hardware pin reinitialization on reset()

[ponyatov]

I have problems in reset() function running on STM32F429-based board with external SDRAM enabled: #7765

I found a code point where system deathly halts:
reset() -> jsvReset -> jspSoftInit -> jsiSemiInit

Looking in a hardware debugger, I see that FMC controller gets resetted in this call chain, so the whole jsVar[] memory becomes unavailable.

What points in an Espruino core can make FMC reset (or maybe IO pins reassign) ?

[gfwilliams]

I believe it's happening in jshResetPeripherals - that's the only place: https://github.com/espruino/Espruino/blob/c39547da361314385d12ae092e7210feafd38702/targets/stm32/jshardware.c#L1151

All you have to do is make sure the IS_PIN_USED_INTERNALLY definition in platform_config.h lists those pins.

It's created here: https://github.com/espruino/Espruino/blob/c39547da361314385d12ae092e7210feafd38702/scripts/build_platform_config.py#L554-L557

so if you have one of OTHER_DEVICES listed in the BOARD.py file and it's got all the pins you use listed under it, the pins won't get reset:

https://github.com/espruino/Espruino/blob/c39547da361314385d12ae092e7210feafd38702/scripts/pinutils.py#L125-L128

You can see we do it in BANGLEJS with MISC:

https://github.com/espruino/Espruino/blob/c39547da361314385d12ae092e7210feafd38702/boards/BANGLEJS.py#L164-L167