PyPortal crash -- EventCountdown example

[jerryneedell]

I am finding that the PyPortal EventCoundown example causes a (HardFault) reboot to safe mode with the current CP Master build. https://github.com/adafruit/Adafruit_Learning_System_Guides/tree/master/PyPortal_EventCountdown Warning -- it may also corrupt the FS. I have had varying results with this -- Sometimes it runs but the background splash disappears. Sometimes it crashes....

Run on:
Adafruit CircuitPython 4.0.0-beta.5-17-gfce63b17c on 2019-03-21; Adafruit PyPortal with samd51j20

This example ran successfully on previous CP versions -- not sure of latest version tested, but I think I tried Beta5

[dhalbert]

I have tried this a few times. I am consistently seeing the background disappear, but otherwise it hasn't crashed. Does it crash right away, or should I let it run a while?

[jerryneedell]

@dhalbert With the default settings, it does not crash immediately -- infact, it seems to be running OK today,
However, if I cahange the event date to:

14,16c14,16
< EVENT_MONTH = 6
< EVENT_DAY = 28
< EVENT_HOUR = 17
---
> EVENT_MONTH = 4
> EVENT_DAY = 15
> EVENT_HOUR = 9

it crashes immediately on first update. That is, I see the background, then after a few seconds, it crashes.

Update -- I cannot reproduce the failure with the default setting, but it still crashes if it change the date as noted above -- I also tried EVENT_MONTH = 5 and it still crashed...

[jerryneedell]

very strange -- I run eventcountdown.py on the PyPortal, it runs OK -- I then copied eventcountdown.py to new_copy.py -- When I run new_copy.py it crashes...
a diff of the files does not report any difference.

In all cases, I am running this from the REPL not code.py.

even stranger, any of the examples (that is different dates) run OK if loaded as code.py.....
(the background still disappears)
My head hurts...

[deshipu]

I think it might depend on where the file is in the partition table, how it is split into blocks, and other things like that. On a non-express board you could just copy the current.uf2 file to reproduce the whole filesystem exactly, perhaps we could have some tool to dump the filesystem off the flash chip on the express boards as well?

[dhalbert]

The essence of this bug is described in #1695. I don't know about the hard crashes, but that may be yet some other bug, or else it's a manifestation of the #1695 bug. Could close this in favor of #1680.

[dhalbert]

I think this is fixed by #1714. Could you reopen if you are still seeing crashes? Thanks.