drm next fixes for 6.18-rc1

amdgpu:
 - DC DCE6 fixes
 - GPU reset fixes
 - Secure diplay messaging cleanup
 - MES fix
 - GPUVM locking fixes
 - PMFW messaging cleanup
 - PCI US/DS switch handling fix
 - VCN queue reset fix
 - DC FPU handling fix
 - DCN 3.5 fix
 - DC mirroring fix
 
 amdkfd:
 - Fix kfd process ref leak
 - mmap write lock handling fix
 - Fix comments in IOCTL
 
 xe:
 - Fix build with clang 16
 - Fix handling of invalid configfs syntax usage and spell out the
   expected syntax in the documentation
 - Do not try late bind firmware when running as VF since it
   shouldn't handle firmware loading
 - Fix idle assertion for local BOs
 - Fix uninitialized variable for late binding
 - Do not require perfmon_capable to expose free memory at page
   granularity. Handle it like other drm drivers do
 - Fix lock handling on suspend error path
 - Fix I2C controller resume after S3
 
 v3d:
 - fix fence locking
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEEEKbZHaGwW9KfbeusDHTzWXnEhr4FAmjpaS4ACgkQDHTzWXnE
 hr5U7xAAgBZ6kqMGN6EKKloDlpIgpQKF3pXGMo7NfwTTJ3x6jTGx41g74ohKCFn2
 Zms0RbHTjGB+u495p3dadOSwbvSW2vTYWUCMnxAWye8s7A4LA7SCTXFY84gEE8ki
 8THR7AU1L78MWILas2wGmDzoy2HNGLITLbcZO1B5Cdq8SxxxKsXJXXZahLjBB67C
 4a23OlANL16g5NCKEQ6owuRBf5hlyc6A+vQuBcfBbKg2VL83Iv/P6nt3Lv5uMwv6
 Qss3lnUrsNBrBMxx3QP8IGG3/EwHU/YcJYIYPbce6qZrxuG8Vw1hTPdJqr8kq9mv
 8wg8JC9VQtPrTuAiNCgC9QrvaDW1vMc/5abYQMJBRvexnXf4VjriSH0WM7YXFQcB
 JvChn5Vo8Sr+1B7oQUM5SOMXHxGfxsL3ypub4a3Die6G6ymhp6cDNwBq0vbU/LXy
 s1fddlNM/PrUACKcFaEGVZMDyz75QTMWae5OmcuUizdzHChUzlDhq1MbHrh0D2jP
 mNuORIHnrO+oZOyxpnQr5YBfZk+xFrnvSLI/hJP982m9FKJ4PVjLg+4YlRNjW4wO
 k4LjxyMutybAFEH0nEWrdGxry+y6EM/OrXpToXFquNq3NdI133NvDRiH+f4SI+8M
 Y6cZdwVSiO5fd7Zup0xKBi+jUi66cfkKS+3MXGfMDX7y1orMMqQ=
 =tSAX
 -----END PGP SIGNATURE-----

Merge tag 'drm-next-2025-10-11-1' of https://gitlab.freedesktop.org/drm/kernel

Pull more drm fixes from Dave Airlie:
 "Just the follow up fixes for rc1 from the next branch, amdgpu and xe
  mostly with a single v3d fix in there.

  amdgpu:
   - DC DCE6 fixes
   - GPU reset fixes
   - Secure diplay messaging cleanup
   - MES fix
   - GPUVM locking fixes
   - PMFW messaging cleanup
   - PCI US/DS switch handling fix
   - VCN queue reset fix
   - DC FPU handling fix
   - DCN 3.5 fix
   - DC mirroring fix

  amdkfd:
   - Fix kfd process ref leak
   - mmap write lock handling fix
   - Fix comments in IOCTL

  xe:
   - Fix build with clang 16
   - Fix handling of invalid configfs syntax usage and spell out the
     expected syntax in the documentation
   - Do not try late bind firmware when running as VF since it shouldn't
     handle firmware loading
   - Fix idle assertion for local BOs
   - Fix uninitialized variable for late binding
   - Do not require perfmon_capable to expose free memory at page
     granularity. Handle it like other drm drivers do
   - Fix lock handling on suspend error path
   - Fix I2C controller resume after S3

  v3d:
   - fix fence locking"

* tag 'drm-next-2025-10-11-1' of https://gitlab.freedesktop.org/drm/kernel: (34 commits)
  drm/amd/display: Incorrect Mirror Cositing
  drm/amd/display: Enable Dynamic DTBCLK Switch
  drm/amdgpu: Report individual reset error
  drm/amdgpu: partially revert "revert to old status lock handling v3"
  drm/amd/display: Fix unsafe uses of kernel mode FPU
  drm/amd/pm: Disable VCN queue reset on SMU v13.0.6 due to regression
  drm/amdgpu: Fix general protection fault in amdgpu_vm_bo_reset_state_machine
  drm/amdgpu: Check swus/ds for switch state save
  drm/amdkfd: Fix two comments in kfd_ioctl.h
  drm/amd/pm: Avoid interface mismatch messaging
  drm/amdgpu: Merge amdgpu_vm_set_pasid into amdgpu_vm_init
  drm/amd/amdgpu: Fix the mes version that support inv_tlbs
  drm/amd: Check whether secure display TA loaded successfully
  drm/amdkfd: Fix mmap write lock not release
  drm/amdkfd: Fix kfd process ref leaking when userptr unmapping
  drm/amdgpu: Fix for GPU reset being blocked by KIQ I/O.
  drm/amd/display: Disable scaling on DCE6 for now
  drm/amd/display: Properly disable scaling on DCE6
  drm/amd/display: Properly clear SCL_*_FILTER_CONTROL on DCE6
  drm/amd/display: Add missing DCE6 SCL_HORZ_FILTER_INIT* SRIs
  ...

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

@@ -2586,12 +2586,17 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 			 * from the KFD, trigger a segmentation fault in VM debug mode.
 			 */
 			if (amdgpu_ttm_adev(bo->tbo.bdev)->debug_vm_userptr) {
+				struct kfd_process *p;
+
 				pr_err("Pid %d unmapped memory before destroying userptr at GPU addr 0x%llx\n",
 								pid_nr(process_info->pid), mem->va);
 
 				// Send GPU VM fault to user space
-				kfd_signal_vm_fault_event_with_userptr(kfd_lookup_process_by_pid(process_info->pid),
+				p = kfd_lookup_process_by_pid(process_info->pid);
-								mem->va);
+				if (p) {
+					kfd_signal_vm_fault_event_with_userptr(p, mem->va);
+					kfd_unref_process(p);
+				}
 			}
 
 			ret = 0;

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

@@ -6389,23 +6389,28 @@ static int amdgpu_device_sched_resume(struct list_head *device_list,
 		if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled)
 			drm_helper_resume_force_mode(adev_to_drm(tmp_adev));
 
-		if (tmp_adev->asic_reset_res)
+		if (tmp_adev->asic_reset_res) {
-			r = tmp_adev->asic_reset_res;
-
-		tmp_adev->asic_reset_res = 0;
-
-		if (r) {
 			/* bad news, how to tell it to userspace ?
 			 * for ras error, we should report GPU bad status instead of
 			 * reset failure
 			 */
 			if (reset_context->src != AMDGPU_RESET_SRC_RAS ||
 			    !amdgpu_ras_eeprom_check_err_threshold(tmp_adev))
-				dev_info(tmp_adev->dev, "GPU reset(%d) failed\n",
+				dev_info(
-					atomic_read(&tmp_adev->gpu_reset_counter));
+					tmp_adev->dev,
-			amdgpu_vf_error_put(tmp_adev, AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
+					"GPU reset(%d) failed with error %d \n",
+					atomic_read(
+						&tmp_adev->gpu_reset_counter),
+					tmp_adev->asic_reset_res);
+			amdgpu_vf_error_put(tmp_adev,
+					    AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0,
+					    tmp_adev->asic_reset_res);
+			if (!r)
+				r = tmp_adev->asic_reset_res;
+			tmp_adev->asic_reset_res = 0;
 		} else {
-			dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n", atomic_read(&tmp_adev->gpu_reset_counter));
+			dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n",
+				 atomic_read(&tmp_adev->gpu_reset_counter));
 			if (amdgpu_acpi_smart_shift_update(tmp_adev,
 							   AMDGPU_SS_DEV_D0))
 				dev_warn(tmp_adev->dev,
@@ -7157,28 +7162,35 @@ void amdgpu_pci_resume(struct pci_dev *pdev)
 
 static void amdgpu_device_cache_switch_state(struct amdgpu_device *adev)
 {
-	struct pci_dev *parent = pci_upstream_bridge(adev->pdev);
+	struct pci_dev *swus, *swds;
 	int r;
 
-	if (!parent || parent->vendor != PCI_VENDOR_ID_ATI)
+	swds = pci_upstream_bridge(adev->pdev);
+	if (!swds || swds->vendor != PCI_VENDOR_ID_ATI ||
+	    pci_pcie_type(swds) != PCI_EXP_TYPE_DOWNSTREAM)
+		return;
+	swus = pci_upstream_bridge(swds);
+	if (!swus ||
+	    (swus->vendor != PCI_VENDOR_ID_ATI &&
+	     swus->vendor != PCI_VENDOR_ID_AMD) ||
+	    pci_pcie_type(swus) != PCI_EXP_TYPE_UPSTREAM)
 		return;
 
 	/* If already saved, return */
 	if (adev->pcie_reset_ctx.swus)
 		return;
 	/* Upstream bridge is ATI, assume it's SWUS/DS architecture */
-	r = pci_save_state(parent);
+	r = pci_save_state(swds);
 	if (r)
 		return;
-	adev->pcie_reset_ctx.swds_pcistate = pci_store_saved_state(parent);
+	adev->pcie_reset_ctx.swds_pcistate = pci_store_saved_state(swds);
 
-	parent = pci_upstream_bridge(parent);
+	r = pci_save_state(swus);
-	r = pci_save_state(parent);
 	if (r)
 		return;
-	adev->pcie_reset_ctx.swus_pcistate = pci_store_saved_state(parent);
+	adev->pcie_reset_ctx.swus_pcistate = pci_store_saved_state(swus);
 
