Segmentation fault when re-creating the Gstreamer pipeline with nvarguscamerasrc

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

Segmentation fault when recreating the pipeline with gstreamer on Jetson Nano (using nvarguscamerasrc plugin), below is the debug log. nvcamerasrc used to work previously on old SDKs.

0x0000007fb79ce6f0 in __GI___pthread_mutex_lock (mutex=0x58) at pthread_mutex_lock.c:65
#1 0x0000007fb72ba098 in () at /usr/lib/aarch64-linux-gnu/gstreamer-1.0/libgstnvarguscamerasrc.so
#2 0x0000007fb72b8b60 in () at /usr/lib/aarch64-linux-gnu/gstreamer-1.0/libgstnvarguscamerasrc.so
#3 0x0000007fb7d5498c in () at /usr/lib/aarch64-linux-gnu/libglib-2.0.so.0
#4 0x0000007fb342e598 in ()

2

I a, creating the pipeline, destroying and again creating the pipeline. I am doing all the steps in a single program.

3

hello ram469,

could you please share more details, may I know which JetPack release you’re working with.
please also share the commands that you’re able to see the segmentation fault failure for review.
thanks

1 Like
4

Hi Jerry,
I am using OpenCV, please find the code below. I am trying to create a pipeline and destroy it and again create a new one. In this process, I am experiencing the crash. Thanks for the kind reply.

#include <opencv2/opencv.hpp>
// #include <opencv2/videoio.hpp>
// #include <opencv2/highgui.hpp>

std::string gstreamer_pipeline (int capture_width, int capture_height, int display_width, int display_height, int framerate, int flip_method) {
    return "nvarguscamerasrc ! video/x-raw(memory:NVMM), width=(int)" + std::to_string(capture_width) + ", height=(int)" +
           std::to_string(capture_height) + ", format=(string)NV12, framerate=(fraction)" + std::to_string(framerate) +
           "/1 ! nvvidconv flip-method=" + std::to_string(flip_method) + " ! video/x-raw, width=(int)" + std::to_string(display_width) + ", height=(int)" +
           std::to_string(display_height) + ", format=(string)BGRx ! videoconvert ! video/x-raw, format=(string)BGR ! appsink";
}

int main()
{
    int capture_width = 1280 ;
    int capture_height = 720 ;
    int display_width = 1280 ;
    int display_height = 720 ;
    int framerate = 60 ;
    int flip_method = 0 ;

    std::string pipeline = gstreamer_pipeline(capture_width,
	capture_height,
	display_width,
	display_height,
	framerate,
	flip_method);
    std::cout << "Using pipeline: \n\t" << pipeline << "\n";
   
   /* Create a VideoCapture element */
    cv::VideoCapture  cap1(pipeline, cv::CAP_GSTREAMER);; 
    if(!cap1.isOpened()) {
	std::cout<<"Failed to open camera."<<std::endl;
	return (-1);
    }
    cap1.release(); // released the first capture element
    
    /* Create a another VideoCapture element */
    cv::VideoCapture  cap(pipeline, cv::CAP_GSTREAMER);; 
    if(!cap.isOpened()) {
	std::cout<<"Failed to open camera."<<std::endl;
	return (-1);
    }
    if(!cap.isOpened()) {
	std::cout<<"Failed to open camera."<<std::endl;
	return (-1);
    }
    cv::namedWindow("CSI Camera", cv::WINDOW_AUTOSIZE);
    cv::Mat img;

    std::cout << "Hit ESC to exit" << "\n" ;
    while(true)
    {
    	if (!cap.read(img)) {
		std::cout<<"Capture read error"<<std::endl;
		break;
	}
	
	cv::imshow("CSI Camera",img);
	int keycode = cv::waitKey(30) & 0xff ; 
        if (keycode == 27) break ;
    }

    cap.release();
    cv::destroyAllWindows() ;
    return 0;
}
5

I am using Jetpack 4.2.

6

hello ram469,

there’s known issue that nvargus-daemon hang when calling pipeline stop with JetPack-4.2 / l4t-r32.1,
you might upgrade your working process to the latest JetPack release, i.e. JetPack-4.2.2 / l4t-32.2.1
BTW,
please also check similar discussion thread, Topic 1062196 for reference.
thanks

7

Aprart from upgrading the SDK, I did the second method i.e sent an EOS.

According to the thread you posted, I sent an EOS signal on the bus but no luck.
I used below APIs to do this but no luck still seeing the crash.

gst_bus_post (data->bus,gst_message_new_eos (NULL));

Should there be another way of doing this?

Thanks for the reply.

8

hello ram469,

it’s a l4t-r32.1 known failure to terminate the pipeline.
since we had significant modifications to fix the issue, suggest upgrade to l4t-r32.2 if it’s possible.
thanks

9

Thanks Jerry.

10

I am just curious to know whether the old versions are also facing this issue or only this version.