Error reported in nvosd plugin

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

Hi,My hardware platform is Jetson Xavier NX,Deepstream sdk version is the latest version 5.0.1.When my pipeline is running,nvosd plugin reports an error as shown below：
libnvosd (281):(ERROR) : nvll_osd_vic_draw_rects_without_alpha: Rectangle processing failed: ffffffff
ERROR from element nv-onscreendisplay: Unable to draw rectangles
Error details: plugins/gst-nvdsosd/gstnvdsosd.c(490): gst_nvds_osd_transform_ip (): /GstPipeline:media-pipeline/GstBin:composite_bin/GstNvDsOsd:nv-onscreendisplay
ERROR from element composite-nvmultistreamtiler: Internal data stream error.

This error will cause the pipeline to crash.And Kernel reports an warning as shown below:
kernel: [81336.561456] nvmap_alloc_handle: PID 9602: app: WARNING: All NvMap Allocations must have a tag to identify the subsystem allocating memory.Please pass the tag to the API call NvRmMemHanldeAllocAttr() or relevant.

Through my investigation，this crash is caused by other modules.But I don’t know the underlying cause of this crash，hope to help check it.

#3

Hey, what’s your pipeline and have you modified the OSD plugin or other DS plugins?

#4

Hello，The whole function of my pipeline is quite complicated. Let me give a brief introduction. The main function of my pipeline is to access the front-end H264 code stream, decode and generate nv12 image, then perform algorithm processing and OSD superposition, and then display it.In the source code of nvosd plugin，I only add a line of code in gst_nvds_osd_transform_ip function to force the number of batch input by the plug-in to be 1.The changes are shown in the picture below。

gst_nvds_osd_transform_ip
gst_nvds_osd_transform_ip1132×1364 284 KB

#5

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Can you revert your change and see if the issue persist, I don’t think you can modify the batch size of the surface.

BTW, it’s better to repro your issue with a simple app and pipeline, so I can help debug locally.