Merge remote-tracking branch 'drm/drm-next' into drm-misc-next

Backmerge to prevent getting out of sync with drm-next too much.

Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>

drivers/gpu/drm/amd/amdgpu/Makefile

@@ -37,7 +37,8 @@ ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \
 	-I$(FULL_AMD_DISPLAY_PATH)/modules/inc \
 	-I$(FULL_AMD_DISPLAY_PATH)/dc \
 	-I$(FULL_AMD_DISPLAY_PATH)/amdgpu_dm \
-	-I$(FULL_AMD_PATH)/amdkfd
+	-I$(FULL_AMD_PATH)/amdkfd \
+	-I$(FULL_AMD_PATH)/ras/ras_mgr
 
 # Locally disable W=1 warnings enabled in drm subsystem Makefile
 subdir-ccflags-y += -Wno-override-init
@@ -324,4 +325,9 @@ amdgpu-y += \
 	isp_v4_1_1.o
 endif
 
+AMD_GPU_RAS_PATH := ../ras
+AMD_GPU_RAS_FULL_PATH := $(FULL_AMD_PATH)/ras
+include $(AMD_GPU_RAS_FULL_PATH)/Makefile
+amdgpu-y += $(AMD_GPU_RAS_FILES)
+
 obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o

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

@@ -88,6 +88,10 @@ static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev)
 	uint32_t ip_block;
 	int r, i;
 
+	/* Skip suspend of SDMA IP versions >= 4.4.2. They are multi-aid */
+	if (adev->aid_mask)
+		ip_block_mask &= ~BIT(AMD_IP_BLOCK_TYPE_SDMA);
+
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 

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

@@ -372,13 +372,15 @@ void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
 					    u64 *flags);
 int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
 				   enum amd_ip_block_type block_type);
+bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev,
+			    enum amd_ip_block_type block_type);
 bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
 			      enum amd_ip_block_type block_type);
 int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block);
 
 int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block);
 
-#define AMDGPU_MAX_IP_NUM 16
+#define AMDGPU_MAX_IP_NUM AMD_IP_BLOCK_TYPE_NUM
 
 struct amdgpu_ip_block_status {
 	bool valid;
@@ -839,8 +841,6 @@ struct amd_powerplay {
 	const struct amd_pm_funcs *pp_funcs;
 };
 
-struct ip_discovery_top;
-
 /* polaris10 kickers */
 #define ASICID_IS_P20(did, rid)		(((did == 0x67DF) && \
 					 ((rid == 0xE3) || \
@@ -972,8 +972,7 @@ struct amdgpu_device {
 	struct notifier_block		acpi_nb;
 	struct notifier_block		pm_nb;
 	struct amdgpu_i2c_chan		*i2c_bus[AMDGPU_MAX_I2C_BUS];
-	struct debugfs_blob_wrapper     debugfs_vbios_blob;
-	struct debugfs_blob_wrapper     debugfs_discovery_blob;
+	struct debugfs_blob_wrapper debugfs_vbios_blob;
 	struct mutex			srbm_mutex;
 	/* GRBM index mutex. Protects concurrent access to GRBM index */
 	struct mutex                    grbm_idx_mutex;
@@ -1063,6 +1062,9 @@ struct amdgpu_device {
 		u32			log2_max_MBps;
 	} mm_stats;
 
+	/* discovery*/
+	struct amdgpu_discovery_info discovery;
+
 	/* display */
 	bool				enable_virtual_display;
 	struct amdgpu_vkms_output       *amdgpu_vkms_output;
@@ -1174,6 +1176,12 @@ struct amdgpu_device {
 	 * queue fence.
 	 */
 	struct xarray			userq_xa;
+	/**
+	 * @userq_doorbell_xa: Global user queue map (doorbell index → queue)
+	 * Key: doorbell_index (unique global identifier for the queue)
+	 * Value: struct amdgpu_usermode_queue
+	 */
+	struct xarray userq_doorbell_xa;
 
 	/* df */
 	struct amdgpu_df                df;
@@ -1265,8 +1273,6 @@ struct amdgpu_device {
 
 	struct list_head                ras_list;
 
-	struct ip_discovery_top         *ip_top;
-
 	struct amdgpu_reset_domain	*reset_domain;
 
 	struct mutex			benchmark_mutex;
@@ -1309,8 +1315,6 @@ struct amdgpu_device {
 	 */
 	bool                            apu_prefer_gtt;
 
-	struct list_head		userq_mgr_list;
-	struct mutex                    userq_mutex;
 	bool                            userq_halt_for_enforce_isolation;
 	struct amdgpu_uid *uid_info;
 
@@ -1638,7 +1642,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 				 struct drm_file *file_priv);
 void amdgpu_driver_release_kms(struct drm_device *dev);
 
