Merge tag 'amd-drm-next-6.9-2024-03-08-1' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.9-2024-03-08-1:

amdgpu:
- DCN 3.5.1 support
- Fixes for IOMMUv2 removal
- UAF fix
- Misc small fixes and cleanups
- SR-IOV fixes
- MCBP cleanup
- devcoredump update
- NBIF 6.3.1 support
- VPE 6.1.1 support

amdkfd:
- Misc fixes and cleanups
- GFX10.1 trap fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240308170741.3691166-1-alexander.deucher@amd.com

1 files changed, 119 insertions, 71 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index ed4a8c5d26d7..4299ce386322 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -513,8 +513,14 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		bo = bo_base->bo;
 
 		if (dma_resv_locking_ctx(bo->tbo.base.resv) != ticket) {
-			pr_warn_ratelimited("Evicted user BO is not reserved in pid %d\n",
-					    vm->task_info.pid);
+			struct amdgpu_task_info *ti = amdgpu_vm_get_task_info_vm(vm);
+
+			pr_warn_ratelimited("Evicted user BO is not reserved\n");
+			if (ti) {
+				pr_warn_ratelimited("pid %d\n", ti->pid);
+				amdgpu_vm_put_task_info(ti);
+			}
+
 			return -EINVAL;
 		}
 
@@ -652,7 +658,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	bool vm_flush_needed = job->vm_needs_flush;
 	struct dma_fence *fence = NULL;
 	bool pasid_mapping_needed = false;
-	unsigned patch_offset = 0;
+	unsigned int patch;
 	int r;
 
 	if (amdgpu_vmid_had_gpu_reset(adev, id)) {
@@ -679,7 +685,8 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 
 	amdgpu_ring_ib_begin(ring);
 	if (ring->funcs->init_cond_exec)
-		patch_offset = amdgpu_ring_init_cond_exec(ring);
+		patch = amdgpu_ring_init_cond_exec(ring,
+						   ring->cond_exe_gpu_addr);
 
 	if (need_pipe_sync)
 		amdgpu_ring_emit_pipeline_sync(ring);
@@ -727,8 +734,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	}
 	dma_fence_put(fence);
 
-	if (ring->funcs->patch_cond_exec)
-		amdgpu_ring_patch_cond_exec(ring, patch_offset);
+	amdgpu_ring_patch_cond_exec(ring, patch);
 
 	/* the double SWITCH_BUFFER here *cannot* be skipped by COND_EXEC */
 	if (ring->funcs->emit_switch_buffer) {
@@ -1385,10 +1391,6 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
 
-		if (vm->pte_support_ats &&
-		    mapping->start < AMDGPU_GMC_HOLE_START)
-			init_pte_value = AMDGPU_PTE_DEFAULT_ATC;
-
 		r = amdgpu_vm_update_range(adev, vm, false, false, true, false,
 					   resv, mapping->start, mapping->last,
 					   init_pte_value, 0, 0, NULL, NULL,
@@ -2225,6 +2227,108 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
 	return dma_fence_wait_timeout(vm->last_unlocked, true, timeout);
 }
 
+static void amdgpu_vm_destroy_task_info(struct kref *kref)
+{
+	struct amdgpu_task_info *ti = container_of(kref, struct amdgpu_task_info, refcount);
+
+	kfree(ti);
+}
+
+static inline struct amdgpu_vm *
+amdgpu_vm_get_vm_from_pasid(struct amdgpu_device *adev, u32 pasid)
+{
+	struct amdgpu_vm *vm;
+	unsigned long flags;
+
+	xa_lock_irqsave(&adev->vm_manager.pasids, flags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
+
+	return vm;
+}
+
+/**
+ * amdgpu_vm_put_task_info - reference down the vm task_info ptr
+ *
+ * @task_info: task_info struct under discussion.
+ *
+ * frees the vm task_info ptr at the last put
+ */
+void amdgpu_vm_put_task_info(struct amdgpu_task_info *task_info)
+{
+	kref_put(&task_info->refcount, amdgpu_vm_destroy_task_info);
+}
+
+/**
+ * amdgpu_vm_get_task_info_vm - Extracts task info for a vm.
+ *
+ * @vm: VM to get info from
+ *
+ * Returns the reference counted task_info structure, which must be
+ * referenced down with amdgpu_vm_put_task_info.
+ */
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_vm(struct amdgpu_vm *vm)
+{
+	struct amdgpu_task_info *ti = NULL;
+
+	if (vm) {
+		ti = vm->task_info;
+		kref_get(&vm->task_info->refcount);
+	}
+
+	return ti;
+}
+
+/**
+ * amdgpu_vm_get_task_info_pasid - Extracts task info for a PASID.
+ *
+ * @adev: drm device pointer
+ * @pasid: PASID identifier for VM
+ *
+ * Returns the reference counted task_info structure, which must be
+ * referenced down with amdgpu_vm_put_task_info.
+ */
+struct amdgpu_task_info *
+amdgpu_vm_get_task_info_pasid(struct amdgpu_device *adev, u32 pasid)
+{
+	return amdgpu_vm_get_task_info_vm(
+			amdgpu_vm_get_vm_from_pasid(adev, pasid));
+}
+
+static int amdgpu_vm_create_task_info(struct amdgpu_vm *vm)
+{
+	vm->task_info = kzalloc(sizeof(struct amdgpu_task_info), GFP_KERNEL);
+	if (!vm->task_info)
+		return -ENOMEM;
+
+	kref_init(&vm->task_info->refcount);
+	return 0;
+}
+
+/**
+ * amdgpu_vm_set_task_info - Sets VMs task info.
+ *
+ * @vm: vm for which to set the info
+ */
+void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
+{
+	if (!vm->task_info)
+		return;
+
+	if (vm->task_info->pid == current->pid)
+		return;
+
+	vm->task_info->pid = current->pid;
+	get_task_comm(vm->task_info->task_name, current);
+
+	if (current->group_leader->mm != current->mm)
+		return;
+
+	vm->task_info->tgid = current->group_leader->pid;
+	get_task_comm(vm->task_info->process_name, current->group_leader);
+}
+
 /**
  * amdgpu_vm_init - initialize a vm instance
  *
@@ -2264,7 +2368,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		return r;
 
