stream->dst.x to their original
+ * values.
+ *
+ * shift_border_left_to_dst() will shift the amount of h_border_left to
+ * stream->dst.x and set h_border_left to 0.  restore_border_left_from_dst()
+ * will restore h_border_left and stream->dst.x back to their original values
+ * We also need to make sure pipe_ctx->plane_res.scl_data.h_active uses the
+ * original h_border_left value in its calculation.
+ */
+int shift_border_left_to_dst(struct pipe_ctx *pipe_ctx)
 {
-	unsigned int integer_multiple = 1;
+	int store_h_border_left = pipe_ctx->stream->timing.h_border_left;
 
-	if (pipe_ctx->plane_state->scaling_quality.integer_scaling) {
+	if (store_h_border_left) {
-		// calculate maximum # of replication of src onto addressable
+		pipe_ctx->stream->timing.h_border_left = 0;
-		integer_multiple = min(
+		pipe_ctx->stream->dst.x += store_h_border_left;
-				pipe_ctx->stream->timing.h_addressable / pipe_ctx->stream->src.width,
-				pipe_ctx->stream->timing.v_addressable  / pipe_ctx->stream->src.height);
-
-		//scale dst
-		pipe_ctx->stream->dst.width  = integer_multiple * pipe_ctx->stream->src.width;
-		pipe_ctx->stream->dst.height = integer_multiple * pipe_ctx->stream->src.height;
-
-		//center dst onto addressable
-		pipe_ctx->stream->dst.x = (pipe_ctx->stream->timing.h_addressable - pipe_ctx->stream->dst.width)/2;
-		pipe_ctx->stream->dst.y = (pipe_ctx->stream->timing.v_addressable - pipe_ctx->stream->dst.height)/2;
-
-		//We are guaranteed that we are scaling in integer ratio
-		pipe_ctx->plane_state->scaling_quality.v_taps = 1;
-		pipe_ctx->plane_state->scaling_quality.h_taps = 1;
-		pipe_ctx->plane_state->scaling_quality.v_taps_c = 1;
-		pipe_ctx->plane_state->scaling_quality.h_taps_c = 1;
 	}
+	return store_h_border_left;
+}
+
+void restore_border_left_from_dst(struct pipe_ctx *pipe_ctx,
+                                  int store_h_border_left)
+{
+	pipe_ctx->stream->dst.x -= store_h_border_left;
+	pipe_ctx->stream->timing.h_border_left = store_h_border_left;
 }
 
 bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
@@ -971,6 +984,7 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
 	const struct dc_plane_state *plane_state = pipe_ctx->plane_state;
 	struct dc_crtc_timing *timing = &pipe_ctx->stream->timing;
 	bool res = false;
+	int store_h_border_left = shift_border_left_to_dst(pipe_ctx);
 	DC_LOGGER_INIT(pipe_ctx->stream->ctx->logger);
 	/* Important: scaling ratio calculation requires pixel format,
 	 * lb depth calculation requires recout and taps require scaling ratios.
@@ -979,14 +993,18 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
 	pipe_ctx->plane_res.scl_data.format = convert_pixel_format_to_dalsurface(
 			pipe_ctx->plane_state->format);
 
-	calculate_integer_scaling(pipe_ctx);
-
 	calculate_scaling_ratios(pipe_ctx);
 
 	calculate_viewport(pipe_ctx);
 
-	if (pipe_ctx->plane_res.scl_data.viewport.height < 16 || pipe_ctx->plane_res.scl_data.viewport.width < 16)
+	if (pipe_ctx->plane_res.scl_data.viewport.height < 16 ||
+		pipe_ctx->plane_res.scl_data.viewport.width < 16) {
+		if (store_h_border_left) {
+			restore_border_left_from_dst(pipe_ctx,
+				store_h_border_left);
+		}
 		return false;
+	}
 
 	calculate_recout(pipe_ctx);
 
