[PATCH] drm/amdgpu: improve HMM error -ENOMEM and -EBUSY handling

[Kuehling, Felix]

On 2019-06-14 9:52 p.m., Yang, Philip wrote:
> Under memory pressure, hmm_range_fault may return error code -ENOMEM
> or -EBUSY, change pr_info to pr_debug to remove unnecessary kernel log
> message because we will retry restore again.
>
> Call get_user_pages_done if TTM get user pages failed will have
> WARN_ONCE kernel calling stack dump log.
>
> Change-Id: I086f92944630f9d1a70365c00417cb9440662464
> Signed-off-by: Philip Yang <Philip.Yang at amd.com>

Reviewed-by: Felix Kuehling <Felix.Kuehling at amd.com>


> ---
>   .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c  | 38 +++----------------
>   1 file changed, 6 insertions(+), 32 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> index 74e86952553f..10abae398e51 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
> @@ -1731,35 +1731,17 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
>   		ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
>   						   bo->tbo.ttm->pages);
>   		if (ret) {
> -			bo->tbo.ttm->pages[0] = NULL;
> -			pr_info("%s: Failed to get user pages: %d\n",
> +			pr_debug("%s: Failed to get user pages: %d\n",
>   				__func__, ret);
> -			/* Pretend it succeeded. It will fail later
> -			 * with a VM fault if the GPU tries to access
> -			 * it. Better than hanging indefinitely with
> -			 * stalled user mode queues.
> -			 */
> -		}
> -	}
> -
> -	return 0;
> -}
>   
> -/* Remove invalid userptr BOs from hmm track list
> - *
> - * Stop HMM track the userptr update
> - */
> -static void untrack_invalid_user_pages(struct amdkfd_process_info *process_info)
> -{
> -	struct kgd_mem *mem, *tmp_mem;
> -	struct amdgpu_bo *bo;
> +			/* Return error -EBUSY or -ENOMEM, retry restore */
> +			return ret;
> +		}
>   
> -	list_for_each_entry_safe(mem, tmp_mem,
> -				 &process_info->userptr_inval_list,
> -				 validate_list.head) {
> -		bo = mem->bo;
>   		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
>   	}
> +
> +	return 0;
>   }
>   
>   /* Validate invalid userptr BOs
> @@ -1841,13 +1823,6 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
>   		list_move_tail(&mem->validate_list.head,
>   			       &process_info->userptr_valid_list);
>   
> -		/* Stop HMM track the userptr update. We dont check the return
> -		 * value for concurrent CPU page table update because we will
> -		 * reschedule the restore worker if process_info->evicted_bos
> -		 * is updated.
> -		 */
> -		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
> -
>   		/* Update mapping. If the BO was not validated
>   		 * (because we couldn't get user pages), this will
>   		 * clear the page table entries, which will result in
> @@ -1946,7 +1921,6 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
>   	}
>   
>   unlock_out:
> -	untrack_invalid_user_pages(process_info);
>   	mutex_unlock(&process_info->lock);
>   	mmput(mm);
>   	put_task_struct(usertask);