Fixes SAMD PDMIn

[dhalbert]

• Fixes metro-m4 call to PDMIn.record() never returns #5797 (both on SAMx5x and SAMD21). Tested with a sound level program on CPX and with a simpler program on Metro M4.

SAMD PDMIn has not worked right for a while.

• The main problem was that the needed IRQ was never actually enabled.
• Made a flag variable volatile, as it should have been.
• Called to pdmin_reset() was commented out, maybe left over from some debugging.

I will probably backport this to 7.1.x, but there is maybe one other unrelated regression that should be checked for a 7.1.1 release first.

[ladyada]

didnt test but.... is there any linter that will warn if irq code is accessing a non-volatile variable? i feel like its such a killer bug that keeps popping up in microcontroller code :/

[dhalbert]

didnt test but.... is there any linter that will warn if irq code is accessing a non-volatile variable? i feel like its such a killer bug that keeps popping up in microcontroller code :/

Totally agree. I looked at the ones in https://github.com/caramelomartins/awesome-linters and also did some general searching; I could not find any checkers, but it would not be so hard as long as IRQ routines were labeled as such.

[tannewt]

Thank you for the fix!