-	adev->pcie_reset_ctx.swus = parent;
+	adev->pcie_reset_ctx.swus = swus;
 }
 
 static void amdgpu_device_load_switch_state(struct amdgpu_device *adev)

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

@@ -1102,6 +1102,9 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_
 
 	might_sleep();
 	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_read;
+
 		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
 		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
 	}
@@ -1171,6 +1174,8 @@ void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint3
 
 	might_sleep();
 	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_write;
 
 		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
 		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);

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

@@ -1421,14 +1421,10 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 
 	amdgpu_debugfs_vm_init(file_priv);
 
-	r = amdgpu_vm_init(adev, &fpriv->vm, fpriv->xcp_id);
+	r = amdgpu_vm_init(adev, &fpriv->vm, fpriv->xcp_id, pasid);
 	if (r)
 		goto error_pasid;
 
-	r = amdgpu_vm_set_pasid(adev, &fpriv->vm, pasid);
-	if (r)
-		goto error_vm;
-
 	fpriv->prt_va = amdgpu_vm_bo_add(adev, &fpriv->vm, NULL);
 	if (!fpriv->prt_va) {
 		r = -ENOMEM;
@@ -1468,10 +1464,8 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 	amdgpu_vm_fini(adev, &fpriv->vm);
 
 error_pasid:
-	if (pasid) {
+	if (pasid)
 		amdgpu_pasid_free(pasid);
-		amdgpu_vm_set_pasid(adev, &fpriv->vm, 0);
-	}
 
 	kfree(fpriv);
 

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

@@ -2352,7 +2352,7 @@ static int psp_securedisplay_initialize(struct psp_context *psp)
 	}
 
 	ret = psp_ta_load(psp, &psp->securedisplay_context.context);
-	if (!ret) {
+	if (!ret && !psp->securedisplay_context.context.resp_status) {
 		psp->securedisplay_context.context.initialized = true;
 		mutex_init(&psp->securedisplay_context.mutex);
 	} else

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

@@ -726,12 +726,12 @@ amdgpu_userq_bo_validate(struct amdgpu_device *adev, struct drm_exec *exec,
 	struct amdgpu_bo *bo;
 	int ret;
 
-	spin_lock(&vm->invalidated_lock);
+	spin_lock(&vm->status_lock);
 	while (!list_empty(&vm->invalidated)) {
 		bo_va = list_first_entry(&vm->invalidated,
 					 struct amdgpu_bo_va,
 					 base.vm_status);
-		spin_unlock(&vm->invalidated_lock);
+		spin_unlock(&vm->status_lock);
 
 		bo = bo_va->base.bo;
 		ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 2);
@@ -748,9 +748,9 @@ amdgpu_userq_bo_validate(struct amdgpu_device *adev, struct drm_exec *exec,
 		if (ret)
 			return ret;
 
-		spin_lock(&vm->invalidated_lock);
+		spin_lock(&vm->status_lock);
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	return 0;
 }

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

@@ -138,48 +138,6 @@ static void amdgpu_vm_assert_locked(struct amdgpu_vm *vm)
 	dma_resv_assert_held(vm->root.bo->tbo.base.resv);
 }
 
-/**
- * amdgpu_vm_set_pasid - manage pasid and vm ptr mapping
- *
- * @adev: amdgpu_device pointer
- * @vm: amdgpu_vm pointer
- * @pasid: the pasid the VM is using on this GPU
- *
- * Set the pasid this VM is using on this GPU, can also be used to remove the
- * pasid by passing in zero.
- *
- */
-int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			u32 pasid)
-{
-	int r;
-
-	amdgpu_vm_assert_locked(vm);
-
-	if (vm->pasid == pasid)
-		return 0;
-
-	if (vm->pasid) {
-		r = xa_err(xa_erase_irq(&adev->vm_manager.pasids, vm->pasid));
-		if (r < 0)
-			return r;
-
-		vm->pasid = 0;
-	}
-
-	if (pasid) {
-		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm,
-					GFP_KERNEL));
-		if (r < 0)
-			return r;
-
-		vm->pasid = pasid;
-	}
-
-
-	return 0;
-}
-
 /**
  * amdgpu_vm_bo_evicted - vm_bo is evicted
  *
@@ -195,10 +153,12 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
 
 	vm_bo->moved = true;
 	amdgpu_vm_assert_locked(vm);
+	spin_lock(&vm_bo->vm->status_lock);
 	if (bo->tbo.type == ttm_bo_type_kernel)
 		list_move(&vm_bo->vm_status, &vm->evicted);
 	else
 		list_move_tail(&vm_bo->vm_status, &vm->evicted);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 /**
  * amdgpu_vm_bo_moved - vm_bo is moved
@@ -211,7 +171,9 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
 static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
 {
 	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -225,7 +187,9 @@ static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
 static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
 {
 	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->idle);
+	spin_unlock(&vm_bo->vm->status_lock);
 	vm_bo->moved = false;
 }
 
@@ -239,9 +203,9 @@ static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
 {
-	spin_lock(&vm_bo->vm->invalidated_lock);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->invalidated);
-	spin_unlock(&vm_bo->vm->invalidated_lock);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -254,9 +218,10 @@ static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_evicted_user(struct amdgpu_vm_bo_base *vm_bo)
 {
-	amdgpu_vm_assert_locked(vm_bo->vm);
 	vm_bo->moved = true;
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->evicted_user);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -270,10 +235,13 @@ static void amdgpu_vm_bo_evicted_user(struct amdgpu_vm_bo_base *vm_bo)
 static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
 {
 	amdgpu_vm_assert_locked(vm_bo->vm);
-	if (vm_bo->bo->parent)
+	if (vm_bo->bo->parent) {
+		spin_lock(&vm_bo->vm->status_lock);
 		list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
-	else
+		spin_unlock(&vm_bo->vm->status_lock);
+	} else {
 		amdgpu_vm_bo_idle(vm_bo);
+	}
 }
 
 /**
@@ -287,7 +255,9 @@ static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
 static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
 {
 	amdgpu_vm_assert_locked(vm_bo->vm);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->done);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -301,13 +271,13 @@ static void amdgpu_vm_bo_reset_state_machine(struct amdgpu_vm *vm)
 {
 	struct amdgpu_vm_bo_base *vm_bo, *tmp;
 
-	spin_lock(&vm->invalidated_lock);
+	amdgpu_vm_assert_locked(vm);
+
+	spin_lock(&vm->status_lock);
 	list_splice_init(&vm->done, &vm->invalidated);
 	list_for_each_entry(vm_bo, &vm->invalidated, vm_status)
 		vm_bo->moved = true;
-	spin_unlock(&vm->invalidated_lock);
 
-	amdgpu_vm_assert_locked(vm_bo->vm);
 	list_for_each_entry_safe(vm_bo, tmp, &vm->idle, vm_status) {
 		struct amdgpu_bo *bo = vm_bo->bo;
 
@@ -317,13 +287,14 @@ static void amdgpu_vm_bo_reset_state_machine(struct amdgpu_vm *vm)
 		else if (bo->parent)
 			list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
 	}
+	spin_unlock(&vm->status_lock);
 }
 
 /**
  * amdgpu_vm_update_shared - helper to update shared memory stat
  * @base: base structure for tracking BO usage in a VM
  *
- * Takes the vm stats_lock and updates the shared memory stat. If the basic
+ * Takes the vm status_lock and updates the shared memory stat. If the basic
  * stat changed (e.g. buffer was moved) amdgpu_vm_update_stats need to be called
  * as well.
  */
@@ -336,7 +307,7 @@ static void amdgpu_vm_update_shared(struct amdgpu_vm_bo_base *base)
 	bool shared;
 
 	dma_resv_assert_held(bo->tbo.base.resv);
-	spin_lock(&vm->stats_lock);
+	spin_lock(&vm->status_lock);
 	shared = drm_gem_object_is_shared_for_memory_stats(&bo->tbo.base);
 	if (base->shared != shared) {
 		base->shared = shared;
@@ -348,7 +319,7 @@ static void amdgpu_vm_update_shared(struct amdgpu_vm_bo_base *base)
 			vm->stats[bo_memtype].drm.private += size;
 		}
 	}
-	spin_unlock(&vm->stats_lock);
+	spin_unlock(&vm->status_lock);
 }
 
 /**
@@ -373,7 +344,7 @@ void amdgpu_vm_bo_update_shared(struct amdgpu_bo *bo)
  *        be bo->tbo.resource
  * @sign: if we should add (+1) or subtract (-1) from the stat
  *
- * Caller need to have the vm stats_lock held. Useful for when multiple update
+ * Caller need to have the vm status_lock held. Useful for when multiple update
  * need to happen at the same time.
  */
 static void amdgpu_vm_update_stats_locked(struct amdgpu_vm_bo_base *base,
@@ -401,8 +372,7 @@ static void amdgpu_vm_update_stats_locked(struct amdgpu_vm_bo_base *base,
 		 */
 		if (bo->flags & AMDGPU_GEM_CREATE_DISCARDABLE)
 			vm->stats[res_memtype].drm.purgeable += size;
-		if (!(bo->preferred_domains &
+		if (!(bo->preferred_domains & amdgpu_mem_type_to_domain(res_memtype)))
-		      amdgpu_mem_type_to_domain(res_memtype)))
 			vm->stats[bo_memtype].evicted += size;
 	}
 }
@@ -421,9 +391,9 @@ void amdgpu_vm_update_stats(struct amdgpu_vm_bo_base *base,
 {
 	struct amdgpu_vm *vm = base->vm;
 
-	spin_lock(&vm->stats_lock);
+	spin_lock(&vm->status_lock);
 	amdgpu_vm_update_stats_locked(base, res, sign);
-	spin_unlock(&vm->stats_lock);
+	spin_unlock(&vm->status_lock);
 }
 
 /**
@@ -449,10 +419,10 @@ void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base,
 	base->next = bo->vm_bo;
 	bo->vm_bo = base;
 
-	spin_lock(&vm->stats_lock);
+	spin_lock(&vm->status_lock);
 	base->shared = drm_gem_object_is_shared_for_memory_stats(&bo->tbo.base);
 	amdgpu_vm_update_stats_locked(base, bo->tbo.resource, +1);
-	spin_unlock(&vm->stats_lock);
+	spin_unlock(&vm->status_lock);
 
 	if (!amdgpu_vm_is_bo_always_valid(vm, bo))
 		return;
@@ -511,10 +481,10 @@ int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
 	int ret;
 