-int amdgpu_device_ip_suspend(struct amdgpu_device *adev);
 int amdgpu_device_prepare(struct drm_device *dev);
 void amdgpu_device_complete(struct drm_device *dev);
 int amdgpu_device_suspend(struct drm_device *dev, bool fbcon);

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

@@ -507,7 +507,6 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev,
 				pm_runtime_get_sync(adev_to_drm(adev)->dev);
 				/* Just fire off a uevent and let userspace tell us what to do */
 				drm_helper_hpd_irq_event(adev_to_drm(adev));
-				pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 				pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			}
 		}

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

@@ -36,6 +36,7 @@
 #include "amdgpu_ras.h"
 #include "amdgpu_umc.h"
 #include "amdgpu_reset.h"
+#include "amdgpu_ras_mgr.h"
 
 /* Total memory size in system memory and all GPU VRAM. Used to
  * estimate worst case amount of memory to reserve for page tables
@@ -746,6 +747,20 @@ void amdgpu_amdkfd_ras_pasid_poison_consumption_handler(struct amdgpu_device *ad
 				enum amdgpu_ras_block block, uint16_t pasid,
 				pasid_notify pasid_fn, void *data, uint32_t reset)
 {
+
+	if (amdgpu_uniras_enabled(adev)) {
+		struct ras_ih_info ih_info;
+
+		memset(&ih_info, 0, sizeof(ih_info));
+		ih_info.block = block;
+		ih_info.pasid = pasid;
+		ih_info.reset = reset;
+		ih_info.pasid_fn = pasid_fn;
+		ih_info.data = data;
+		amdgpu_ras_mgr_handle_consumer_interrupt(adev, &ih_info);
+		return;
+	}
+
 	amdgpu_umc_pasid_poison_handler(adev, block, pasid, pasid_fn, data, reset);
 }
 

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

@@ -71,7 +71,7 @@ struct kgd_mem {
 	struct mutex lock;
 	struct amdgpu_bo *bo;
 	struct dma_buf *dmabuf;
-	struct hmm_range *range;
+	struct amdgpu_hmm_range *range;
 	struct list_head attachments;
 	/* protected by amdkfd_process_info.lock */
 	struct list_head validate_list;

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

@@ -1057,7 +1057,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 	struct amdkfd_process_info *process_info = mem->process_info;
 	struct amdgpu_bo *bo = mem->bo;
 	struct ttm_operation_ctx ctx = { true, false };
-	struct hmm_range *range;
+	struct amdgpu_hmm_range *range;
 	int ret = 0;
 
 	mutex_lock(&process_info->lock);
@@ -1089,8 +1089,15 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 		return 0;
 	}
 
-	ret = amdgpu_ttm_tt_get_user_pages(bo, &range);
+	range = amdgpu_hmm_range_alloc(NULL);
+	if (unlikely(!range)) {
+		ret = -ENOMEM;
+		goto unregister_out;
+	}
+
+	ret = amdgpu_ttm_tt_get_user_pages(bo, range);
 	if (ret) {
+		amdgpu_hmm_range_free(range);
 		if (ret == -EAGAIN)
 			pr_debug("Failed to get user pages, try again\n");
 		else
@@ -1113,7 +1120,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 	amdgpu_bo_unreserve(bo);
 
 release_out:
-	amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range);
+	amdgpu_hmm_range_free(range);
 unregister_out:
 	if (ret)
 		amdgpu_hmm_unregister(bo);
@@ -1916,7 +1923,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	if (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm)) {
 		amdgpu_hmm_unregister(mem->bo);
 		mutex_lock(&process_info->notifier_lock);
-		amdgpu_ttm_tt_discard_user_pages(mem->bo->tbo.ttm, mem->range);
+		amdgpu_hmm_range_free(mem->range);
 		mutex_unlock(&process_info->notifier_lock);
 	}
 
@@ -1954,9 +1961,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	 */
 	if (size) {
 		if (!is_imported &&
-		   (mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM ||
-		   (adev->apu_prefer_gtt &&
-		    mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT)))
+		   mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)
 			*size = bo_size;
 		else
 			*size = 0;
@@ -2542,7 +2547,7 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 
 		bo = mem->bo;
 
-		amdgpu_ttm_tt_discard_user_pages(bo->tbo.ttm, mem->range);
+		amdgpu_hmm_range_free(mem->range);
 		mem->range = NULL;
 
 		/* BO reservations and getting user pages (hmm_range_fault)
@@ -2566,9 +2571,14 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 			}
 		}
 
+		mem->range = amdgpu_hmm_range_alloc(NULL);
+		if (unlikely(!mem->range))
+			return -ENOMEM;
 		/* Get updated user pages */
-		ret = amdgpu_ttm_tt_get_user_pages(bo, &mem->range);
+		ret = amdgpu_ttm_tt_get_user_pages(bo, mem->range);
 		if (ret) {
+			amdgpu_hmm_range_free(mem->range);
+			mem->range = NULL;
 			pr_debug("Failed %d to get user pages\n", ret);
 
 			/* Return -EFAULT bad address error as success. It will
@@ -2741,8 +2751,8 @@ static int confirm_valid_user_pages_locked(struct amdkfd_process_info *process_i
 			continue;
 
 		/* Only check mem with hmm range associated */
-		valid = amdgpu_ttm_tt_get_user_pages_done(
-					mem->bo->tbo.ttm, mem->range);
+		valid = amdgpu_hmm_range_valid(mem->range);
+		amdgpu_hmm_range_free(mem->range);
 
 		mem->range = NULL;
 		if (!valid) {

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

@@ -181,19 +181,22 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
 	u8 frev, crev;
 	int usage_bytes = 0;
 
-	if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) {
-		if (frev == 2 && crev == 1) {
-			fw_usage_v2_1 =
-				(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
-			amdgpu_atomfirmware_allocate_fb_v2_1(adev,
-					fw_usage_v2_1,
-					&usage_bytes);
-		} else if (frev >= 2 && crev >= 2) {
-			fw_usage_v2_2 =
-				(struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset);
-			amdgpu_atomfirmware_allocate_fb_v2_2(adev,
-					fw_usage_v2_2,
-					&usage_bytes);
+	/* Skip atomfirmware allocation for SRIOV VFs when dynamic crit regn is enabled */
+	if (!(amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled)) {
+		if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) {
+			if (frev == 2 && crev == 1) {
+				fw_usage_v2_1 =
+					(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
+				amdgpu_atomfirmware_allocate_fb_v2_1(adev,
+						fw_usage_v2_1,
+						&usage_bytes);
+			} else if (frev >= 2 && crev >= 2) {
+				fw_usage_v2_2 =
+					(struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset);
+				amdgpu_atomfirmware_allocate_fb_v2_2(adev,
+						fw_usage_v2_2,
+						&usage_bytes);
+			}
 		}
 	}
 

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

@@ -96,13 +96,14 @@ void amdgpu_bios_release(struct amdgpu_device *adev)
  * part of the system bios.  On boot, the system bios puts a
  * copy of the igp rom at the start of vram if a discrete card is
  * present.
- * For SR-IOV, the vbios image is also put in VRAM in the VF.
+ * For SR-IOV, if dynamic critical region is not enabled,
+ * the vbios image is also put at the start of VRAM in the VF.
  */
 static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
 {
-	uint8_t __iomem *bios;
+	uint8_t __iomem *bios = NULL;
 	resource_size_t vram_base;
-	resource_size_t size = 256 * 1024; /* ??? */
+	u32 size = 256U * 1024U; /* ??? */
 
 	if (!(adev->flags & AMD_IS_APU))
 		if (amdgpu_device_need_post(adev))
@@ -114,18 +115,33 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
 
 	adev->bios = NULL;
 	vram_base = pci_resource_start(adev->pdev, 0);
-	bios = ioremap_wc(vram_base, size);
-	if (!bios)
-		return false;
 
 	adev->bios = kmalloc(size, GFP_KERNEL);
-	if (!adev->bios) {
-		iounmap(bios);
+	if (!adev->bios)
 		return false;
+
+	/* For SRIOV with dynamic critical region is enabled,
+	 * the vbios image is put at a dynamic offset of VRAM in the VF.
+	 * If dynamic critical region is disabled, follow the existing logic as on baremetal.
+	 */
+	if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) {
+		if (amdgpu_virt_get_dynamic_data_info(adev,
+				AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, &size)) {
+			amdgpu_bios_release(adev);
+			return false;
+		}
+	} else {
+		bios = ioremap_wc(vram_base, size);
+		if (!bios) {
+			amdgpu_bios_release(adev);
+			return false;
+		}
+
+		memcpy_fromio(adev->bios, bios, size);
+		iounmap(bios);
 	}
+
 	adev->bios_size = size;
-	memcpy_fromio(adev->bios, bios, size);
-	iounmap(bios);
 
 	if (!check_atom_bios(adev, size)) {
 		amdgpu_bios_release(adev);

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

@@ -38,7 +38,7 @@ struct amdgpu_bo_list_entry {
 	struct amdgpu_bo		*bo;
 	struct amdgpu_bo_va		*bo_va;
 	uint32_t			priority;
-	struct hmm_range		*range;
+	struct amdgpu_hmm_range		*range;
 	bool				user_invalidated;
 };
 

