Using context and sensor in user class hangs on destruction #7553
Labels
T260 series
Intel® T265 library
Comments
|
Just as an update: there is a potential bug in my IntelRealSenseControl.cpp source, thus I update it here: In createCallBack(), gyro_data is now defined within the functor. This DOES NOT change the issue described above. |
krazycoder2k
added a commit
to krazycoder2k/librealsense
that referenced
this issue
Mar 12, 2021
This PR is effectively the suggested fix in the GH issue below. Thank you @ankyur. IntelRealSense#7276 I've validated the fix works using the following code: // Reproduces T265 Hand on Exit. int main(int, char**) { constexpr std::chrono::seconds timeout{ 1 }; while (true) { // Start rs2::config config; rs2::pipeline pipeline; std::cout << "Entering pipeline.start()" << std::endl; pipeline.start(); std::cout << "Exiting pipeline.start()" << std::endl; std::cout << "Sleeping for 1 second..." << std::endl; std::this_thread::sleep_for(timeout); std::cout << "Entering pipeline.stop()" << std::endl; pipeline.stop(); std::cout << "Exiting pipeline.stop()" << std::endl; } return 0; } Suspect this fix potentially addresses the following open T265 issues as well: IntelRealSense#7553 IntelRealSense#5807 IntelRealSense#6272 IntelRealSense#7555 IntelRealSense#7750
|
Fixed in latest via PR #8561 Thanks |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Issue Description
I have built a minimal project consisting of a main program file and two classes. It simply streams gyroscope data and
can detect if the device is disconnected and reconnected and can restart streaming. See attached text files for the program:
IntelRealSenseControl_cpp.txt
IntelRealSenseControl_h.txt
IntelRealSenseSensor_cpp.txt
IntelRealSenseSensor_h.txt
main_cpp.txt
SDK warnings and errors are enabled and print to output.
The program stops if you type 'c' and press enter. It happens quite often (typically at least once in 20 trials) that the program hangs during destruction in the mp_sensorManager.reset() step of the IntelRealSenseControl::controlLoop().
The pointer mp_sensorManager holds a class with a context and a sensor member.
Thus, the reset destroys these SDK objects. First sensor, then context.
Further Details
It seems to happen in the destruction of the rs2_delete_sensor of rs.cpp.
There, VALIDATE_NOT_NULL(device) evaluates sometimes to false and a throw is issued
which leads to skipping the call of the context destructor.
However, the context destructor would stop the thread, where the device watcher is running (_device_watcher->stop()). This thread was started when the program was handed over a callback function to the API member function of context: set_devices_changed_callback.
Thus, this thread is never stopped and the controlLoop thread cannot be joined back and the program hangs.
Does this program use the API in a wrong way or is it indeed an API bug?
Thanks for help.
Asbjoern
The text was updated successfully, but these errors were encountered: