ESP32-S2: Crash into the HardFault_Handler

[ctmorrison]

CircuitPython version

Adafruit CircuitPython 7.3.2 on 2022-07-20; Adafruit QT Py ESP32S2 with ESP32S2

Code/REPL

# Garage Vehicle Space Detection

import alarm
import time
import board
import adafruit_vl53l1x
import neopixel
from digitalio import DigitalInOut, Direction, Pull
from analogio import AnalogIn

# This is the latching relay
unoccupied = DigitalInOut(board.A0)
unoccupied.direction = Direction.OUTPUT
occupied = DigitalInOut(board.A1)
occupied.direction = Direction.OUTPUT

# This is the button used to put the device into setup mode
setup = DigitalInOut(board.A2)
setup.direction = Direction.INPUT
setup.pull = Pull.DOWN
# read value with setup.value, knowing it has a pullup
# We'll use this to know the state of the relay
occupiedStatus = DigitalInOut(board.SDA)
occupiedStatus.direction = Direction.INPUT

# This is the potentiometer input
pot = AnalogIn(board.A3)
tripPoint = pot.value * 400 / 65536

# Length of time in seconds to sleep between checks
sleepPeriod = 15
# Length of time in seconds to turn on neopixel to indicate if space is occupied or not
spaceStatusPeriod = 0.5
# Amount of time to energize the relay (in seconds)
energizeTime = 0.25

# We're using the Stemma interface to connect to the sensor
i2c = board.STEMMA_I2C()

# This code works with the VL53L1CX ToF sensor
vl53 = adafruit_vl53l1x.VL53L1X(i2c)

# The onboard neopixel will be used for visual signalling
pixels = neopixel.NeoPixel(board.NEOPIXEL, 1, brightness=0.1)

# OPTIONAL: can set non-default values
vl53.distance_mode = 2
vl53.timing_budget = 100

print("Garage Space Detection")
print("Version 20220812.1")
print("--------------------")
model_id, module_type, mask_rev = vl53.model_info
print("Model ID: 0x{:0X}".format(model_id))
print("Module Type: 0x{:0X}".format(module_type))
print("Mask Revision: 0x{:0X}".format(mask_rev))
print("Distance Mode: ", end="")
if vl53.distance_mode == 1:
    print("SHORT")
elif vl53.distance_mode == 2:
    print("LONG")
else:
    print("UNKNOWN")
print("Timing Budget: {}".format(vl53.timing_budget))
print("Sleep period: {}".format(sleepPeriod))
print("Trip point: {}".format(tripPoint))
print("Space status period: {}".format(spaceStatusPeriod))
print("--------------------")
vl53.start_ranging()

# Attempt to read a distance
while vl53.data_ready is not True:
    print("not ready")
    time.sleep(0.5)
print("Sensor is ready to read")

# Let's get a distance
distance = vl53.distance
while distance is None:
    distance = vl53.distance
    pixels.fill((0, 0, 255))
    time.sleep(0.1)

while setup.value is True:
    # the Setup button is being pressed and we need to set the  occupied distance
    # the occupied distance should be just above the floor
    # this assumes the space is EMPTY!
    print("In setup mode")
    tripPoint = pot.value * 400 / 65536
    # check to see if sensed distance is
    print("Floor: {}".format(distance))
    print("tripPoint: {}".format(tripPoint))
    if tripPoint > distance:
        print("tripPoint is beyond floor - adjust to lower value")
        pixels.fill((255, 0, 0))
    else:
        print("tripPoint is above floor - make sure it's close to floor")
        pixels.fill((0, 255, 0))
    time.sleep(1)
print("Setup finished")

try:
    pixels.fill((0, 0, 0))
except RuntimeError:
    print("10***************************************************pixels.fill failed")

# Read the distance to the nearest object
distance = vl53.distance

# test for no distance read to avoid error
while distance is None:
    distance = vl53.distance
    pixels.fill((0, 0, 255))
    time.sleep(0.1)
try:
    pixels.fill((0, 0, 0))
except RuntimeError:
    print("20***************************************************pixels.fill failed")