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

@@ -734,10 +734,8 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -919,10 +917,8 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 out:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -1146,10 +1142,8 @@ out:
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 exit:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -1486,10 +1480,8 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
 out:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
 	    connector->connector_type == DRM_MODE_CONNECTOR_eDP)

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

@@ -1,4 +1,4 @@
-// SPDX-License-Identifier: GPL-2.0
+// SPDX-License-Identifier: MIT
 /*
  * Copyright 2025 Advanced Micro Devices, Inc.
  *

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

@@ -1,4 +1,4 @@
-/* SPDX-License-Identifier: GPL-2.0 */
+/* SPDX-License-Identifier: MIT */
 /*
  * Copyright 2025 Advanced Micro Devices, Inc.
  *

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

@@ -29,7 +29,6 @@
 #include <linux/pagemap.h>
 #include <linux/sync_file.h>
 #include <linux/dma-buf.h>
-#include <linux/hmm.h>
 
 #include <drm/amdgpu_drm.h>
 #include <drm/drm_syncobj.h>
@@ -41,6 +40,7 @@
 #include "amdgpu_gmc.h"
 #include "amdgpu_gem.h"
 #include "amdgpu_ras.h"
+#include "amdgpu_hmm.h"
 
 static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p,
 				 struct amdgpu_device *adev,
@@ -891,12 +891,17 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		bool userpage_invalidated = false;
 		struct amdgpu_bo *bo = e->bo;
 
-		r = amdgpu_ttm_tt_get_user_pages(bo, &e->range);
+		e->range = amdgpu_hmm_range_alloc(NULL);
+		if (unlikely(!e->range))
+			return -ENOMEM;
+
+		r = amdgpu_ttm_tt_get_user_pages(bo, e->range);
 		if (r)
 			goto out_free_user_pages;
 
 		for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
-			if (bo->tbo.ttm->pages[i] != hmm_pfn_to_page(e->range->hmm_pfns[i])) {
+			if (bo->tbo.ttm->pages[i] !=
+				hmm_pfn_to_page(e->range->hmm_range.hmm_pfns[i])) {
 				userpage_invalidated = true;
 				break;
 			}
@@ -990,9 +995,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 out_free_user_pages:
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-		struct amdgpu_bo *bo = e->bo;
-
-		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range);
+		amdgpu_hmm_range_free(e->range);
 		e->range = NULL;
 	}
 	mutex_unlock(&p->bo_list->bo_list_mutex);
@@ -1323,8 +1326,8 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	 */
 	r = 0;
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-		r |= !amdgpu_ttm_tt_get_user_pages_done(e->bo->tbo.ttm,
-							e->range);
+		r |= !amdgpu_hmm_range_valid(e->range);
+		amdgpu_hmm_range_free(e->range);
 		e->range = NULL;
 	}
 	if (r) {

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

@@ -129,7 +129,6 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
 	if (use_bank) {
 		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
 		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines)) {
-			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			amdgpu_virt_disable_access_debugfs(adev);
 			return -EINVAL;
@@ -179,7 +178,6 @@ end:
 	if (pm_pg_lock)
 		mutex_unlock(&adev->pm.mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	amdgpu_virt_disable_access_debugfs(adev);
@@ -255,7 +253,6 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off
 	if (rd->id.use_grbm) {
 		if ((rd->id.grbm.sh != 0xFFFFFFFF && rd->id.grbm.sh >= adev->gfx.config.max_sh_per_se) ||
 		    (rd->id.grbm.se != 0xFFFFFFFF && rd->id.grbm.se >= adev->gfx.config.max_shader_engines)) {
-			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			amdgpu_virt_disable_access_debugfs(adev);
 			mutex_unlock(&rd->lock);
@@ -310,7 +307,6 @@ end:
 
 	mutex_unlock(&rd->lock);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	amdgpu_virt_disable_access_debugfs(adev);
@@ -446,7 +442,6 @@ static ssize_t amdgpu_debugfs_gprwave_read(struct file *f, char __user *buf, siz
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, rd->id.xcc_id);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (!x) {
@@ -557,7 +552,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -617,7 +611,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -676,7 +669,6 @@ static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -736,7 +728,6 @@ static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -795,7 +786,6 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -855,7 +845,6 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -1003,7 +992,6 @@ static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf,
 
 	r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (r) {
@@ -1094,7 +1082,6 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (!x) {
@@ -1192,7 +1179,6 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	while (size) {
@@ -1266,7 +1252,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_read(struct file *f, char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1315,7 +1300,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_write(struct file *f, const char
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1365,7 +1349,6 @@ static ssize_t amdgpu_debugfs_gfxoff_count_read(struct file *f, char __user *buf
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1414,7 +1397,6 @@ static ssize_t amdgpu_debugfs_gfxoff_write(struct file *f, const char __user *bu
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1460,7 +1442,6 @@ static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1501,7 +1482,6 @@ static ssize_t amdgpu_debugfs_gfxoff_status_read(struct file *f, char __user *bu
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1701,7 +1681,6 @@ static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused)
 
 	up_write(&adev->reset_domain->sem);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1721,7 +1700,6 @@ static int amdgpu_debugfs_evict_vram(void *data, u64 *val)
 