 	/* We can only trust prev->next while holding the lock */
-	spin_lock(&vm->invalidated_lock);
+	spin_lock(&vm->status_lock);
 	while (!list_is_head(prev->next, &vm->done)) {
 		bo_va = list_entry(prev->next, typeof(*bo_va), base.vm_status);
-		spin_unlock(&vm->invalidated_lock);
+		spin_unlock(&vm->status_lock);
 
 		bo = bo_va->base.bo;
 		if (bo) {
@@ -522,10 +492,10 @@ int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
 			if (unlikely(ret))
 				return ret;
 		}
-		spin_lock(&vm->invalidated_lock);
+		spin_lock(&vm->status_lock);
 		prev = prev->next;
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	return 0;
 }
@@ -621,7 +591,7 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		       void *param)
 {
 	uint64_t new_vm_generation = amdgpu_vm_generation(adev, vm);
-	struct amdgpu_vm_bo_base *bo_base, *tmp;
+	struct amdgpu_vm_bo_base *bo_base;
 	struct amdgpu_bo *bo;
 	int r;
 
@@ -634,7 +604,13 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			return r;
 	}
 
-	list_for_each_entry_safe(bo_base, tmp, &vm->evicted, vm_status) {
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->evicted)) {
+		bo_base = list_first_entry(&vm->evicted,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		spin_unlock(&vm->status_lock);
+
 		bo = bo_base->bo;
 
 		r = validate(param, bo);
@@ -647,11 +623,14 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			vm->update_funcs->map_table(to_amdgpu_bo_vm(bo));
 			amdgpu_vm_bo_relocated(bo_base);
 		}
+		spin_lock(&vm->status_lock);
 	}
+	while (ticket && !list_empty(&vm->evicted_user)) {
+		bo_base = list_first_entry(&vm->evicted_user,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		spin_unlock(&vm->status_lock);
 
-	if (ticket) {
-		list_for_each_entry_safe(bo_base, tmp, &vm->evicted_user,
-					 vm_status) {
 		bo = bo_base->bo;
 		dma_resv_assert_held(bo->tbo.base.resv);
 
@@ -660,8 +639,10 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			return r;
 
 		amdgpu_vm_bo_invalidated(bo_base);
+
+		spin_lock(&vm->status_lock);
 	}
-	}
+	spin_unlock(&vm->status_lock);
 
 	amdgpu_vm_eviction_lock(vm);
 	vm->evicting = false;
@@ -690,7 +671,9 @@ bool amdgpu_vm_ready(struct amdgpu_vm *vm)
 	ret = !vm->evicting;
 	amdgpu_vm_eviction_unlock(vm);
 
+	spin_lock(&vm->status_lock);
 	ret &= list_empty(&vm->evicted);
+	spin_unlock(&vm->status_lock);
 
 	spin_lock(&vm->immediate.lock);
 	ret &= !vm->immediate.stopped;
@@ -981,13 +964,18 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 			  struct amdgpu_vm *vm, bool immediate)
 {
 	struct amdgpu_vm_update_params params;
-	struct amdgpu_vm_bo_base *entry, *tmp;
+	struct amdgpu_vm_bo_base *entry;
 	bool flush_tlb_needed = false;
+	LIST_HEAD(relocated);
 	int r, idx;
 
 	amdgpu_vm_assert_locked(vm);
 
-	if (list_empty(&vm->relocated))
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->relocated, &relocated);
+	spin_unlock(&vm->status_lock);
+
+	if (list_empty(&relocated))
 		return 0;
 
 	if (!drm_dev_enter(adev_to_drm(adev), &idx))
@@ -1003,7 +991,7 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
-	list_for_each_entry(entry, &vm->relocated, vm_status) {
+	list_for_each_entry(entry, &relocated, vm_status) {
 		/* vm_flush_needed after updating moved PDEs */
 		flush_tlb_needed |= entry->moved;
 
@@ -1019,7 +1007,9 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	if (flush_tlb_needed)
 		atomic64_inc(&vm->tlb_seq);
 
-	list_for_each_entry_safe(entry, tmp, &vm->relocated, vm_status) {
+	while (!list_empty(&relocated)) {
+		entry = list_first_entry(&relocated, struct amdgpu_vm_bo_base,
+					 vm_status);
 		amdgpu_vm_bo_idle(entry);
 	}
 
@@ -1246,9 +1236,9 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
 			  struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM])
 {
-	spin_lock(&vm->stats_lock);
+	spin_lock(&vm->status_lock);
 	memcpy(stats, vm->stats, sizeof(*stats) * __AMDGPU_PL_NUM);
-	spin_unlock(&vm->stats_lock);
+	spin_unlock(&vm->status_lock);
 }
 
 /**
@@ -1615,24 +1605,29 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 			   struct amdgpu_vm *vm,
 			   struct ww_acquire_ctx *ticket)
 {
-	struct amdgpu_bo_va *bo_va, *tmp;
+	struct amdgpu_bo_va *bo_va;
 	struct dma_resv *resv;
 	bool clear, unlock;
 	int r;
 
-	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->moved)) {
+		bo_va = list_first_entry(&vm->moved, struct amdgpu_bo_va,
+					 base.vm_status);
+		spin_unlock(&vm->status_lock);
+
 		/* Per VM BOs never need to bo cleared in the page tables */
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
+		spin_lock(&vm->status_lock);
 	}
 
-	spin_lock(&vm->invalidated_lock);
 	while (!list_empty(&vm->invalidated)) {
 		bo_va = list_first_entry(&vm->invalidated, struct amdgpu_bo_va,
 					 base.vm_status);
 		resv = bo_va->base.bo->tbo.base.resv;
-		spin_unlock(&vm->invalidated_lock);
+		spin_unlock(&vm->status_lock);
 
 		/* Try to reserve the BO to avoid clearing its ptes */
 		if (!adev->debug_vm && dma_resv_trylock(resv)) {
@@ -1664,9 +1659,9 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 		     bo_va->base.bo->tbo.resource->mem_type == TTM_PL_SYSTEM))
 			amdgpu_vm_bo_evicted_user(&bo_va->base);
 
-		spin_lock(&vm->invalidated_lock);
+		spin_lock(&vm->status_lock);
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	return 0;
 }
@@ -2195,9 +2190,9 @@ void amdgpu_vm_bo_del(struct amdgpu_device *adev,
 		}
 	}
 
-	spin_lock(&vm->invalidated_lock);
+	spin_lock(&vm->status_lock);
 	list_del(&bo_va->base.vm_status);
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
 		list_del(&mapping->list);
@@ -2305,10 +2300,10 @@ void amdgpu_vm_bo_move(struct amdgpu_bo *bo, struct ttm_resource *new_mem,
 	for (bo_base = bo->vm_bo; bo_base; bo_base = bo_base->next) {
 		struct amdgpu_vm *vm = bo_base->vm;
 
-		spin_lock(&vm->stats_lock);
+		spin_lock(&vm->status_lock);
 		amdgpu_vm_update_stats_locked(bo_base, bo->tbo.resource, -1);
 		amdgpu_vm_update_stats_locked(bo_base, new_mem, +1);
-		spin_unlock(&vm->stats_lock);
+		spin_unlock(&vm->status_lock);
 	}
 
 	amdgpu_vm_bo_invalidate(bo, evicted);
@@ -2554,6 +2549,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
  * @adev: amdgpu_device pointer
  * @vm: requested vm
  * @xcp_id: GPU partition selection id
+ * @pasid: the pasid the VM is using on this GPU
  *
  * Init @vm fields.
  *
@@ -2561,7 +2557,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
  * 0 for success, error for failure.
  */
 int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		   int32_t xcp_id)
+		   int32_t xcp_id, uint32_t pasid)
 {
 	struct amdgpu_bo *root_bo;
 	struct amdgpu_bo_vm *root;
@@ -2575,12 +2571,11 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	INIT_LIST_HEAD(&vm->relocated);
 	INIT_LIST_HEAD(&vm->moved);
 	INIT_LIST_HEAD(&vm->idle);
-	spin_lock_init(&vm->invalidated_lock);
 	INIT_LIST_HEAD(&vm->invalidated);
+	spin_lock_init(&vm->status_lock);
 	INIT_LIST_HEAD(&vm->freed);
 	INIT_LIST_HEAD(&vm->done);
 	INIT_KFIFO(vm->faults);
-	spin_lock_init(&vm->stats_lock);
 
 	r = amdgpu_vm_init_entities(adev, vm);
 	if (r)
@@ -2638,12 +2633,26 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		dev_dbg(adev->dev, "Failed to create task info for VM\n");
 
+	/* Store new PASID in XArray (if non-zero) */
+	if (pasid != 0) {
+		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm, GFP_KERNEL));
+		if (r < 0)
+			goto error_free_root;
+
+		vm->pasid = pasid;
+	}
+
 	amdgpu_bo_unreserve(vm->root.bo);
 	amdgpu_bo_unref(&root_bo);
 
 	return 0;
 
 error_free_root:
+	/* If PASID was partially set, erase it from XArray before failing */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
 	amdgpu_vm_pt_free_root(adev, vm);
 	amdgpu_bo_unreserve(vm->root.bo);
 	amdgpu_bo_unref(&root_bo);
@@ -2749,7 +2758,11 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 	root = amdgpu_bo_ref(vm->root.bo);
 	amdgpu_bo_reserve(root, true);
-	amdgpu_vm_set_pasid(adev, vm, 0);
+	/* Remove PASID mapping before destroying VM */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
 	dma_fence_wait(vm->last_unlocked, false);
 	dma_fence_put(vm->last_unlocked);
 	dma_fence_wait(vm->last_tlb_flush, false);
@@ -3038,6 +3051,7 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 
 	amdgpu_vm_assert_locked(vm);
 
+	spin_lock(&vm->status_lock);
 	seq_puts(m, "\tIdle BOs:\n");
 	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
 		if (!bo_va->base.bo)
@@ -3075,13 +3089,11 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 	id = 0;
 
 	seq_puts(m, "\tInvalidated BOs:\n");
-	spin_lock(&vm->invalidated_lock);
 	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
 		if (!bo_va->base.bo)
 			continue;
 		total_invalidated += amdgpu_bo_print_info(id++,	bo_va->base.bo, m);
 	}
-	spin_unlock(&vm->invalidated_lock);
 	total_invalidated_objs = id;
 	id = 0;
 
