1 files changed, 28 insertions, 50 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 32e5db509560..00350eccd571 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -247,14 +247,15 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 	q_properties->priority = args->queue_priority;
 	q_properties->queue_address = args->ring_base_address;
 	q_properties->queue_size = args->ring_size;
-	q_properties->read_ptr = (uint32_t *) args->read_pointer_address;
-	q_properties->write_ptr = (uint32_t *) args->write_pointer_address;
+	q_properties->read_ptr = (void __user *)args->read_pointer_address;
+	q_properties->write_ptr = (void __user *)args->write_pointer_address;
 	q_properties->eop_ring_buffer_address = args->eop_buffer_address;
 	q_properties->eop_ring_buffer_size = args->eop_buffer_size;
 	q_properties->ctx_save_restore_area_address =
 			args->ctx_save_restore_address;
 	q_properties->ctx_save_restore_area_size = args->ctx_save_restore_size;
 	q_properties->ctl_stack_size = args->ctl_stack_size;
+	q_properties->sdma_engine_id = args->sdma_engine_id;
 	if (args->queue_type == KFD_IOC_QUEUE_TYPE_COMPUTE ||
 		args->queue_type == KFD_IOC_QUEUE_TYPE_COMPUTE_AQL)
 		q_properties->type = KFD_QUEUE_TYPE_COMPUTE;
@@ -262,6 +263,8 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 		q_properties->type = KFD_QUEUE_TYPE_SDMA;
 	else if (args->queue_type == KFD_IOC_QUEUE_TYPE_SDMA_XGMI)
 		q_properties->type = KFD_QUEUE_TYPE_SDMA_XGMI;
+	else if (args->queue_type == KFD_IOC_QUEUE_TYPE_SDMA_BY_ENG_ID)
+		q_properties->type = KFD_QUEUE_TYPE_SDMA_BY_ENG_ID;
 	else
 		return -ENOTSUPP;
 
@@ -306,7 +309,6 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 	struct kfd_process_device *pdd;
 	struct queue_properties q_properties;
 	uint32_t doorbell_offset_in_process = 0;
-	struct amdgpu_bo *wptr_bo = NULL;
 
 	memset(&q_properties, 0, sizeof(struct queue_properties));
 
@@ -334,6 +336,18 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 		goto err_bind_process;
 	}
 
+	if (q_properties.type == KFD_QUEUE_TYPE_SDMA_BY_ENG_ID) {
+		int max_sdma_eng_id = kfd_get_num_sdma_engines(dev) +
+				      kfd_get_num_xgmi_sdma_engines(dev) - 1;
+
+		if (q_properties.sdma_engine_id > max_sdma_eng_id) {
+			err = -EINVAL;
+			pr_err("sdma_engine_id %i exceeds maximum id of %i\n",
+			       q_properties.sdma_engine_id, max_sdma_eng_id);
+			goto err_sdma_engine_id;
+		}
+	}
+
 	if (!pdd->qpd.proc_doorbells) {
 		err = kfd_alloc_process_doorbells(dev->kfd, pdd);
 		if (err) {
@@ -342,53 +356,17 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 		}
 	}
 
-	/* Starting with GFX11, wptr BOs must be mapped to GART for MES to determine work
-	 * on unmapped queues for usermode queue oversubscription (no aggregated doorbell)
-	 */
-	if (dev->kfd->shared_resources.enable_mes &&
-			((dev->adev->mes.sched_version & AMDGPU_MES_API_VERSION_MASK)
-			>> AMDGPU_MES_API_VERSION_SHIFT) >= 2) {
-		struct amdgpu_bo_va_mapping *wptr_mapping;
-		struct amdgpu_vm *wptr_vm;
-
-		wptr_vm = drm_priv_to_vm(pdd->drm_priv);
-		err = amdgpu_bo_reserve(wptr_vm->root.bo, false);
-		if (err)
-			goto err_wptr_map_gart;
-
-		wptr_mapping = amdgpu_vm_bo_lookup_mapping(
-				wptr_vm, args->write_pointer_address >> PAGE_SHIFT);
-		amdgpu_bo_unreserve(wptr_vm->root.bo);
-		if (!wptr_mapping) {
-			pr_err("Failed to lookup wptr bo\n");
-			err = -EINVAL;
-			goto err_wptr_map_gart;
-		}
-
-		wptr_bo = wptr_mapping->bo_va->base.bo;
-		if (wptr_bo->tbo.base.size > PAGE_SIZE) {
-			pr_err("Requested GART mapping for wptr bo larger than one page\n");
-			err = -EINVAL;
-			goto err_wptr_map_gart;
-		}
-		if (dev->adev != amdgpu_ttm_adev(wptr_bo->tbo.bdev)) {
-			pr_err("Queue memory allocated to wrong device\n");
-			err = -EINVAL;
-			goto err_wptr_map_gart;
-		}
-
-		err = amdgpu_amdkfd_map_gtt_bo_to_gart(wptr_bo);
-		if (err) {
-			pr_err("Failed to map wptr bo to GART\n");
-			goto err_wptr_map_gart;
-		}
+	err = kfd_queue_acquire_buffers(pdd, &q_properties);
+	if (err) {
+		pr_debug("failed to acquire user queue buffers\n");
+		goto err_acquire_queue_buf;
 	}
 
 	pr_debug("Creating queue for PASID 0x%x on gpu 0x%x\n",
 			p->pasid,
 			dev->id);
 
-	err = pqm_create_queue(&p->pqm, dev, filep, &q_properties, &queue_id, wptr_bo,
+	err = pqm_create_queue(&p->pqm, dev, filep, &q_properties, &queue_id,
 			NULL, NULL, NULL, &doorbell_offset_in_process);
 	if (err != 0)
 		goto err_create_queue;
@@ -422,9 +400,10 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
 	return 0;
 
 err_create_queue:
-	if (wptr_bo)
-		amdgpu_amdkfd_free_gtt_mem(dev->adev, wptr_bo);
-err_wptr_map_gart:
+	kfd_queue_unref_bo_vas(pdd, &q_properties);
+	kfd_queue_release_buffers(pdd, &q_properties);
+err_acquire_queue_buf:
+err_sdma_engine_id:
 err_bind_process:
 err_pdd:
 	mutex_unlock(&p->mutex);
@@ -1422,8 +1401,7 @@ static int kfd_ioctl_unmap_memory_from_gpu(struct file *filep,
 		err = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
 			peer_pdd->dev->adev, (struct kgd_mem *)mem, peer_pdd->drm_priv);
 		if (err) {
-			pr_err("Failed to unmap from gpu %d/%d\n",
-			       i, args->n_devices);
+			pr_debug("Failed to unmap from gpu %d/%d\n", i, args->n_devices);
 			goto unmap_memory_from_gpu_failed;
 		}
 		args->n_success = i+1;
@@ -1963,7 +1941,7 @@ static int criu_checkpoint_bos(struct kfd_process *p,
 				bo_bucket->offset = amdgpu_bo_mmap_offset(dumper_bo);
 
 			for (i = 0; i < p->n_pdds; i++) {
-				if (amdgpu_amdkfd_bo_mapped_to_dev(p->pdds[i]->dev->adev, kgd_mem))
+				if (amdgpu_amdkfd_bo_mapped_to_dev(p->pdds[i]->drm_priv, kgd_mem))
 					bo_priv->mapped_gpuids[dev_idx++] = p->pdds[i]->user_gpu_id;
 			}