 	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1742,7 +1720,6 @@ static int amdgpu_debugfs_evict_gtt(void *data, u64 *val)
 
 	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_TT);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1762,7 +1739,6 @@ static int amdgpu_debugfs_benchmark(void *data, u64 val)
 
 	r = amdgpu_benchmark(adev, val);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return r;
@@ -1902,7 +1878,7 @@ no_preempt:
 			continue;
 		}
 		job = to_amdgpu_job(s_job);
-		if (preempted && (&job->hw_fence.base) == fence)
+		if (preempted && (&job->hw_fence->base) == fence)
 			/* mark the job as preempted */
 			job->preemption_status |= AMDGPU_IB_PREEMPTED;
 	}
@@ -2014,7 +1990,6 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val)
 		ret = -EINVAL;
 
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return ret;
@@ -2123,10 +2098,9 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
 	debugfs_create_blob("amdgpu_vbios", 0444, root,
 			    &adev->debugfs_vbios_blob);
 
-	adev->debugfs_discovery_blob.data = adev->mman.discovery_bin;
-	adev->debugfs_discovery_blob.size = adev->mman.discovery_tmr_size;
-	debugfs_create_blob("amdgpu_discovery", 0444, root,
-			    &adev->debugfs_discovery_blob);
+	if (adev->discovery.debugfs_blob.size)
+		debugfs_create_blob("amdgpu_discovery", 0444, root,
+				    &adev->discovery.debugfs_blob);
 
 	return 0;
 }

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

@@ -2387,7 +2387,7 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_device_ip_is_valid - is the hardware IP enabled
+ * amdgpu_device_ip_is_hw - is the hardware IP enabled
  *
  * @adev: amdgpu_device pointer
  * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
@@ -2395,6 +2395,27 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
  * Check if the hardware IP is enable or not.
  * Returns true if it the IP is enable, false if not.
  */
+bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev,
+			    enum amd_ip_block_type block_type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].version->type == block_type)
+			return adev->ip_blocks[i].status.hw;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_device_ip_is_valid - is the hardware IP valid
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Check if the hardware IP is valid or not.
+ * Returns true if it the IP is valid, false if not.
+ */
 bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
 			       enum amd_ip_block_type block_type)
 {
@@ -2633,7 +2654,7 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 		chip_name = "arcturus";
 		break;
 	case CHIP_NAVI12:
-		if (adev->mman.discovery_bin)
+		if (adev->discovery.bin)
 			return 0;
 		chip_name = "navi12";
 		break;
@@ -2761,6 +2782,10 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		r = amdgpu_virt_request_full_gpu(adev, true);
 		if (r)
 			return r;
+
+		r = amdgpu_virt_init_critical_region(adev);
+		if (r)
+			return r;
 	}
 
 	switch (adev->asic_type) {
@@ -3780,7 +3805,6 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_DCE)
 			continue;
 
-		/* XXX handle errors */
 		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
 		if (r)
 			return r;
@@ -3863,9 +3887,9 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
 			continue;
 
-		/* XXX handle errors */
 		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
-		adev->ip_blocks[i].status.hw = false;
+		if (r)
+			return r;
 
 		/* handle putting the SMC in the appropriate state */
 		if (!amdgpu_sriov_vf(adev)) {
@@ -3895,7 +3919,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
  * in each IP into a state suitable for suspend.
  * Returns 0 on success, negative error code on failure.
  */
-int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
+static int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
 {
 	int r;
 
@@ -4191,7 +4215,6 @@ bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev,
 #else
 		return false;
 #endif
-	case CHIP_BONAIRE:
 	case CHIP_KAVERI:
 	case CHIP_KABINI:
 	case CHIP_MULLINS:
@@ -4285,58 +4308,53 @@ static int amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev)
 	long timeout;
 	int ret = 0;
 
-	/*
-	 * By default timeout for jobs is 10 sec
-	 */
-	adev->compute_timeout = adev->gfx_timeout = msecs_to_jiffies(10000);
-	adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
+	/* By default timeout for all queues is 2 sec */
+	adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout =
+		adev->video_timeout = msecs_to_jiffies(2000);
 
-	if (strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
-		while ((timeout_setting = strsep(&input, ",")) &&
-				strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
-			ret = kstrtol(timeout_setting, 0, &timeout);
-			if (ret)
-				return ret;
+	if (!strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH))
+		return 0;
 