@@ -3091,6 +3103,7 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 			continue;
 		total_done += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
 	}
+	spin_unlock(&vm->status_lock);
 	total_done_objs = id;
 
 	seq_printf(m, "\tTotal idle size:        %12lld\tobjs:\t%d\n", total_idle,

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

@@ -203,11 +203,11 @@ struct amdgpu_vm_bo_base {
 	/* protected by bo being reserved */
 	struct amdgpu_vm_bo_base	*next;
 
-	/* protected by vm reservation and invalidated_lock */
+	/* protected by vm status_lock */
 	struct list_head		vm_status;
 
 	/* if the bo is counted as shared in mem stats
-	 * protected by vm BO being reserved */
+	 * protected by vm status_lock */
 	bool				shared;
 
 	/* protected by the BO being reserved */
@@ -343,8 +343,10 @@ struct amdgpu_vm {
 	bool			evicting;
 	unsigned int		saved_flags;
 
-	/* Memory statistics for this vm, protected by stats_lock */
+	/* Lock to protect vm_bo add/del/move on all lists of vm */
-	spinlock_t		stats_lock;
+	spinlock_t		status_lock;
+
+	/* Memory statistics for this vm, protected by status_lock */
 	struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM];
 
 	/*
@@ -352,8 +354,6 @@ struct amdgpu_vm {
 	 * PDs, PTs or per VM BOs. The state transits are:
 	 *
 	 * evicted -> relocated (PDs, PTs) or moved (per VM BOs) -> idle
-	 *
-	 * Lists are protected by the root PD dma_resv lock.
 	 */
 
 	/* Per-VM and PT BOs who needs a validation */
@@ -374,10 +374,7 @@ struct amdgpu_vm {
 	 * state transits are:
 	 *
 	 * evicted_user or invalidated -> done
-	 *
-	 * Lists are protected by the invalidated_lock.
 	 */
-	spinlock_t		invalidated_lock;
 
 	/* BOs for user mode queues that need a validation */
 	struct list_head	evicted_user;
@@ -503,11 +500,8 @@ extern const struct amdgpu_vm_update_funcs amdgpu_vm_sdma_funcs;
 void amdgpu_vm_manager_init(struct amdgpu_device *adev);
 void amdgpu_vm_manager_fini(struct amdgpu_device *adev);
 
-int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			u32 pasid);
-
 long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout);
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id, uint32_t pasid);
 int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,

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

@@ -543,7 +543,9 @@ static void amdgpu_vm_pt_free(struct amdgpu_vm_bo_base *entry)
 	entry->bo->vm_bo = NULL;
 	ttm_bo_set_bulk_move(&entry->bo->tbo, NULL);
 
+	spin_lock(&entry->vm->status_lock);
 	list_del(&entry->vm_status);
+	spin_unlock(&entry->vm->status_lock);
 	amdgpu_bo_unref(&entry->bo);
 }
 
@@ -587,6 +589,7 @@ static void amdgpu_vm_pt_add_list(struct amdgpu_vm_update_params *params,
 	struct amdgpu_vm_pt_cursor seek;
 	struct amdgpu_vm_bo_base *entry;
 
+	spin_lock(&params->vm->status_lock);
 	for_each_amdgpu_vm_pt_dfs_safe(params->adev, params->vm, cursor, seek, entry) {
 		if (entry && entry->bo)
 			list_move(&entry->vm_status, &params->tlb_flush_waitlist);
@@ -594,6 +597,7 @@ static void amdgpu_vm_pt_add_list(struct amdgpu_vm_update_params *params,
 
 	/* enter start node now */
 	list_move(&cursor->entry->vm_status, &params->tlb_flush_waitlist);
+	spin_unlock(&params->vm->status_lock);
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c

@@ -337,7 +337,7 @@ static void gmc_v12_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 	int vmid, i;
 
 	if (adev->enable_uni_mes && adev->mes.ring[AMDGPU_MES_SCHED_PIPE].sched.ready &&
-	    (adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x83) {
+	    (adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x84) {
 		struct mes_inv_tlbs_pasid_input input = {0};
 		input.pasid = pasid;
 		input.flush_type = flush_type;

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

@@ -3045,6 +3045,8 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
 	if (svms->checkpoint_ts[gpuidx] != 0) {
 		if (amdgpu_ih_ts_after_or_equal(ts,  svms->checkpoint_ts[gpuidx])) {
 			pr_debug("draining retry fault, drop fault 0x%llx\n", addr);
+			if (write_locked)
+				mmap_write_downgrade(mm);
 			r = -EAGAIN;
 			goto out_unlock_svms;
 		} else {

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

@@ -2000,6 +2000,10 @@ static int amdgpu_dm_init(struct amdgpu_device *adev)
 
 	init_data.flags.disable_ips_in_vpb = 0;
 
+	/* DCN35 and above supports dynamic DTBCLK switch */
+	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 5, 0))
+		init_data.flags.allow_0_dtb_clk = true;
+
 	/* Enable DWB for tested platforms only */
 	if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 0, 0))
 		init_data.num_virtual_links = 1;

drivers/gpu/drm/amd/display/dc/dce/dce_transform.c

@@ -154,10 +154,13 @@ static bool dce60_setup_scaling_configuration(
 	REG_SET(SCL_BYPASS_CONTROL, 0, SCL_BYPASS_MODE, 0);
 
 	if (data->taps.h_taps + data->taps.v_taps <= 2) {
-		/* Set bypass */
+		/* Disable scaler functionality */
-
+		REG_WRITE(SCL_SCALER_ENABLE, 0);
-		/* DCE6 has no SCL_MODE register, skip scale mode programming */
 
+		/* Clear registers that can cause glitches even when the scaler is off */
+		REG_WRITE(SCL_TAP_CONTROL, 0);
+		REG_WRITE(SCL_AUTOMATIC_MODE_CONTROL, 0);
+		REG_WRITE(SCL_F_SHARP_CONTROL, 0);
 		return false;
 	}
 
@@ -165,7 +168,7 @@ static bool dce60_setup_scaling_configuration(
 			SCL_H_NUM_OF_TAPS, data->taps.h_taps - 1,
 			SCL_V_NUM_OF_TAPS, data->taps.v_taps - 1);
 
-	/* DCE6 has no SCL_MODE register, skip scale mode programming */
+	REG_WRITE(SCL_SCALER_ENABLE, 1);
 
 	/* DCE6 has no SCL_BOUNDARY_MODE bit, skip replace out of bound pixels */
 
@@ -502,6 +505,8 @@ static void dce60_transform_set_scaler(
 	REG_SET(DC_LB_MEM_SIZE, 0,
 		DC_LB_MEM_SIZE, xfm_dce->lb_memory_size);
 
+	REG_WRITE(SCL_UPDATE, 0x00010000);
+
 	/* Clear SCL_F_SHARP_CONTROL value to 0 */
 	REG_WRITE(SCL_F_SHARP_CONTROL, 0);
 
@@ -527,8 +532,7 @@ static void dce60_transform_set_scaler(
 		if (coeffs_v != xfm_dce->filter_v || coeffs_h != xfm_dce->filter_h) {
 			/* 4. Program vertical filters */
 			if (xfm_dce->filter_v == NULL)
-				REG_SET(SCL_VERT_FILTER_CONTROL, 0,
+				REG_WRITE(SCL_VERT_FILTER_CONTROL, 0);
-						SCL_V_2TAP_HARDCODE_COEF_EN, 0);
 			program_multi_taps_filter(
 					xfm_dce,
 					data->taps.v_taps,
@@ -542,8 +546,7 @@ static void dce60_transform_set_scaler(
 
 			/* 5. Program horizontal filters */
 			if (xfm_dce->filter_h == NULL)
-				REG_SET(SCL_HORZ_FILTER_CONTROL, 0,
+				REG_WRITE(SCL_HORZ_FILTER_CONTROL, 0);
-						SCL_H_2TAP_HARDCODE_COEF_EN, 0);
 			program_multi_taps_filter(
 					xfm_dce,
 					data->taps.h_taps,
@@ -566,6 +569,8 @@ static void dce60_transform_set_scaler(
 	/* DCE6 has no SCL_COEF_UPDATE_COMPLETE bit to flip to new coefficient memory */
 
 	/* DCE6 DATA_FORMAT register does not support ALPHA_EN */
+
+	REG_WRITE(SCL_UPDATE, 0);
 }
 #endif
 

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

@@ -155,6 +155,9 @@
 	SRI(SCL_COEF_RAM_TAP_DATA, SCL, id), \
 	SRI(VIEWPORT_START, SCL, id), \
 	SRI(VIEWPORT_SIZE, SCL, id), \
+	SRI(SCL_SCALER_ENABLE, SCL, id), \
+	SRI(SCL_HORZ_FILTER_INIT_RGB_LUMA, SCL, id), \
+	SRI(SCL_HORZ_FILTER_INIT_CHROMA, SCL, id), \
 	SRI(SCL_HORZ_FILTER_SCALE_RATIO, SCL, id), \
 	SRI(SCL_VERT_FILTER_SCALE_RATIO, SCL, id), \
 	SRI(SCL_VERT_FILTER_INIT, SCL, id), \
@@ -590,6 +593,7 @@ struct dce_transform_registers {
 	uint32_t SCL_VERT_FILTER_SCALE_RATIO;
 	uint32_t SCL_HORZ_FILTER_INIT;
 #if defined(CONFIG_DRM_AMD_DC_SI)
+	uint32_t SCL_SCALER_ENABLE;
 	uint32_t SCL_HORZ_FILTER_INIT_RGB_LUMA;
 	uint32_t SCL_HORZ_FILTER_INIT_CHROMA;
 #endif

drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.c

@@ -808,6 +808,8 @@ void dcn316_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_param
 
 int dcn_get_max_non_odm_pix_rate_100hz(struct _vcs_dpi_soc_bounding_box_st *soc)
 {
+	dc_assert_fp_enabled();
+
 	return soc->clock_limits[0].dispclk_mhz * 10000.0 / (1.0 + soc->dcn_downspread_percent / 100.0);
 }
 
