LinuxCNC disables Mesa 5i24 - not as serious as it sounds

This is kind of a tips 'n tricks. Maybe it will help someone with a similar problem. Note that the text below was specific to my hardware and configuration so you may have to make changes.

Problem:
After maximizing the LinuxCNC Axis window after startup, the PWM EN line would remain high (on the Mesa 5i24, (enable is active low) and the motor would no longer run. Toggling Machine Power and Emergency Stop had no effect. LinuxCNC was still alive, but unable to process the job.

Reason:
The PC (almost 20 years old now), could not maximize the window and at the same time pet the watchdog timer. A message stating "Unexpected real time delay: check dmesg for details" appears at the lower right corner of the LinuxCNC Axis window.

Using dmesg > hm.txt, the last message is: "hm2/hm2_5i24.0: Watchdog has bit! (set the .has_bit pin to false to resume).

The watchdog timer is documented here . A key line from that document is: "When the watchdog bites, all the board’s I/O pins are disconnected from their Module instances and become high-impedance inputs (pulled high), and all communication with the board stops."

Solution:
The long term solution is to eliminate the real time delay. This may involve new hardware, such as moving to a PC that was produced in this century.

The short term solution is to reset the .has_bit to TRUE after the real time error has occurred. The status of the has_bit flag can be confirmed by opening Machine/Show Hal Configuration/Pins/hm2_5i24/0/watch_dog/has_bit. It should show that it is TRUE. To set it to FALSE execute the following command in the "Test HAL command:" dialog box: "setp hm2_5i24.0.watchdog.has_bit FALSE".

LinuxCNC should now operate normally….at least until the next real time error. A new PC is almost ready.

Just remember an "Unexpected real time delay" warning message is only ever displayed once per Linuxcnc session. If further real time delay excursions occur, you will not be informed of them. You really should address that problem, because it opens the door for other even nastier things to potentially occur. (an uncontrolled runaway can be dangerous to life, limb, and property.)

Todd,
Agreed.

The PC that caused this problem is being used in a lab environment - the motor is not connected to anything. This system is being used to test out some custom built hardware prior to the arrival of an item needed for the new, much faster, PC.

Its also a bug in some versions of LinuxCNC that the watchdog bite is not reported (it should be reported like the real time error)