-			if (timeout == 0) {
-				index++;
-				continue;
-			} else if (timeout < 0) {
-				timeout = MAX_SCHEDULE_TIMEOUT;
-				dev_warn(adev->dev, "lockup timeout disabled");
-				add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
-			} else {
-				timeout = msecs_to_jiffies(timeout);
-			}
+	while ((timeout_setting = strsep(&input, ",")) &&
+	       strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
+		ret = kstrtol(timeout_setting, 0, &timeout);
+		if (ret)
+			return ret;
 
-			switch (index++) {
-			case 0:
-				adev->gfx_timeout = timeout;
-				break;
-			case 1:
-				adev->compute_timeout = timeout;
-				break;
-			case 2:
-				adev->sdma_timeout = timeout;
-				break;
-			case 3:
-				adev->video_timeout = timeout;
-				break;
-			default:
-				break;
-			}
+		if (timeout == 0) {
+			index++;
+			continue;
+		} else if (timeout < 0) {
+			timeout = MAX_SCHEDULE_TIMEOUT;
+			dev_warn(adev->dev, "lockup timeout disabled");
+			add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
+		} else {
+			timeout = msecs_to_jiffies(timeout);
 		}
-		/*
-		 * There is only one value specified and
-		 * it should apply to all non-compute jobs.
-		 */
-		if (index == 1) {
-			adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
-			if (amdgpu_sriov_vf(adev) || amdgpu_passthrough(adev))
-				adev->compute_timeout = adev->gfx_timeout;
+
+		switch (index++) {
+		case 0:
+			adev->gfx_timeout = timeout;
+			break;
+		case 1:
+			adev->compute_timeout = timeout;
+			break;
+		case 2:
+			adev->sdma_timeout = timeout;
+			break;
+		case 3:
+			adev->video_timeout = timeout;
+			break;
+		default:
+			break;
 		}
 	}
 
+	/* When only one value specified apply it to all queues. */
+	if (index == 1)
+		adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout =
+			adev->video_timeout = timeout;
+
 	return ret;
 }
 
@@ -4391,6 +4409,55 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
 		dev_info(adev->dev, "MCBP is enabled\n");
 }
 
+static int amdgpu_device_sys_interface_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_atombios_sysfs_init(adev);
+	if (r)
+		drm_err(&adev->ddev,
+			"registering atombios sysfs failed (%d).\n", r);
+
+	r = amdgpu_pm_sysfs_init(adev);
+	if (r)
+		dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r);
+
+	r = amdgpu_ucode_sysfs_init(adev);
+	if (r) {
+		adev->ucode_sysfs_en = false;
+		dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r);
+	} else
+		adev->ucode_sysfs_en = true;
+
+	r = amdgpu_device_attr_sysfs_init(adev);
+	if (r)
+		dev_err(adev->dev, "Could not create amdgpu device attr\n");
+
+	r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group);
+	if (r)
+		dev_err(adev->dev,
+			"Could not create amdgpu board attributes\n");
+
+	amdgpu_fru_sysfs_init(adev);
+	amdgpu_reg_state_sysfs_init(adev);
+	amdgpu_xcp_sysfs_init(adev);
+
+	return r;
+}
+
+static void amdgpu_device_sys_interface_fini(struct amdgpu_device *adev)
+{
+	if (adev->pm.sysfs_initialized)
+		amdgpu_pm_sysfs_fini(adev);
+	if (adev->ucode_sysfs_en)
+		amdgpu_ucode_sysfs_fini(adev);
+	amdgpu_device_attr_sysfs_fini(adev);
+	amdgpu_fru_sysfs_fini(adev);
+
+	amdgpu_reg_state_sysfs_fini(adev);
+	amdgpu_xcp_sysfs_fini(adev);
+}
+
 /**
  * amdgpu_device_init - initialize the driver
  *
@@ -4490,7 +4557,6 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	mutex_init(&adev->gfx.userq_sch_mutex);
 	mutex_init(&adev->gfx.workload_profile_mutex);
 	mutex_init(&adev->vcn.workload_profile_mutex);
-	mutex_init(&adev->userq_mutex);
 
 	amdgpu_device_init_apu_flags(adev);
 
@@ -4518,7 +4584,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 
 	INIT_LIST_HEAD(&adev->pm.od_kobj_list);
 
-	INIT_LIST_HEAD(&adev->userq_mgr_list);
+	xa_init(&adev->userq_doorbell_xa);
 
 	INIT_DELAYED_WORK(&adev->delayed_init_work,
 			  amdgpu_device_delayed_init_work_handler);
@@ -4814,39 +4880,14 @@ fence_driver_init:
 		flush_delayed_work(&adev->delayed_init_work);
 	}
 
+	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
+		amdgpu_xgmi_reset_on_init(adev);
 	/*
 	 * Place those sysfs registering after `late_init`. As some of those
 	 * operations performed in `late_init` might affect the sysfs
 	 * interfaces creating.
 	 */
-	r = amdgpu_atombios_sysfs_init(adev);
-	if (r)
-		drm_err(&adev->ddev,
-			"registering atombios sysfs failed (%d).\n", r);
-
-	r = amdgpu_pm_sysfs_init(adev);
-	if (r)
-		dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r);
-
-	r = amdgpu_ucode_sysfs_init(adev);
-	if (r) {
-		adev->ucode_sysfs_en = false;
-		dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r);
-	} else
-		adev->ucode_sysfs_en = true;
-
-	r = amdgpu_device_attr_sysfs_init(adev);
-	if (r)
-		dev_err(adev->dev, "Could not create amdgpu device attr\n");
-
-	r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group);
-	if (r)
-		dev_err(adev->dev,
-			"Could not create amdgpu board attributes\n");
-
-	amdgpu_fru_sysfs_init(adev);
-	amdgpu_reg_state_sysfs_init(adev);
-	amdgpu_xcp_sysfs_init(adev);
+	r = amdgpu_device_sys_interface_init(adev);
 
 	if (IS_ENABLED(CONFIG_PERF_EVENTS))
 		r = amdgpu_pmu_init(adev);
@@ -4874,9 +4915,6 @@ fence_driver_init:
 	if (px)
 		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
 
-	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
-		amdgpu_xgmi_reset_on_init(adev);
-
 	amdgpu_device_check_iommu_direct_map(adev);
 
 	adev->pm_nb.notifier_call = amdgpu_device_pm_notifier;
@@ -4968,15 +5006,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 	}
 	amdgpu_fence_driver_hw_fini(adev);
 
-	if (adev->pm.sysfs_initialized)
-		amdgpu_pm_sysfs_fini(adev);
-	if (adev->ucode_sysfs_en)
-		amdgpu_ucode_sysfs_fini(adev);
-	amdgpu_device_attr_sysfs_fini(adev);
-	amdgpu_fru_sysfs_fini(adev);
-
-	amdgpu_reg_state_sysfs_fini(adev);
-	amdgpu_xcp_sysfs_fini(adev);
+	amdgpu_device_sys_interface_fini(adev);
 
 	/* disable ras feature must before hw fini */
 	amdgpu_ras_pre_fini(adev);
@@ -5051,7 +5081,7 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 
 	if (IS_ENABLED(CONFIG_PERF_EVENTS))
 		amdgpu_pmu_fini(adev);
-	if (adev->mman.discovery_bin)
+	if (adev->discovery.bin)
 		amdgpu_discovery_fini(adev);
 
 	amdgpu_reset_put_reset_domain(adev->reset_domain);
@@ -5225,10 +5255,14 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 
 	amdgpu_ras_suspend(adev);
 
-	amdgpu_device_ip_suspend_phase1(adev);
+	r = amdgpu_device_ip_suspend_phase1(adev);
+	if (r)
+		return r;
 
 	amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
-	amdgpu_userq_suspend(adev);
+	r = amdgpu_userq_suspend(adev);
+	if (r)
+		return r;
 
 	r = amdgpu_device_evict_resources(adev);
 	if (r)
@@ -5238,7 +5272,9 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 
 	amdgpu_fence_driver_hw_fini(adev);
 
-	amdgpu_device_ip_suspend_phase2(adev);
+	r = amdgpu_device_ip_suspend_phase2(adev);
+	if (r)
+		return r;
 
 	if (amdgpu_sriov_vf(adev))
 		amdgpu_virt_release_full_gpu(adev, false);
@@ -5809,11 +5845,6 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 		if (!amdgpu_ring_sched_ready(ring))
 			continue;
 
-		/* Clear job fence from fence drv to avoid force_completion
-		 * leave NULL and vm flush fence in fence drv
-		 */
-		amdgpu_fence_driver_clear_job_fences(ring);
-
 		/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
 		amdgpu_fence_driver_force_completion(ring);
 	}
@@ -6542,7 +6573,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	 *
 	 * job->base holds a reference to parent fence
 	 */
-	if (job && dma_fence_is_signaled(&job->hw_fence.base)) {
+	if (job && dma_fence_is_signaled(&job->hw_fence->base)) {
 		job_signaled = true;
 		dev_info(adev->dev, "Guilty job already signaled, skipping HW reset");
 		goto skip_hw_reset;
@@ -7286,10 +7317,17 @@ void amdgpu_device_flush_hdp(struct amdgpu_device *adev,
 	if (adev->gmc.xgmi.connected_to_cpu)
 		return;
 