@@ -815,6 +817,8 @@ int dcn_get_approx_det_segs_required_for_pstate(
 		struct _vcs_dpi_soc_bounding_box_st *soc,
 		int pix_clk_100hz, int bpp, int seg_size_kb)
 {
+	dc_assert_fp_enabled();
+
 	/* Roughly calculate required crb to hide latency. In practice there is slightly
 	 * more buffer available for latency hiding
 	 */

drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c

@@ -445,6 +445,8 @@ int dcn35_populate_dml_pipes_from_context_fpu(struct dc *dc,
 	bool upscaled = false;
 	const unsigned int max_allowed_vblank_nom = 1023;
 
+	dc_assert_fp_enabled();
+
 	dcn31_populate_dml_pipes_from_context(dc, context, pipes,
 					      validate_mode);
 
@@ -498,9 +500,7 @@ int dcn35_populate_dml_pipes_from_context_fpu(struct dc *dc,
 
 		pipes[pipe_cnt].pipe.src.unbounded_req_mode = false;
 
-		DC_FP_START();
 		dcn31_zero_pipe_dcc_fraction(pipes, pipe_cnt);
-		DC_FP_END();
 
 		pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch;
 		pipes[pipe_cnt].pipe.src.dcc_rate = 3;
@@ -581,6 +581,8 @@ void dcn35_decide_zstate_support(struct dc *dc, struct dc_state *context)
 	unsigned int i, plane_count = 0;
 	DC_LOGGER_INIT(dc->ctx->logger);
 
+	dc_assert_fp_enabled();
+
 	for (i = 0; i < dc->res_pool->pipe_count; i++) {
 		if (context->res_ctx.pipe_ctx[i].plane_state)
 			plane_count++;

drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c

@@ -478,6 +478,8 @@ int dcn351_populate_dml_pipes_from_context_fpu(struct dc *dc,
 	bool upscaled = false;
 	const unsigned int max_allowed_vblank_nom = 1023;
 
+	dc_assert_fp_enabled();
+
 	dcn31_populate_dml_pipes_from_context(dc, context, pipes,
 					      validate_mode);
 
@@ -531,9 +533,7 @@ int dcn351_populate_dml_pipes_from_context_fpu(struct dc *dc,
 
 		pipes[pipe_cnt].pipe.src.unbounded_req_mode = false;
 
-		DC_FP_START();
 		dcn31_zero_pipe_dcc_fraction(pipes, pipe_cnt);
-		DC_FP_END();
 
 		pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch;
 		pipes[pipe_cnt].pipe.src.dcc_rate = 3;

drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c

@@ -404,13 +404,13 @@ static const struct dc_plane_cap plane_cap = {
 	},
 
 	.max_upscale_factor = {
-			.argb8888 = 16000,
+			.argb8888 = 1,
 			.nv12 = 1,
 			.fp16 = 1
 	},
 
 	.max_downscale_factor = {
-			.argb8888 = 250,
+			.argb8888 = 1,
 			.nv12 = 1,
 			.fp16 = 1
 	}

drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c

@@ -1760,6 +1760,20 @@ enum dc_status dcn35_patch_unknown_plane_state(struct dc_plane_state *plane_stat
 }
 
 
+static int populate_dml_pipes_from_context_fpu(struct dc *dc,
+					       struct dc_state *context,
+					       display_e2e_pipe_params_st *pipes,
+					       enum dc_validate_mode validate_mode)
+{
+	int ret;
+
+	DC_FP_START();
+	ret = dcn35_populate_dml_pipes_from_context_fpu(dc, context, pipes, validate_mode);
+	DC_FP_END();
+
+	return ret;
+}
+
 static struct resource_funcs dcn35_res_pool_funcs = {
 	.destroy = dcn35_destroy_resource_pool,
 	.link_enc_create = dcn35_link_encoder_create,
@@ -1770,7 +1784,7 @@ static struct resource_funcs dcn35_res_pool_funcs = {
 	.validate_bandwidth = dcn35_validate_bandwidth,
 	.calculate_wm_and_dlg = NULL,
 	.update_soc_for_wm_a = dcn31_update_soc_for_wm_a,
-	.populate_dml_pipes = dcn35_populate_dml_pipes_from_context_fpu,
+	.populate_dml_pipes = populate_dml_pipes_from_context_fpu,
 	.acquire_free_pipe_as_secondary_dpp_pipe = dcn20_acquire_free_pipe_for_layer,
 	.release_pipe = dcn20_release_pipe,
 	.add_stream_to_ctx = dcn30_add_stream_to_ctx,

drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c

@@ -1732,6 +1732,21 @@ static enum dc_status dcn351_validate_bandwidth(struct dc *dc,
 	return out ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE;
 }
 
+static int populate_dml_pipes_from_context_fpu(struct dc *dc,
+					       struct dc_state *context,
+					       display_e2e_pipe_params_st *pipes,
+					       enum dc_validate_mode validate_mode)
+{
+	int ret;
+
+	DC_FP_START();
+	ret = dcn351_populate_dml_pipes_from_context_fpu(dc, context, pipes, validate_mode);
+	DC_FP_END();
+
+	return ret;
+
+}
+
 static struct resource_funcs dcn351_res_pool_funcs = {
 	.destroy = dcn351_destroy_resource_pool,
 	.link_enc_create = dcn35_link_encoder_create,
@@ -1742,7 +1757,7 @@ static struct resource_funcs dcn351_res_pool_funcs = {
 	.validate_bandwidth = dcn351_validate_bandwidth,
 	.calculate_wm_and_dlg = NULL,
 	.update_soc_for_wm_a = dcn31_update_soc_for_wm_a,
-	.populate_dml_pipes = dcn351_populate_dml_pipes_from_context_fpu,
+	.populate_dml_pipes = populate_dml_pipes_from_context_fpu,
 	.acquire_free_pipe_as_secondary_dpp_pipe = dcn20_acquire_free_pipe_for_layer,
 	.release_pipe = dcn20_release_pipe,
 	.add_stream_to_ctx = dcn30_add_stream_to_ctx,

drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c

@@ -1734,6 +1734,20 @@ static enum dc_status dcn35_validate_bandwidth(struct dc *dc,
 }
 
 
+static int populate_dml_pipes_from_context_fpu(struct dc *dc,
+					       struct dc_state *context,
+					       display_e2e_pipe_params_st *pipes,
+					       enum dc_validate_mode validate_mode)
+{
+	int ret;
+
+	DC_FP_START();
+	ret = dcn35_populate_dml_pipes_from_context_fpu(dc, context, pipes, validate_mode);
+	DC_FP_END();
+
+	return ret;
+}
+
 static struct resource_funcs dcn36_res_pool_funcs = {
 	.destroy = dcn36_destroy_resource_pool,
 	.link_enc_create = dcn35_link_encoder_create,
@@ -1744,7 +1758,7 @@ static struct resource_funcs dcn36_res_pool_funcs = {
 	.validate_bandwidth = dcn35_validate_bandwidth,
 	.calculate_wm_and_dlg = NULL,
 	.update_soc_for_wm_a = dcn31_update_soc_for_wm_a,
-	.populate_dml_pipes = dcn35_populate_dml_pipes_from_context_fpu,
+	.populate_dml_pipes = populate_dml_pipes_from_context_fpu,
 	.acquire_free_pipe_as_secondary_dpp_pipe = dcn20_acquire_free_pipe_for_layer,
 	.release_pipe = dcn20_release_pipe,
 	.add_stream_to_ctx = dcn30_add_stream_to_ctx,

drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c

@@ -641,16 +641,16 @@ static void spl_calculate_inits_and_viewports(struct spl_in *spl_in,
 		/* this gives the direction of the cositing (negative will move
 		 * left, right otherwise)
 		 */
-		int sign = 1;
+		int h_sign = flip_horz_scan_dir ? -1 : 1;
+		int v_sign = flip_vert_scan_dir ? -1 : 1;
 
 		switch (spl_in->basic_in.cositing) {
-
 		case CHROMA_COSITING_TOPLEFT:
-			init_adj_h = spl_fixpt_from_fraction(sign, 4);
+			init_adj_h = spl_fixpt_from_fraction(h_sign, 4);
-			init_adj_v = spl_fixpt_from_fraction(sign, 4);
+			init_adj_v = spl_fixpt_from_fraction(v_sign, 4);
 			break;
 		case CHROMA_COSITING_LEFT:
-			init_adj_h = spl_fixpt_from_fraction(sign, 4);
+			init_adj_h = spl_fixpt_from_fraction(h_sign, 4);
 			init_adj_v = spl_fixpt_zero;
 			break;
 		case CHROMA_COSITING_NONE:

drivers/gpu/drm/amd/include/asic_reg/dce/dce_6_0_d.h

@@ -4115,6 +4115,7 @@
 #define mmSCL0_SCL_COEF_RAM_CONFLICT_STATUS 0x1B55
 #define mmSCL0_SCL_COEF_RAM_SELECT 0x1B40
 #define mmSCL0_SCL_COEF_RAM_TAP_DATA 0x1B41
+#define mmSCL0_SCL_SCALER_ENABLE 0x1B42
 #define mmSCL0_SCL_CONTROL 0x1B44
 #define mmSCL0_SCL_DEBUG 0x1B6A
 #define mmSCL0_SCL_DEBUG2 0x1B69
@@ -4144,6 +4145,7 @@
 #define mmSCL1_SCL_COEF_RAM_CONFLICT_STATUS 0x1E55
 #define mmSCL1_SCL_COEF_RAM_SELECT 0x1E40
 #define mmSCL1_SCL_COEF_RAM_TAP_DATA 0x1E41
+#define mmSCL1_SCL_SCALER_ENABLE 0x1E42
 #define mmSCL1_SCL_CONTROL 0x1E44
 #define mmSCL1_SCL_DEBUG 0x1E6A
 #define mmSCL1_SCL_DEBUG2 0x1E69
@@ -4173,6 +4175,7 @@
 #define mmSCL2_SCL_COEF_RAM_CONFLICT_STATUS 0x4155
 #define mmSCL2_SCL_COEF_RAM_SELECT 0x4140
 #define mmSCL2_SCL_COEF_RAM_TAP_DATA 0x4141
+#define mmSCL2_SCL_SCALER_ENABLE 0x4142
 #define mmSCL2_SCL_CONTROL 0x4144
 #define mmSCL2_SCL_DEBUG 0x416A
 #define mmSCL2_SCL_DEBUG2 0x4169
@@ -4202,6 +4205,7 @@
 #define mmSCL3_SCL_COEF_RAM_CONFLICT_STATUS 0x4455
 #define mmSCL3_SCL_COEF_RAM_SELECT 0x4440
 #define mmSCL3_SCL_COEF_RAM_TAP_DATA 0x4441
+#define mmSCL3_SCL_SCALER_ENABLE 0x4442
 #define mmSCL3_SCL_CONTROL 0x4444
 #define mmSCL3_SCL_DEBUG 0x446A
 #define mmSCL3_SCL_DEBUG2 0x4469
@@ -4231,6 +4235,7 @@
 #define mmSCL4_SCL_COEF_RAM_CONFLICT_STATUS 0x4755
 #define mmSCL4_SCL_COEF_RAM_SELECT 0x4740
 #define mmSCL4_SCL_COEF_RAM_TAP_DATA 0x4741
+#define mmSCL4_SCL_SCALER_ENABLE 0x4742
 #define mmSCL4_SCL_CONTROL 0x4744
 #define mmSCL4_SCL_DEBUG 0x476A
 #define mmSCL4_SCL_DEBUG2 0x4769
@@ -4260,6 +4265,7 @@
 #define mmSCL5_SCL_COEF_RAM_CONFLICT_STATUS 0x4A55
 #define mmSCL5_SCL_COEF_RAM_SELECT 0x4A40
 #define mmSCL5_SCL_COEF_RAM_TAP_DATA 0x4A41
+#define mmSCL5_SCL_SCALER_ENABLE 0x4A42
 #define mmSCL5_SCL_CONTROL 0x4A44
 #define mmSCL5_SCL_DEBUG 0x4A6A
 #define mmSCL5_SCL_DEBUG2 0x4A69
@@ -4287,6 +4293,7 @@
 #define mmSCL_COEF_RAM_CONFLICT_STATUS 0x1B55
 #define mmSCL_COEF_RAM_SELECT 0x1B40
 #define mmSCL_COEF_RAM_TAP_DATA 0x1B41
+#define mmSCL_SCALER_ENABLE 0x1B42
 #define mmSCL_CONTROL 0x1B44
 #define mmSCL_DEBUG 0x1B6A
 #define mmSCL_DEBUG2 0x1B69

drivers/gpu/drm/amd/include/asic_reg/dce/dce_6_0_sh_mask.h

@@ -8650,6 +8650,8 @@
 #define REGAMMA_LUT_INDEX__REGAMMA_LUT_INDEX__SHIFT 0x00000000
 #define REGAMMA_LUT_WRITE_EN_MASK__REGAMMA_LUT_WRITE_EN_MASK_MASK 0x00000007L
 #define REGAMMA_LUT_WRITE_EN_MASK__REGAMMA_LUT_WRITE_EN_MASK__SHIFT 0x00000000
+#define SCL_SCALER_ENABLE__SCL_SCALE_EN_MASK 0x00000001L
+#define SCL_SCALER_ENABLE__SCL_SCALE_EN__SHIFT 0x00000000
 #define SCL_ALU_CONTROL__SCL_ALU_DISABLE_MASK 0x00000001L
 #define SCL_ALU_CONTROL__SCL_ALU_DISABLE__SHIFT 0x00000000
 #define SCL_BYPASS_CONTROL__SCL_BYPASS_MODE_MASK 0x00000003L

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c

@@ -288,7 +288,8 @@ int smu_v13_0_check_fw_version(struct smu_context *smu)
 	 * Considering above, we just leave user a verbal message instead
 	 * of halt driver loading.
 	 */
-	if (if_version != smu->smc_driver_if_version) {
+	if (smu->smc_driver_if_version != SMU_IGNORE_IF_VERSION &&
+	    if_version != smu->smc_driver_if_version) {
 		dev_info(adev->dev, "smu driver if version = 0x%08x, smu fw if version = 0x%08x, "
 			 "smu fw program = %d, smu fw version = 0x%08x (%d.%d.%d)\n",
 			 smu->smc_driver_if_version, if_version,

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c

@@ -450,8 +450,7 @@ static void smu_v13_0_6_init_caps(struct smu_context *smu)
 	    ((pgm == 4) && (fw_ver >= 0x4557000)))
 		smu_v13_0_6_cap_set(smu, SMU_CAP(SDMA_RESET));
 
-	if (((pgm == 0) && (fw_ver >= 0x00558200)) ||
+	if ((pgm == 0) && (fw_ver >= 0x00558200))
-	    ((pgm == 4) && (fw_ver >= 0x04557100)))
 		smu_v13_0_6_cap_set(smu, SMU_CAP(VCN_RESET));
 }
 
@@ -3933,7 +3932,7 @@ void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu)
 	smu->feature_map = (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12)) ?
 		smu_v13_0_12_feature_mask_map : smu_v13_0_6_feature_mask_map;
 	smu->table_map = smu_v13_0_6_table_map;
-	smu->smc_driver_if_version = SMU13_0_6_DRIVER_IF_VERSION;
+	smu->smc_driver_if_version = SMU_IGNORE_IF_VERSION;
 	smu->smc_fw_caps |= SMU_FW_CAP_RAS_PRI;
 	smu_v13_0_set_smu_mailbox_registers(smu);
 	smu_v13_0_6_set_temp_funcs(smu);

drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h

@@ -40,6 +40,8 @@
 #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_ABNORMAL        0x8
 #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_RECOVERY        0x9
 
+#define SMU_IGNORE_IF_VERSION 0xFFFFFFFF
+
 #define smu_cmn_init_soft_gpu_metrics(ptr, frev, crev)                   \
 	do {                                                             \
 		typecheck(struct gpu_metrics_v##frev##_##crev *, (ptr)); \

drivers/gpu/drm/drm_gpusvm.c

@@ -361,7 +361,6 @@ static const struct mmu_interval_notifier_ops drm_gpusvm_notifier_ops = {
  * @name: Name of the GPU SVM.
  * @drm: Pointer to the DRM device structure.
  * @mm: Pointer to the mm_struct for the address space.
- * @device_private_page_owner: Device private pages owner.
  * @mm_start: Start address of GPU SVM.
  * @mm_range: Range of the GPU SVM.
  * @notifier_size: Size of individual notifiers.
@@ -383,7 +382,7 @@ static const struct mmu_interval_notifier_ops drm_gpusvm_notifier_ops = {
  */
 int drm_gpusvm_init(struct drm_gpusvm *gpusvm,
 		    const char *name, struct drm_device *drm,
-		    struct mm_struct *mm, void *device_private_page_owner,
+		    struct mm_struct *mm,
 		    unsigned long mm_start, unsigned long mm_range,
 		    unsigned long notifier_size,
 		    const struct drm_gpusvm_ops *ops,
@@ -395,15 +394,13 @@ int drm_gpusvm_init(struct drm_gpusvm *gpusvm,
 		mmgrab(mm);
 	} else {
 		/* No full SVM mode, only core drm_gpusvm_pages API. */
-		if (ops || num_chunks || mm_range || notifier_size ||
+		if (ops || num_chunks || mm_range || notifier_size)
-		    device_private_page_owner)
 			return -EINVAL;
 	}
 
 	gpusvm->name = name;
 	gpusvm->drm = drm;
 	gpusvm->mm = mm;
-	gpusvm->device_private_page_owner = device_private_page_owner;
 	gpusvm->mm_start = mm_start;
 	gpusvm->mm_range = mm_range;
 	gpusvm->notifier_size = notifier_size;
@@ -684,6 +681,7 @@ static unsigned int drm_gpusvm_hmm_pfn_to_order(unsigned long hmm_pfn,
  * @notifier: Pointer to the GPU SVM notifier structure
  * @start: Start address
  * @end: End address
+ * @dev_private_owner: The device private page owner
  *
  * Check if pages between start and end have been faulted in on the CPU. Use to
  * prevent migration of pages without CPU backing store.
@@ -692,14 +690,15 @@ static unsigned int drm_gpusvm_hmm_pfn_to_order(unsigned long hmm_pfn,
  */
 static bool drm_gpusvm_check_pages(struct drm_gpusvm *gpusvm,
 				   struct drm_gpusvm_notifier *notifier,
-				   unsigned long start, unsigned long end)
+				   unsigned long start, unsigned long end,
+				   void *dev_private_owner)
 {
 	struct hmm_range hmm_range = {
 		.default_flags = 0,
 		.notifier = &notifier->notifier,
 		.start = start,
 		.end = end,
-		.dev_private_owner = gpusvm->device_private_page_owner,
+		.dev_private_owner = dev_private_owner,
 	};
 	unsigned long timeout =
 		jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
@@ -753,6 +752,7 @@ static bool drm_gpusvm_check_pages(struct drm_gpusvm *gpusvm,
  * @gpuva_start: Start address of GPUVA which mirrors CPU
  * @gpuva_end: End address of GPUVA which mirrors CPU
  * @check_pages_threshold: Check CPU pages for present threshold
+ * @dev_private_owner: The device private page owner
  *
  * This function determines the chunk size for the GPU SVM range based on the
  * fault address, GPU SVM chunk sizes, existing GPU SVM ranges, and the virtual
@@ -767,7 +767,8 @@ drm_gpusvm_range_chunk_size(struct drm_gpusvm *gpusvm,
 			    unsigned long fault_addr,
 			    unsigned long gpuva_start,
 			    unsigned long gpuva_end,
-			    unsigned long check_pages_threshold)
+			    unsigned long check_pages_threshold,
+			    void *dev_private_owner)
 {
 	unsigned long start, end;
 	int i = 0;
@@ -814,7 +815,7 @@ drm_gpusvm_range_chunk_size(struct drm_gpusvm *gpusvm,
 		 * process-many-malloc' mallocs at least 64k at a time.
 		 */
 		if (end - start <= check_pages_threshold &&
-		    !drm_gpusvm_check_pages(gpusvm, notifier, start, end)) {
+		    !drm_gpusvm_check_pages(gpusvm, notifier, start, end, dev_private_owner)) {
 			++i;
 			goto retry;
 		}
@@ -957,7 +958,8 @@ drm_gpusvm_range_find_or_insert(struct drm_gpusvm *gpusvm,
 	chunk_size = drm_gpusvm_range_chunk_size(gpusvm, notifier, vas,
 						 fault_addr, gpuva_start,
 						 gpuva_end,
-						 ctx->check_pages_threshold);
+						 ctx->check_pages_threshold,
+						 ctx->device_private_page_owner);
 	if (chunk_size == LONG_MAX) {
 		err = -EINVAL;
 		goto err_notifier_remove;
@@ -1268,7 +1270,7 @@ int drm_gpusvm_get_pages(struct drm_gpusvm *gpusvm,
 		.notifier = notifier,
 		.start = pages_start,
 		.end = pages_end,
-		.dev_private_owner = gpusvm->device_private_page_owner,
+		.dev_private_owner = ctx->device_private_page_owner,
 	};
 	void *zdd;
 	unsigned long timeout =

drivers/gpu/drm/v3d/v3d_drv.h

@@ -62,6 +62,8 @@ struct v3d_queue_state {
 	/* Currently active job for this queue */
 	struct v3d_job *active_job;
 	spinlock_t queue_lock;
+	/* Protect dma fence for signalling job completion */
+	spinlock_t fence_lock;
 };
 