@@ -999,8 +1017,10 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
 	pipe_ctx->plane_res.scl_data.recout.x += timing->h_border_left;
 	pipe_ctx->plane_res.scl_data.recout.y += timing->v_border_top;
 
-	pipe_ctx->plane_res.scl_data.h_active = timing->h_addressable + timing->h_border_left + timing->h_border_right;
+	pipe_ctx->plane_res.scl_data.h_active = timing->h_addressable +
-	pipe_ctx->plane_res.scl_data.v_active = timing->v_addressable + timing->v_border_top + timing->v_border_bottom;
+		store_h_border_left + timing->h_border_right;
+	pipe_ctx->plane_res.scl_data.v_active = timing->v_addressable +
+		timing->v_border_top + timing->v_border_bottom;
 
 	/* Taps calculations */
 	if (pipe_ctx->plane_res.xfm != NULL)
@@ -1047,6 +1067,9 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx)
 				plane_state->dst_rect.x,
 				plane_state->dst_rect.y);
 
+	if (store_h_border_left)
+		restore_border_left_from_dst(pipe_ctx, store_h_border_left);
+
 	return res;
 }
 
@@ -1894,8 +1917,26 @@ static int acquire_resource_from_hw_enabled_state(
 		pipe_ctx->plane_res.dpp = pool->dpps[tg_inst];
 		pipe_ctx->stream_res.opp = pool->opps[tg_inst];
 
-		if (pool->dpps[tg_inst])
+		if (pool->dpps[tg_inst]) {
 			pipe_ctx->plane_res.mpcc_inst = pool->dpps[tg_inst]->inst;
+
+			// Read DPP->MPCC->OPP Pipe from HW State
+			if (pool->mpc->funcs->read_mpcc_state) {
+				struct mpcc_state s = {0};
+
+				pool->mpc->funcs->read_mpcc_state(pool->mpc, pipe_ctx->plane_res.mpcc_inst, &s);
+
+				if (s.dpp_id < MAX_MPCC)
+					pool->mpc->mpcc_array[pipe_ctx->plane_res.mpcc_inst].dpp_id = s.dpp_id;
+
+				if (s.bot_mpcc_id < MAX_MPCC)
+					pool->mpc->mpcc_array[pipe_ctx->plane_res.mpcc_inst].mpcc_bot =
+							&pool->mpc->mpcc_array[s.bot_mpcc_id];
+
+				if (s.opp_id < MAX_OPP)
+					pipe_ctx->stream_res.opp->mpc_tree_params.opp_id = s.opp_id;
+			}
+		}
 		pipe_ctx->pipe_idx = tg_inst;
 
