SerialCommS300: more than one packet is buffered, measurements are potentially stale!

[mertmzk]

Hi

When I kill the node using "killall -9 rosmaster" and restart the node it gives above warning and does not read from lidar. I have to reboot the computer to get it working again.

Any suggestion to solve the problem.

Thanks

Mert

[bohlender]

Hey Mert,

unfortunately I'm working at a different place now, and don't have access to a S300 anymore. As a result I cannot debug your issue but merely give an educated guess on where to start digging ¯\_(ツ)_/¯

The warning you're getting comes from here and suggests that restarting the node after a hard termination (SIGKILL) does not re-initialise the serial I/O properly - the buffer is still full/in use. It might suffice to replace the current flushing of the device byioctl(fd, TCFLSH, 2);, however there's no way for me to try.

If you manage to identify the issue, I'd be happy to merge your fix.

[mertmzk]

Hi Bohlender,

Thanks for your quick response. I have tried what you suggested but it didnt worked either.

I found a way around to it; just killing the node using "rosnode kill laser" does not produce the warning on the next run.

Thanks

[bohlender]

I'm glad to hear that using rosnode kill laser is a workaround. However I'll keep the issue open - it would be nice if someone were to fix it to work as expected even in the case of hard termination.