[Ovmsdev] OVMS hang/crash

[Stephen Casner]

Today I found that the OVMS V3.0 had again become comatose.  Again
when I connected the USB cable to my laptop there was no response.

Does the gdb stub depend upon the USB being connected at the time a
crash occurs?

I didn't like having to disconnect the car cable to power cycle the
module when it gets hung like this, so I implemented an external reset
button by repurposing the red plastic "RESET" button from a recycled
hair dryer GFCI plug.  (This may be a US-only oddity so those of you
who live elsewhere may not have GFCI plugs on hair dryers.)  In the
GFCI the reset button has a shaft that pushes on electrical contacts.
To use it for OVMS I needed to shorten that shaft slightly to match
the distance between the inside of the case and the top of the reset
button and then cut a slot in the OVMS box to fit the oblong shape of
the button.

                                                        -- Steve

On Fri, 18 May 2018, Stephen Casner wrote:

> I took a three-day road trip to Yosemite this week with the v3.0
> module in the car.  It seemed to work fine, communicating over the
> modem except for the spotty reception in the Yosemite Valley due to
> the high stone cliffs.  But back at home, some time after I had
> plugged in to start charging, the module stopped communicating via
> modem or wifi.  I tried plugging in the USB cable to see what was
> wrong, but the /dev/tty.SLAB_USBtoUART device did not get created on
> my laptop so I could not see what went wrong.
>
> At that point, I decided I needed to cycle power to reboot, then the
> module came back up as verified via the home wifi, but the USB was
> still not responding.  Now I have taken the module out of the car and
> brought it in to connect to the laptop and USB is working again.  I
> guess that must imply that the USB physical connection, although the
> cable seemed solidly inserted while in the car.
>
> I have had the module configured to go to gdb rather than reboot on
> crashes since that was helpful for development, but I may need to
> change that setting to use the module in the car.  On the other hand,
> with small RAM of the v3.0 module I don't think I can afford to have
> the SD card plugged in for logging, so being able to investigate
> crashes via USB might be best for now if I can figure out why that
> didn't work this morning.
>
>                                                         -- Steve