Deepstream 5.0, after upgrade, getting "nvbuf_utils: dmabuf_fd 1308 mapped entry NOT found"

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)

Something new in 5.0? I now see the following message nvbuf_utils: dmabuf_fd 1308 mapped entry NOT found… this happens when the Pipeline is explicitly set back to a “READY” state, after receiving a EOS message.

Is there anything I should be doing differently?


Is there possible share one sample which can run in our environments to repro your issue?

Hi @amycao,

I’ll try to put something together. And sorry, I should have provided a little more detail I’ve looked through many of the examples but I haven’t found a case where a Pipeline is explicitly set to a Ready State from a Playing State. For example,
using a non-live stream.

  1. link up the components
  2. set the pipeline to GST_STATE_PAUSED
  3. set the pipeline to GST_STATE_PLAYING
  4. on EOS stream message - bus watch handler
  5. set the pipeline to GST_STATE_READY
  6. update/add/remove pipeline components
  7. return to step 1.

I haven’t been able to find any official gstreamer examples for the above, but there are a number of gstreamer questions/answers on Stack Overflow of suggesting the above, unrelated to Deepstream.

It is on step 5. above that I see the error message, and I wanted to know if you see anything in the above scenario that would concern you? Do you know of any examples that do at least something similar?


I see you are using JetPack 4.0 - it will not work - you need move to Jetpack 4.4

@amycao Sorry, that was a typo… I’m using 4.4 … just be to be clear, There is no problem playing the pipeline. It is just an error message I see as I stop the pipeline by setting the state back to ready.

The error doesn’t seem to affect the behavior as this happens during Pipeline stop,… the question is more, should I be stopping the Pipeline using a different method? Should I worry about the error.?

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.

Can you share your source code?