diff mbox series

[FFmpeg-devel,v2] libavfilter/dnn: fix openvino async mode

Message ID PH7PR11MB58878E868D9EE9845AC141C5DAE69@PH7PR11MB5887.namprd11.prod.outlook.com
State New
Headers show
Series [FFmpeg-devel,v2] libavfilter/dnn: fix openvino async mode | expand

Checks

Context Check Description
yinshiyou/make_loongarch64 success Make finished
yinshiyou/make_fate_loongarch64 success Make fate finished
andriy/make_x86 success Make finished
andriy/make_fate_x86 success Make fate finished

Commit Message

Saliev, Rafik F Dec. 16, 2022, 10:34 a.m. UTC 
Bugfix: The OpenVino DNN backend in the 'async' mode sets
'task->inference_done' to 'complete' prior to data copy from
OpenVino output buffer to task's output frame.
This order causes task destroy in ff_dnn_get_result_common()
prior to model output processing.

Signed-off-by: Rafik Saliev <rafik.f.saliev@intel.com>
---
 libavfilter/dnn/dnn_backend_openvino.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Guo, Yejun Dec. 16, 2022, 11:02 a.m. UTC | #1 
> -----Original Message-----
> From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of
> Saliev, Rafik F
> Sent: Friday, December 16, 2022 6:35 PM
> To: ffmpeg-devel@ffmpeg.org
> Subject: [FFmpeg-devel] [PATCH v2] libavfilter/dnn: fix openvino async mode
> 
> Bugfix: The OpenVino DNN backend in the 'async' mode sets 'task-
> >inference_done' to 'complete' prior to data copy from OpenVino output
> buffer to task's output frame.
> This order causes task destroy in ff_dnn_get_result_common() prior to model
> output processing.
> 
> Signed-off-by: Rafik Saliev <rafik.f.saliev@intel.com>
> ---
>  libavfilter/dnn/dnn_backend_openvino.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libavfilter/dnn/dnn_backend_openvino.c
> b/libavfilter/dnn/dnn_backend_openvino.c
> index b494f26f55..b67f288336 100644
> --- a/libavfilter/dnn/dnn_backend_openvino.c
> +++ b/libavfilter/dnn/dnn_backend_openvino.c
> @@ -244,7 +244,6 @@ static void infer_completion_callback(void *args)
>      av_assert0(request->lltask_count >= 1);
>      for (int i = 0; i < request->lltask_count; ++i) {
>          task = request->lltasks[i]->task;
> -        task->inference_done++;
> 
>          switch (ov_model->model->func_type) {
>          case DFT_PROCESS_FRAME:
> @@ -278,6 +277,7 @@ static void infer_completion_callback(void *args)
>              break;
>          }
> 
> +        task->inference_done++;
>          av_freep(&request->lltasks[i]);
>          output.data = (uint8_t *)output.data
>                        + output.width * output.height * output.channels *
> get_datatype_size(output.dt);
> --
LGTM, will push soon, thanks.
diff mbox series

Patch

diff --git a/libavfilter/dnn/dnn_backend_openvino.c b/libavfilter/dnn/dnn_backend_openvino.c
index b494f26f55..b67f288336 100644
--- a/libavfilter/dnn/dnn_backend_openvino.c
+++ b/libavfilter/dnn/dnn_backend_openvino.c
@@ -244,7 +244,6 @@  static void infer_completion_callback(void *args)
     av_assert0(request->lltask_count >= 1);
     for (int i = 0; i < request->lltask_count; ++i) {
         task = request->lltasks[i]->task;
-        task->inference_done++;
 
         switch (ov_model->model->func_type) {
         case DFT_PROCESS_FRAME:
@@ -278,6 +277,7 @@  static void infer_completion_callback(void *args)
             break;
         }
 
+        task->inference_done++;
         av_freep(&request->lltasks[i]);
         output.data = (uint8_t *)output.data
                       + output.width * output.height * output.channels * get_datatype_size(output.dt);