hardware_reset generates "failed to set power state" in C++ application

[jpfsaunders]

|---------------------------------|------------------------------------------- |
| Camera Model = D455
| Firmware Version = 5.13.00.50
| Operating System & Version Linux = Ubuntu 18.04
| Kernel Version (Linux Only) = ?
| Platform = nVidia Jetson Nano
| SDK Version = ?
| Language = C++ |
|

Issue Description

My C++ app operates three D455 cameras. Normally it starts and runs correctly. I would like to add a feature that monitors some info and then may decide to perform a hardware reset on a single camera (not all three), and then re-enable the steams and re-start the pipeline for that one camera (leaving the other two running as they originally were).

My attempt to do this is not working. All I did was run my code to determine if a reset is warranted (this part does work), and if it is, perform a hardware reset and then repeat the same code I used to originally (successfully) start up the camera.
When I try the hardware_reset(), the OS throws an exception "failed to set power state" and the app stops. The reset/restart code I tried using appears below:

// Below, variables g_list_of_cameras, serials, and ctx have been previously defined before this code block executes
// This next block assumes that the devices of g_list_of_cameras are in the order Camera 1, 2, 3.  If not, it won't work. 
// Note that each "for" loop should only do anything for 1 device and skip all others:
int cn = 0;
for (auto&& dev : g_list_of_cameras)
{
	cn += 1;
	if (cn != camera) continue;

	cout << "resetting camera #" << camera << ", please wait ..." << endl;
	dev.hardware_reset();
	rs2::device_hub hub(ctx);
	dev = hub.wait_for_device();				
	sleep(3);
}

// Re-Start streaming pipe for the reset device 
cn = 0;
for (auto&& serial : serials)
{
	cn += 1;
	if (cn != camera) continue;

	cout << serial << "\n\r";
	rs2::pipeline pipe(ctx);
	rs2::config cfg;
	cfg.enable_device(serial);     
	cfg.disable_all_streams();	 
	waitKey(10);

	cfg.enable_stream(RS2_STREAM_COLOR,    _FWIDTH, _FHEIGHT, RS2_FORMAT_BGR8, _CAM_FRAME_RATE);  
	cfg.enable_stream(RS2_STREAM_DEPTH,    _FWIDTH, _FHEIGHT, RS2_FORMAT_Z16,  _CAM_FRAME_RATE);  
	cfg.enable_stream(RS2_STREAM_INFRARED, _FWIDTH, _FHEIGHT, RS2_FORMAT_Y8,   _CAM_FRAME_RATE);  

	pipeline_profile selection = pipe.start(cfg);
}

Should it be possible to use hardware_reset() on a camera more than once after initial powerup? What might be wrong with what I am doing?

[MartyG-RealSense]

Hi @jpfsaunders There is not a C++ example of resetting a single specific camera by its serial number, though a RealSense team member provided one for Python at #5428 (comment) that takes the approach of generating a list of all attached cameras with the ctx.query_devices() instruction and then querying a serial number in that list.

There is also a C++ reset script at #9287 (comment) that cycles through all attached cameras. This script also uses ctx.query_devices()

[jpfsaunders]

Hi @MartyG-RealSense ,
Thank you for the quick response. The Python example is exactly what I am doing in C++; for me it only works once on powerup and then if I try it a second time after that I get an error. I will look through the C++ reset script to see what they are doing but it looks more involved so it may take me a bit to figure out what is happening there.

[MartyG-RealSense]

Thanks very much, @jpfsaunders - I look forward to your next report. Good luck!

[MartyG-RealSense]

Hi @jpfsaunders Do you have an update about this case that you can provide, please? Thanks!

[jpfsaunders]

Hi @MartyG-RealSense ,
I have not made any progress, and this particular issue has been bumper down on my priority list. The C++ script does not appear to be doing anything different than what I am already doing. About the only thing I can think of is to try to reset the USB controller on my hardware (Jetson Nano) separately, either before or at the same time as performing the hardware_reset(). I am not sure how to do that, but I will experiment once I can get back to looking at this.

I would say we can close this issue. I will open a new one later if I can't figure out a workaround.

[MartyG-RealSense]

Okay, thanks very much @jpfsaunders for the update.