 /* Performance monitor object. The perform lifetime is controlled by userspace

drivers/gpu/drm/v3d/v3d_fence.c

@@ -15,7 +15,7 @@ struct dma_fence *v3d_fence_create(struct v3d_dev *v3d, enum v3d_queue q)
 	fence->dev = &v3d->drm;
 	fence->queue = q;
 	fence->seqno = ++queue->emit_seqno;
-	dma_fence_init(&fence->base, &v3d_fence_ops, &queue->queue_lock,
+	dma_fence_init(&fence->base, &v3d_fence_ops, &queue->fence_lock,
 		       queue->fence_context, fence->seqno);
 
 	return &fence->base;

drivers/gpu/drm/v3d/v3d_gem.c

@@ -273,6 +273,7 @@ v3d_gem_init(struct drm_device *dev)
 		seqcount_init(&queue->stats.lock);
 
 		spin_lock_init(&queue->queue_lock);
+		spin_lock_init(&queue->fence_lock);
 	}
 
 	spin_lock_init(&v3d->mm_lock);

drivers/gpu/drm/xe/tests/xe_pci.c

@@ -211,15 +211,15 @@ static void xe_ip_kunit_desc(const struct xe_ip *param, char *desc)
  * param generator can be used for both
  */
 static const struct xe_ip pre_gmdid_graphics_ips[] = {
-	graphics_ip_xelp,
+	{ 1200, "Xe_LP", &graphics_xelp },
-	graphics_ip_xelpp,
+	{ 1210, "Xe_LP+", &graphics_xelp },
-	graphics_ip_xehpg,
+	{ 1255, "Xe_HPG", &graphics_xehpg },
-	graphics_ip_xehpc,
+	{ 1260, "Xe_HPC", &graphics_xehpc },
 };
 
 static const struct xe_ip pre_gmdid_media_ips[] = {
-	media_ip_xem,
+	{ 1200, "Xe_M", &media_xem },
-	media_ip_xehpm,
+	{ 1255, "Xe_HPM", &media_xem },
 };
 
 KUNIT_ARRAY_PARAM(pre_gmdid_graphics_ip, pre_gmdid_graphics_ips, xe_ip_kunit_desc);

drivers/gpu/drm/xe/xe_bo.c

@@ -1737,6 +1737,24 @@ static bool should_migrate_to_smem(struct xe_bo *bo)
 	       bo->attr.atomic_access == DRM_XE_ATOMIC_CPU;
 }
 
+static int xe_bo_wait_usage_kernel(struct xe_bo *bo, struct ttm_operation_ctx *ctx)
+{
+	long lerr;
+
+	if (ctx->no_wait_gpu)
+		return dma_resv_test_signaled(bo->ttm.base.resv, DMA_RESV_USAGE_KERNEL) ?
+			0 : -EBUSY;
+
+	lerr = dma_resv_wait_timeout(bo->ttm.base.resv, DMA_RESV_USAGE_KERNEL,
+				     ctx->interruptible, MAX_SCHEDULE_TIMEOUT);
+	if (lerr < 0)
+		return lerr;
+	if (lerr == 0)
+		return -EBUSY;
+
+	return 0;
+}
+
 /* Populate the bo if swapped out, or migrate if the access mode requires that. */
 static int xe_bo_fault_migrate(struct xe_bo *bo, struct ttm_operation_ctx *ctx,
 			       struct drm_exec *exec)
@@ -1745,9 +1763,8 @@ static int xe_bo_fault_migrate(struct xe_bo *bo, struct ttm_operation_ctx *ctx,
 	int err = 0;
 
 	if (ttm_manager_type(tbo->bdev, tbo->resource->mem_type)->use_tt) {
-		xe_assert(xe_bo_device(bo),
+		err = xe_bo_wait_usage_kernel(bo, ctx);
-			  dma_resv_test_signaled(tbo->base.resv, DMA_RESV_USAGE_KERNEL) ||
+		if (!err)
-			  (tbo->ttm && ttm_tt_is_populated(tbo->ttm)));
 			err = ttm_bo_populate(&bo->ttm, ctx);
 	} else if (should_migrate_to_smem(bo)) {
 		xe_assert(xe_bo_device(bo), bo->flags & XE_BO_FLAG_SYSTEM);
@@ -1922,7 +1939,6 @@ static vm_fault_t xe_bo_cpu_fault(struct vm_fault *vmf)
 			.no_wait_gpu = false,
 			.gfp_retry_mayfail = retry_after_wait,
 		};
-		long lerr;
 
 		err = drm_exec_lock_obj(&exec, &tbo->base);
 		drm_exec_retry_on_contention(&exec);
@@ -1942,13 +1958,9 @@ static vm_fault_t xe_bo_cpu_fault(struct vm_fault *vmf)
 			break;
 		}
 
-		lerr = dma_resv_wait_timeout(tbo->base.resv,
+		err = xe_bo_wait_usage_kernel(bo, &tctx);
-					     DMA_RESV_USAGE_KERNEL, true,
+		if (err)
-					     MAX_SCHEDULE_TIMEOUT);
-		if (lerr < 0) {
-			err = lerr;
 			break;
-		}
 
 		if (!retry_after_wait)
 			ret = __xe_bo_cpu_fault(vmf, xe, bo);

drivers/gpu/drm/xe/xe_configfs.c

@@ -126,8 +126,20 @@
  * not intended for normal execution and will taint the kernel with TAINT_TEST
  * when used.
  *
- * Currently this is implemented only for post and mid context restore.
+ * The syntax allows to pass straight instructions to be executed by the engine
- * Examples:
+ * in a batch buffer or set specific registers.
+ *
+ * #. Generic instruction::
+ *
+ *	<engine-class> cmd <instr> [[dword0] [dword1] [...]]
+ *
+ * #. Simple register setting::
+ *
+ *	<engine-class> reg <address> <value>
+ *
+ * Commands are saved per engine class: all instances of that class will execute
+ * those commands during context switch. The instruction, dword arguments,
+ * addresses and values are in hex format like in the examples below.
  *
  * #. Execute a LRI command to write 0xDEADBEEF to register 0x4f10 after the
  *    normal context restore::
@@ -154,7 +166,8 @@
  *       When using multiple lines, make sure to use a command that is
  *       implemented with a single write syscall, like HEREDOC.
  *
- * These attributes can only be set before binding to the device.
+ * Currently this is implemented only for post and mid context restore and
+ * these attributes can only be set before binding to the device.
  *
  * Remove devices
  * ==============
@@ -324,8 +337,8 @@ static const struct engine_info *lookup_engine_info(const char *pattern, u64 *ma
 			continue;
 
 		pattern += strlen(engine_info[i].cls);
-		if (!mask && !*pattern)
+		if (!mask)
-			return &engine_info[i];
+			return *pattern ? NULL : &engine_info[i];
 
 		if (!strcmp(pattern, "*")) {
 			*mask = engine_info[i].mask;

drivers/gpu/drm/xe/xe_device.c

@@ -685,16 +685,16 @@ static int wait_for_lmem_ready(struct xe_device *xe)
 }
 ALLOW_ERROR_INJECTION(wait_for_lmem_ready, ERRNO); /* See xe_pci_probe() */
 
-static void sriov_update_device_info(struct xe_device *xe)
+static void vf_update_device_info(struct xe_device *xe)
 {
+	xe_assert(xe, IS_SRIOV_VF(xe));
 	/* disable features that are not available/applicable to VFs */
-	if (IS_SRIOV_VF(xe)) {
 	xe->info.probe_display = 0;
 	xe->info.has_heci_cscfi = 0;
 	xe->info.has_heci_gscfi = 0;
+	xe->info.has_late_bind = 0;
 	xe->info.skip_guc_pc = 1;
 	xe->info.skip_pcode = 1;
-	}
 }
 
 static int xe_device_vram_alloc(struct xe_device *xe)
@@ -735,7 +735,8 @@ int xe_device_probe_early(struct xe_device *xe)
 
 	xe_sriov_probe_early(xe);
 
-	sriov_update_device_info(xe);
+	if (IS_SRIOV_VF(xe))
+		vf_update_device_info(xe);
 
 	err = xe_pcode_probe_early(xe);
 	if (err || xe_survivability_mode_is_requested(xe)) {

drivers/gpu/drm/xe/xe_hw_engine_group.c

@@ -213,17 +213,13 @@ static int xe_hw_engine_group_suspend_faulting_lr_jobs(struct xe_hw_engine_group
 
 		err = q->ops->suspend_wait(q);
 		if (err)
-			goto err_suspend;
+			return err;
 	}
 
 	if (need_resume)
 		xe_hw_engine_group_resume_faulting_lr_jobs(group);
 
 	return 0;
-
-err_suspend:
-	up_write(&group->mode_sem);
-	return err;
 }
 
 /**

drivers/gpu/drm/xe/xe_late_bind_fw.c

@@ -60,7 +60,7 @@ static int parse_cpd_header(struct xe_late_bind_fw *lb_fw,
 	const struct gsc_manifest_header *manifest;
 	const struct gsc_cpd_entry *entry;
 	size_t min_size = sizeof(*header);
-	u32 offset;
+	u32 offset = 0;
 	int i;
 