-	if (ring && ring->funcs->emit_hdp_flush)
+	if (ring && ring->funcs->emit_hdp_flush) {
 		amdgpu_ring_emit_hdp_flush(ring);
-	else
-		amdgpu_asic_flush_hdp(adev, ring);
+		return;
+	}
+
+	if (!ring && amdgpu_sriov_runtime(adev)) {
+		if (!amdgpu_kiq_hdp_flush(adev))
+			return;
+	}
+
+	amdgpu_asic_flush_hdp(adev, ring);
 }
 
 void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,

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

@@ -24,9 +24,21 @@
 #ifndef __AMDGPU_DISCOVERY__
 #define __AMDGPU_DISCOVERY__
 
+#include <linux/debugfs.h>
+
 #define DISCOVERY_TMR_SIZE      (10 << 10)
 #define DISCOVERY_TMR_OFFSET    (64 << 10)
 
+struct ip_discovery_top;
+
+struct amdgpu_discovery_info {
+	struct debugfs_blob_wrapper debugfs_blob;
+	struct ip_discovery_top *ip_top;
+	uint32_t size;
+	uint8_t *bin;
+	bool reserve_tmr;
+};
+
 void amdgpu_discovery_fini(struct amdgpu_device *adev);
 int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev);
 

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

@@ -332,8 +332,6 @@ int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
 		if (crtc->enabled)
 			active = true;
 
-	pm_runtime_mark_last_busy(dev->dev);
-
 	adev = drm_to_adev(dev);
 	/* if we have active crtcs and we don't have a power ref,
 	 * take the current one
@@ -1365,6 +1363,64 @@ static const struct drm_prop_enum_list amdgpu_dither_enum_list[] = {
 	{ AMDGPU_FMT_DITHER_ENABLE, "on" },
 };
 
+/**
+ * DOC: property for adaptive backlight modulation
+ *
+ * The 'adaptive backlight modulation' property is used for the compositor to
+ * directly control the adaptive backlight modulation power savings feature
+ * that is part of DCN hardware.
+ *
+ * The property will be attached specifically to eDP panels that support it.
+ *
+ * The property is by default set to 'sysfs' to allow the sysfs file 'panel_power_savings'
+ * to be able to control it.
+ * If set to 'off' the compositor will ensure it stays off.
+ * The other values 'min', 'bias min', 'bias max', and 'max' will control the
+ * intensity of the power savings.
+ *
+ * Modifying this value can have implications on color accuracy, so tread
+ * carefully.
+ */
+static int amdgpu_display_setup_abm_prop(struct amdgpu_device *adev)
+{
+	const struct drm_prop_enum_list props[] = {
+		{ ABM_SYSFS_CONTROL, "sysfs" },
+		{ ABM_LEVEL_OFF, "off" },
+		{ ABM_LEVEL_MIN, "min" },
+		{ ABM_LEVEL_BIAS_MIN, "bias min" },
+		{ ABM_LEVEL_BIAS_MAX, "bias max" },
+		{ ABM_LEVEL_MAX, "max" },
+	};
+	struct drm_property *prop;
+	int i;
+
+	if (!adev->dc_enabled)
+		return 0;
+
+	prop = drm_property_create(adev_to_drm(adev), DRM_MODE_PROP_ENUM,
+				"adaptive backlight modulation",
+				6);
+	if (!prop)
+		return -ENOMEM;
+
+	for (i = 0; i < ARRAY_SIZE(props); i++) {
+		int ret;
+
+		ret = drm_property_add_enum(prop, props[i].type,
+						props[i].name);
+
+		if (ret) {
+			drm_property_destroy(adev_to_drm(adev), prop);
+
+			return ret;
+		}
+	}
+
+	adev->mode_info.abm_level_property = prop;
+
+	return 0;
+}
+
 int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
 {
 	int sz;
@@ -1411,7 +1467,7 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
 					 "dither",
 					 amdgpu_dither_enum_list, sz);
 
-	return 0;
+	return amdgpu_display_setup_abm_prop(adev);
 }
 
 void amdgpu_display_update_priority(struct amdgpu_device *adev)

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

@@ -55,4 +55,11 @@ int amdgpu_display_resume_helper(struct amdgpu_device *adev);
 int amdgpu_display_get_scanout_buffer(struct drm_plane *plane,
 				      struct drm_scanout_buffer *sb);
 
+#define ABM_SYSFS_CONTROL	-1
+#define ABM_LEVEL_OFF		0
+#define ABM_LEVEL_MIN		1
+#define ABM_LEVEL_BIAS_MIN	2
+#define ABM_LEVEL_BIAS_MAX	3
+#define ABM_LEVEL_MAX		4
+
 #endif