BUG REPORT: the deepstream5.0 reference application process sub_bins index incorrectly in demux setting

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) GTX2080 TI
• DeepStream Version 5.0
• JetPack Version (valid for Jetson only)
• TensorRT Version same as offcial devel docker
• NVIDIA GPU Driver Version (valid for GPU only) same as official devel docker

There’s a bug in DeepsStream5.0 reference app. If the demuxer is enabled (tiled display is disabled), and we have sink’s conresponding source-id not equal to 0, then the “NVGSTDS_ELEM_ADD_PROBE” macro will get invalid sink element, resulting in immediate core dump.

File: “deepstreaim_app.c”,
Function:

gboolean create_pipeline (AppCtx * appCtx, bbox_generated_callback bbox_generated_post_analytics_cb, bbox_generated_callback all_bbox_generated_cb, perf_callback perf_cb, overlay_graphics_callback overlay_graphics_cb)

Location:

NVGSTDS_ELEM_ADD_PROBE(latency_probe_id,
   pipeline->instance_bins[i].sink_bin.sub_bins[0].sink, "sink",
   latency_measurement_buf_prob, GST_PAD_PROBE_TYPE_BUFFER,
   appCtx);

Should be changed to:

NVGSTDS_ELEM_ADD_PROBE(latency_probe_id,
    pipeline->instance_bins[i].sink_bin.sub_bins[i].sink, "sink",
    latency_measurement_buf_prob, GST_PAD_PROBE_TYPE_BUFFER,
    appCtx);

This fix is compatible with logic in function “create_sink_bin” from file “deepstream_sink_bin.c”.

@horacehxw
Can you try our patch for this problem? seg_fault_deepstream_app.zip (504 Bytes)

2 Likes

We had encountered this issue and the problem seems to have been fixed with this patch.

1 Like