 	/* manifest_entry is mandatory */
@@ -116,7 +116,7 @@ static int parse_lb_layout(struct xe_late_bind_fw *lb_fw,
 	const struct csc_fpt_header *header = data;
 	const struct csc_fpt_entry *entry;
 	size_t min_size = sizeof(*header);
-	u32 offset;
+	u32 offset = 0;
 	int i;
 
 	/* fpt_entry is mandatory */
@@ -184,17 +184,13 @@ static const char *xe_late_bind_parse_status(uint32_t status)
 	}
 }
 
-static int xe_late_bind_fw_num_fans(struct xe_late_bind *late_bind)
+static int xe_late_bind_fw_num_fans(struct xe_late_bind *late_bind, u32 *num_fans)
 {
 	struct xe_device *xe = late_bind_to_xe(late_bind);
 	struct xe_tile *root_tile = xe_device_get_root_tile(xe);
-	u32 uval;
 
-	if (!xe_pcode_read(root_tile,
+	return xe_pcode_read(root_tile,
-			   PCODE_MBOX(FAN_SPEED_CONTROL, FSC_READ_NUM_FANS, 0), &uval, NULL))
+			     PCODE_MBOX(FAN_SPEED_CONTROL, FSC_READ_NUM_FANS, 0), num_fans, NULL);
-		return uval;
-	else
-		return 0;
 }
 
 void xe_late_bind_wait_for_worker_completion(struct xe_late_bind *late_bind)
@@ -314,7 +310,11 @@ static int __xe_late_bind_fw_init(struct xe_late_bind *late_bind, u32 fw_id)
 	lb_fw->flags &= ~INTEL_LB_FLAG_IS_PERSISTENT;
 
 	if (lb_fw->type == INTEL_LB_TYPE_FAN_CONTROL) {
-		num_fans = xe_late_bind_fw_num_fans(late_bind);
+		ret = xe_late_bind_fw_num_fans(late_bind, &num_fans);
+		if (ret) {
+			drm_dbg(&xe->drm, "Failed to read number of fans: %d\n", ret);
+			return 0; /* Not a fatal error, continue without fan control */
+		}
 		drm_dbg(&xe->drm, "Number of Fans: %d\n", num_fans);
 		if (!num_fans)
 			return 0;

drivers/gpu/drm/xe/xe_pm.c

@@ -201,7 +201,7 @@ int xe_pm_resume(struct xe_device *xe)
 	if (err)
 		goto err;
 
-	xe_i2c_pm_resume(xe, xe->d3cold.allowed);
+	xe_i2c_pm_resume(xe, true);
 
 	xe_irq_resume(xe);
 

drivers/gpu/drm/xe/xe_query.c

@@ -276,7 +276,6 @@ static int query_mem_regions(struct xe_device *xe,
 	mem_regions->mem_regions[0].instance = 0;
 	mem_regions->mem_regions[0].min_page_size = PAGE_SIZE;
 	mem_regions->mem_regions[0].total_size = man->size << PAGE_SHIFT;
-	if (perfmon_capable())
 	mem_regions->mem_regions[0].used = ttm_resource_manager_usage(man);
 	mem_regions->num_mem_regions = 1;
 
@@ -293,13 +292,11 @@ static int query_mem_regions(struct xe_device *xe,
 			mem_regions->mem_regions[mem_regions->num_mem_regions].total_size =
 				man->size;
 
-			if (perfmon_capable()) {
 			xe_ttm_vram_get_used(man,
 					     &mem_regions->mem_regions
 					     [mem_regions->num_mem_regions].used,
 					     &mem_regions->mem_regions
 					     [mem_regions->num_mem_regions].cpu_visible_used);
-			}
 
 			mem_regions->mem_regions[mem_regions->num_mem_regions].cpu_visible_size =
 				xe_ttm_vram_get_cpu_visible_size(man);

drivers/gpu/drm/xe/xe_svm.c

@@ -67,11 +67,6 @@ void xe_svm_range_debug(struct xe_svm_range *range, const char *operation)
 	range_debug(range, operation);
 }
 
-static void *xe_svm_devm_owner(struct xe_device *xe)
-{
-	return xe;
-}
-
 static struct drm_gpusvm_range *
 xe_svm_range_alloc(struct drm_gpusvm *gpusvm)
 {
@@ -744,15 +739,14 @@ int xe_svm_init(struct xe_vm *vm)
 			  xe_svm_garbage_collector_work_func);
 
 		err = drm_gpusvm_init(&vm->svm.gpusvm, "Xe SVM", &vm->xe->drm,
-				      current->mm, xe_svm_devm_owner(vm->xe), 0,
+				      current->mm, 0, vm->size,
-				      vm->size,
 				      xe_modparam.svm_notifier_size * SZ_1M,
 				      &gpusvm_ops, fault_chunk_sizes,
 				      ARRAY_SIZE(fault_chunk_sizes));
 		drm_gpusvm_driver_set_lock(&vm->svm.gpusvm, &vm->lock);
 	} else {
 		err = drm_gpusvm_init(&vm->svm.gpusvm, "Xe SVM (simple)",
-				      &vm->xe->drm, NULL, NULL, 0, 0, 0, NULL,
+				      &vm->xe->drm, NULL, 0, 0, 0, NULL,
 				      NULL, 0);
 	}
 
@@ -1017,6 +1011,7 @@ static int __xe_svm_handle_pagefault(struct xe_vm *vm, struct xe_vma *vma,
 		.devmem_only = need_vram && devmem_possible,
 		.timeslice_ms = need_vram && devmem_possible ?
 			vm->xe->atomic_svm_timeslice_ms : 0,
+		.device_private_page_owner = xe_svm_devm_owner(vm->xe),
 	};
 	struct xe_validation_ctx vctx;
 	struct drm_exec exec;

drivers/gpu/drm/xe/xe_svm.h

@@ -6,6 +6,20 @@
 #ifndef _XE_SVM_H_
 #define _XE_SVM_H_
 
+struct xe_device;
+
+/**
+ * xe_svm_devm_owner() - Return the owner of device private memory
+ * @xe: The xe device.
+ *
+ * Return: The owner of this device's device private memory to use in
+ * hmm_range_fault()-
+ */
+static inline void *xe_svm_devm_owner(struct xe_device *xe)
+{
+	return xe;
+}
+
 #if IS_ENABLED(CONFIG_DRM_XE_GPUSVM)
 
 #include <drm/drm_pagemap.h>

drivers/gpu/drm/xe/xe_userptr.c

@@ -54,6 +54,7 @@ int xe_vma_userptr_pin_pages(struct xe_userptr_vma *uvma)
 	struct xe_device *xe = vm->xe;
 	struct drm_gpusvm_ctx ctx = {
 		.read_only = xe_vma_read_only(vma),
+		.device_private_page_owner = NULL,
 	};
 
 	lockdep_assert_held(&vm->lock);

drivers/gpu/drm/xe/xe_vm.c

@@ -2881,6 +2881,7 @@ static int prefetch_ranges(struct xe_vm *vm, struct xe_vma_op *op)
 	ctx.read_only = xe_vma_read_only(vma);
 	ctx.devmem_possible = devmem_possible;
 	ctx.check_pages_threshold = devmem_possible ? SZ_64K : 0;
+	ctx.device_private_page_owner = xe_svm_devm_owner(vm->xe);
 
 	/* TODO: Threading the migration */
 	xa_for_each(&op->prefetch_range.range, i, svm_range) {

include/drm/drm_gpusvm.h

@@ -179,7 +179,6 @@ struct drm_gpusvm_range {
  * @name: Name of the GPU SVM
  * @drm: Pointer to the DRM device structure
  * @mm: Pointer to the mm_struct for the address space
- * @device_private_page_owner: Device private pages owner
  * @mm_start: Start address of GPU SVM
  * @mm_range: Range of the GPU SVM
  * @notifier_size: Size of individual notifiers
@@ -204,7 +203,6 @@ struct drm_gpusvm {
 	const char *name;
 	struct drm_device *drm;
 	struct mm_struct *mm;
-	void *device_private_page_owner;
 	unsigned long mm_start;
 	unsigned long mm_range;
 	unsigned long notifier_size;
@@ -226,6 +224,8 @@ struct drm_gpusvm {
 /**
  * struct drm_gpusvm_ctx - DRM GPU SVM context
  *
+ * @device_private_page_owner: The device-private page owner to use for
+ * this operation
  * @check_pages_threshold: Check CPU pages for present if chunk is less than or
  *                         equal to threshold. If not present, reduce chunk
  *                         size.
@@ -239,6 +239,7 @@ struct drm_gpusvm {
  * Context that is DRM GPUSVM is operating in (i.e. user arguments).
  */
 struct drm_gpusvm_ctx {
+	void *device_private_page_owner;
 	unsigned long check_pages_threshold;
 	unsigned long timeslice_ms;
 	unsigned int in_notifier :1;
@@ -249,7 +250,7 @@ struct drm_gpusvm_ctx {
 
 int drm_gpusvm_init(struct drm_gpusvm *gpusvm,
 		    const char *name, struct drm_device *drm,
-		    struct mm_struct *mm, void *device_private_page_owner,
+		    struct mm_struct *mm,
 		    unsigned long mm_start, unsigned long mm_range,
 		    unsigned long notifier_size,
 		    const struct drm_gpusvm_ops *ops,

include/uapi/linux/kfd_ioctl.h

@@ -67,8 +67,8 @@ struct kfd_ioctl_get_version_args {
 
 struct kfd_ioctl_create_queue_args {
 	__u64 ring_base_address;	/* to KFD */
-	__u64 write_pointer_address;	/* from KFD */
+	__u64 write_pointer_address;	/* to KFD */
-	__u64 read_pointer_address;	/* from KFD */
+	__u64 read_pointer_address;	/* to KFD */
 	__u64 doorbell_offset;	/* from KFD */
 
 	__u32 ring_size;		/* to KFD */