-	vm->pte_support_ats = false;
 	vm->is_compute_context = false;
 
 	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
@@ -2311,6 +2414,10 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		goto error_free_root;
 
+	r = amdgpu_vm_create_task_info(vm);
+	if (r)
+		DRM_DEBUG("Failed to create task info for VM\n");
+
 	amdgpu_bo_unreserve(vm->root.bo);
 	amdgpu_bo_unref(&root_bo);
 
@@ -2350,30 +2457,12 @@ error_free_delayed:
  */
 int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
-	bool pte_support_ats = (adev->asic_type == CHIP_RAVEN);
 	int r;
 
 	r = amdgpu_bo_reserve(vm->root.bo, true);
 	if (r)
 		return r;
 
-	/* Check if PD needs to be reinitialized and do it before
-	 * changing any other state, in case it fails.
-	 */
-	if (pte_support_ats != vm->pte_support_ats) {
-		/* Sanity checks */
-		if (!amdgpu_vm_pt_is_root_clean(adev, vm)) {
-			r = -EINVAL;
-			goto unreserve_bo;
-		}
-
-		vm->pte_support_ats = pte_support_ats;
-		r = amdgpu_vm_pt_clear(adev, vm, to_amdgpu_bo_vm(vm->root.bo),
-				       false);
-		if (r)
-			goto unreserve_bo;
-	}
-
 	/* Update VM state */
 	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
 				    AMDGPU_VM_USE_CPU_FOR_COMPUTE);
@@ -2450,6 +2539,7 @@ 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_put_task_info(vm->task_info);
 	amdgpu_vm_set_pasid(adev, vm, 0);
 	dma_fence_wait(vm->last_unlocked, false);
 	dma_fence_put(vm->last_unlocked);
@@ -2607,48 +2697,6 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 }
 
 /**
- * amdgpu_vm_get_task_info - Extracts task info for a PASID.
- *
- * @adev: drm device pointer
- * @pasid: PASID identifier for VM
- * @task_info: task_info to fill.
- */
-void amdgpu_vm_get_task_info(struct amdgpu_device *adev, u32 pasid,
-			 struct amdgpu_task_info *task_info)
-{
-	struct amdgpu_vm *vm;
-	unsigned long flags;
-
-	xa_lock_irqsave(&adev->vm_manager.pasids, flags);
-
-	vm = xa_load(&adev->vm_manager.pasids, pasid);
-	if (vm)
-		*task_info = vm->task_info;
-
-	xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
-}
-
-/**
- * amdgpu_vm_set_task_info - Sets VMs task info.
- *
- * @vm: vm for which to set the info
- */
-void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
-{
-	if (vm->task_info.pid)
-		return;
-
-	vm->task_info.pid = current->pid;
-	get_task_comm(vm->task_info.task_name, current);
-
-	if (current->group_leader->mm != current->mm)
-		return;
-
-	vm->task_info.tgid = current->group_leader->pid;
-	get_task_comm(vm->task_info.process_name, current->group_leader);
-}
-
-/**
  * amdgpu_vm_handle_fault - graceful handling of VM faults.
  * @adev: amdgpu device pointer
  * @pasid: PASID of the VM