# here's where we check to see if there's a vehicle detected above the threshold
if distance < tripPoint:
    print("Distance: {} cm -> vehicle sensed".format(distance))
    try:
        pixels.fill((255, 0, 0))
    except RuntimeError:
        print("30***************************************************pixels.fill failed")
    if occupiedStatus.value is False:
        # we only energize if we need to -- it was previously unoccupied
        occupied.value = True
        time.sleep(energizeTime)
        occupied.value = False
    time.sleep(spaceStatusPeriod)
else:
    print("Distance: {} cm -> no vehicle sensed".format(distance))
    try:
        pixels.fill((0, 255, 0))
    except RuntimeError:
        print("40***************************************************pixels.fill failed")
    if occupiedStatus.value is True:
        # we only energize if we need to -- it was previously occupied
        unoccupied.value = True
        time.sleep(energizeTime)
        occupied.value = False
    time.sleep(spaceStatusPeriod)
vl53.clear_interrupt()

# Set up the sleep period for the amount of time specified in sleepPeriod
time_alarm = alarm.time.TimeAlarm(monotonic_time=time.monotonic() + sleepPeriod)
# Exit the program and then deep sleep until the alarm wakes usage
print("Going into Deep Sleep")
alarm.exit_and_deep_sleep_until_alarms(time_alarm)
# Does not get here...program starts at the beginning after sleep

Behavior

This does not always happen. Sometimes, it runs just fine.
Here's the error I got:
You are in safe mode because:
CircuitPython core code crashed hard. Whoops!
Crash into the HardFault_Handler.
Please file an issue with the contents of your CIRCUITPY drive at
https://github.com/adafruit/circuitpython/issues

Description

No response

Additional information

The only error I seem to get often, but not always is when the program executes a call to pixels.fill((255, 0, 0)) or some other color combination

[dhalbert]

How often would you say it happens? If we try to reproduce, it would be good to know how long to wait for a crash.

Ideally we would reduce this program to a very simple example, to see what is significant and what is not. For instance, if you faked the use of the VL53L1X, or if you removed the deep sleep, does it still crash? Thanks.

[ctmorrison]

It usually runs for hours before it crashes. I might try to put some additional print statements in to see if I can zero down where it’s crashing. I’m away for the next few days, but will try when I return. I hope we can resolve this or I’ll be forced to rewrite the app in Arduino—not a bit problem, but it will turn me off on CircuitPython.

…

On Aug 25, 2022, at 8:51 AM, Dan Halbert ***@***.***> wrote: How often would you say it happens? If we try to reproduce, it would be good to know how long to wait for a crash. Ideally we would reduce this program to a very simple example, to see what is significant and what is not. For instance, if you faked the use of the VL53L1X, or if you removed the deep sleep, does it still crash? Thanks. — Reply to this email directly, view it on GitHub <#6793 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABODJI74CDPDHHH3PEVLCMDV25T5BANCNFSM57LTHGYQ>. You are receiving this because you authored the thread.

[DJDevon3]

I'm getting an identical crash error except mine happens upon usb wake or at least when I touch a file with Mu. Admittedly I haven't tried slimming my code to figure out if it's something specific in the code causing it. Will do some testing and report back.

I’ve been over ctmorrisons code and our programs are widely different. About the only similarity I could find was we’re both using try/except within the main while loop. Possible the runtime debug itself is causing a code.py fail to turn into a hardfault crash?

[dhalbert]

@ctmorrison I have been trying to reproduce this. I have been running your test program for about 16 hours, without a crash, on CircuitPython 8.0.0-beta.5. I reduced the sleep time from 15 seconds to 2 seconds to speed up the possibility of crashing.

Is this battery-powered? Might your battery be getting low when it starts crashing? I'm testing with a 5V AC adapter, so I am doing true deep sleep, but the power is steady.

[ctmorrison]

The device is battery powered and the battery is sufficiently charged. I've moved to Arduino code to run the application, so this exercise is rather academic at this point in time. However, I appreciate your attempt to find the issue. Happy Holidays!

[dhalbert]

Thanks for the reply. I will close this for now.