 		pipe_ctx->stream = stream;
@@ -2281,7 +2322,7 @@ static void set_avi_info_frame(
 		if (color_space == COLOR_SPACE_SRGB ||
 			color_space == COLOR_SPACE_2020_RGB_FULLRANGE) {
 			hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_FULL_RANGE;
-			hdmi_info.bits.YQ0_YQ1 = YYC_QUANTIZATION_FULL_RANGE;
+			hdmi_info.bits.YQ0_YQ1 = YYC_QUANTIZATION_LIMITED_RANGE;
 		} else if (color_space == COLOR_SPACE_SRGB_LIMITED ||
 					color_space == COLOR_SPACE_2020_RGB_LIMITEDRANGE) {
 			hdmi_info.bits.Q0_Q1   = RGB_QUANTIZATION_LIMITED_RANGE;
@@ -2811,3 +2852,51 @@ unsigned int resource_pixel_format_to_bpp(enum surface_pixel_format format)
 		return -1;
 	}
 }
+static unsigned int get_max_audio_sample_rate(struct audio_mode *modes)
+{
+	if (modes) {
+		if (modes->sample_rates.rate.RATE_192)
+			return 192000;
+		if (modes->sample_rates.rate.RATE_176_4)
+			return 176400;
+		if (modes->sample_rates.rate.RATE_96)
+			return 96000;
+		if (modes->sample_rates.rate.RATE_88_2)
+			return 88200;
+		if (modes->sample_rates.rate.RATE_48)
+			return 48000;
+		if (modes->sample_rates.rate.RATE_44_1)
+			return 44100;
+		if (modes->sample_rates.rate.RATE_32)
+			return 32000;
+	}
+	/*original logic when no audio info*/
+	return 441000;
+}
+
+void get_audio_check(struct audio_info *aud_modes,
+	struct audio_check *audio_chk)
+{
+	unsigned int i;
+	unsigned int max_sample_rate = 0;
+
+	if (aud_modes) {
+		audio_chk->audio_packet_type = 0x2;/*audio sample packet AP = .25 for layout0, 1 for layout1*/
+
+		audio_chk->max_audiosample_rate = 0;
+		for (i = 0; i < aud_modes->mode_count; i++) {
+			max_sample_rate = get_max_audio_sample_rate(&aud_modes->modes[i]);
+			if (audio_chk->max_audiosample_rate < max_sample_rate)
+				audio_chk->max_audiosample_rate = max_sample_rate;
+			/*dts takes the same as type 2: AP = 0.25*/
+		}
+		/*check which one take more bandwidth*/
+		if (audio_chk->max_audiosample_rate > 192000)
+			audio_chk->audio_packet_type = 0x9;/*AP =1*/
+		audio_chk->acat = 0;/*not support*/
+	}
+}
+
+
+
+

drivers/gpu/drm/amd/display/dc/core/dc_stream.c

@@ -406,25 +406,30 @@ bool dc_stream_add_writeback(struct dc *dc,
 		stream->writeback_info[stream->num_wb_info++] = *wb_info;
 	}
 
-	if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
-		dm_error("DC: update_bandwidth failed!\n");
-		return false;
-	}
-
-	/* enable writeback */
 	if (dc->hwss.enable_writeback) {
 		struct dc_stream_status *stream_status = dc_stream_get_status(stream);
 		struct dwbc *dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst];
+		dwb->otg_inst = stream_status->primary_otg_inst;
+	}
+	if (IS_DIAG_DC(dc->ctx->dce_environment)) {
+		if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
+			dm_error("DC: update_bandwidth failed!\n");
+			return false;
+		}
 
-		if (dwb->funcs->is_enabled(dwb)) {
+		/* enable writeback */
-			/* writeback pipe already enabled, only need to update */
+		if (dc->hwss.enable_writeback) {
-			dc->hwss.update_writeback(dc, stream_status, wb_info, dc->current_state);
+			struct dwbc *dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst];
-		} else {
+
-			/* Enable writeback pipe from scratch*/
+			if (dwb->funcs->is_enabled(dwb)) {
-			dc->hwss.enable_writeback(dc, stream_status, wb_info, dc->current_state);
+				/* writeback pipe already enabled, only need to update */
+				dc->hwss.update_writeback(dc, wb_info, dc->current_state);
+			} else {
+				/* Enable writeback pipe from scratch*/
+				dc->hwss.enable_writeback(dc, wb_info, dc->current_state);
+			}
 		}
 	}
-
 	return true;
 }
 
@@ -463,19 +468,29 @@ bool dc_stream_remove_writeback(struct dc *dc,
 	}
 	stream->num_wb_info = j;
 
-	/* recalculate and apply DML parameters */
+	if (IS_DIAG_DC(dc->ctx->dce_environment)) {
-	if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
+		/* recalculate and apply DML parameters */
-		dm_error("DC: update_bandwidth failed!\n");
+		if (!dc->hwss.update_bandwidth(dc, dc->current_state)) {
-		return false;
+			dm_error("DC: update_bandwidth failed!\n");
+			return false;
+		}
+
+		/* disable writeback */
+		if (dc->hwss.disable_writeback)
+			dc->hwss.disable_writeback(dc, dwb_pipe_inst);
 	}
-
-	/* disable writeback */
-	if (dc->hwss.disable_writeback)
-		dc->hwss.disable_writeback(dc, dwb_pipe_inst);
-
 	return true;
 }
 
