Getting odd values with M0 (other?) boards

[caternuson]

Re this thread: https://forums.adafruit.com/viewtopic.php?f=19&t=175010

I've recreated same behavior on QT PY:

And also Itsy M0:

But seems to work just fine on Itsy M4:

[ladyada]

!? can you try turning on i2c debug in busio?

[caternuson]

This is on QT PY, running the adafruit_scd30_test example, and with #define DEBUG_SERIAL Serial enabled in Adafruit_I2CDevice.cpp, and uploading after a complete power cycle.

Looks like the reset failed maybe?

[ladyada]

can you compare the data trace with an M4?

[caternuson]

Itsy M4

[ladyada]

try initing these to zero https://github.com/adafruit/Adafruit_SCD30/blob/master/Adafruit_SCD30.cpp#L293

[ladyada]

also this is running past the buffer end https://github.com/adafruit/Adafruit_SCD30/blob/master/Adafruit_SCD30.cpp#L286

[caternuson]

OK, back to QT PY with initing those values to zero. But nothing done for the buffer index.

[ladyada]

yay ok! PR plz :)

[caternuson]

nice spot! thanks.

guess we'll see if/when that buffer index issue causes grief.

[caternuson]

Looks resolved: https://forums.adafruit.com/viewtopic.php?f=19&t=175010#p853221

Thanks again @ladyada for the assist!

[kevinjwalters]

This sort of thing (stack variables used before initialisation) could easily be detected by static analysis of the code as a prevention measure in the same way that pylint runs over the CircuitPython libraries.

The gcc/g++ compiler has enough smarts in it to pick this up too with -Wuninitialized and/or -Wmaybe-uninitialized. Antique versions of gcc may incorrectly inhibit this due to confusion from memcpy().

[ladyada]

@kevinjwalters yes, please add a check update to the Arduino CI!