+bool dc_stream_warmup_writeback(struct dc *dc,
+		int num_dwb,
+		struct dc_writeback_info *wb_info)
+{
+	if (dc->hwss.mmhubbub_warmup)
+		return dc->hwss.mmhubbub_warmup(dc, num_dwb, wb_info);
+	else
+		return false;
+}
 uint32_t dc_stream_get_vblank_counter(const struct dc_stream_state *stream)
 {
 	uint8_t i;

drivers/gpu/drm/amd/display/dc/dc.h

@@ -39,7 +39,7 @@
 #include "inc/hw/dmcu.h"
 #include "dml/display_mode_lib.h"
 
-#define DC_VER "3.2.62"
+#define DC_VER "3.2.64"
 
 #define MAX_SURFACES 3
 #define MAX_PLANES 6
@@ -367,6 +367,7 @@ struct dc_debug_options {
 	bool disable_hubp_power_gate;
 	bool disable_dsc_power_gate;
 	int dsc_min_slice_height_override;
+	int dsc_bpp_increment_div;
 	bool native422_support;
 	bool disable_pplib_wm_range;
 	enum wm_report_mode pplib_wm_report_mode;
@@ -513,7 +514,7 @@ struct dc {
 	bool optimized_required;
 
 	/* Require to maintain clocks and bandwidth for UEFI enabled HW */
-	bool optimize_seamless_boot;
+	int optimize_seamless_boot_streams;
 
 	/* FBC compressor */
 	struct compressor *fbc_compressor;

drivers/gpu/drm/amd/display/dc/dc_dsc.h

@@ -53,7 +53,8 @@ struct dc_dsc_policy {
 	uint32_t min_target_bpp;
 };
 
-bool dc_dsc_parse_dsc_dpcd(const uint8_t *dpcd_dsc_basic_data,
+bool dc_dsc_parse_dsc_dpcd(const struct dc *dc,
+		const uint8_t *dpcd_dsc_basic_data,
 		const uint8_t *dpcd_dsc_ext_data,
 		struct dsc_dec_dpcd_caps *dsc_sink_caps);
 
@@ -77,4 +78,6 @@ bool dc_dsc_compute_config(
 void dc_dsc_get_policy_for_timing(const struct dc_crtc_timing *timing,
 		struct dc_dsc_policy *policy);
 
+void dc_dsc_policy_set_max_target_bpp_limit(uint32_t limit);
+
 #endif

drivers/gpu/drm/amd/display/dc/dc_link.h

@@ -133,6 +133,7 @@ struct dc_link {
 	struct link_flags {
 		bool dp_keep_receiver_powered;
 		bool dp_skip_DID2;
+		bool dp_skip_reset_segment;
 	} wa_flags;
 	struct link_mst_stream_allocation_table mst_stream_alloc_table;
 

drivers/gpu/drm/amd/display/dc/dc_stream.h

@@ -344,10 +344,17 @@ bool dc_add_all_planes_for_stream(
 bool dc_stream_add_writeback(struct dc *dc,
 		struct dc_stream_state *stream,
 		struct dc_writeback_info *wb_info);
+
 bool dc_stream_remove_writeback(struct dc *dc,
 		struct dc_stream_state *stream,
 		uint32_t dwb_pipe_inst);
+
+bool dc_stream_warmup_writeback(struct dc *dc,
+		int num_dwb,
+		struct dc_writeback_info *wb_info);
+
 bool dc_stream_dmdata_status_done(struct dc *dc, struct dc_stream_state *stream);
+
 bool dc_stream_set_dynamic_metadata(struct dc *dc,
 		struct dc_stream_state *stream,
 		struct dc_dmdata_attributes *dmdata_attr);