Consider having deep sleep preserve_dios preserve the pin state through code.py starting

[dhalbert]

Currently preserve_dios only preserves pin state during deep sleep. Once the board awakens, the pin state is reset to default. Consider preserving the state.

I think the concept of preserve_dios should be revisited if Pin.Configuration or something similar is implemented: #9845

[bablokb]

Hm, there are two problems:

• where would you save the state? E.g the rp2040 does not have enough scratch registers, so you would need some sleep_memory that is not reinitialized.
• even if you could save the state: if you configure and setup the pins, how would user code get hold of them again?

I think it would be more efficient to invest time and effort instead into the "third sleep-mode", be it a deep-sleep without reset or a light-sleep with aggressive power-saving. Preserving dios is just one aspect of keeping state.

[dhalbert]

@bablokb It is an aspirational goal, maybe not possible, and I'm not how needed it is. I remember one user was surprised that the pin state was not maintained on restart. Do people need glitch-free pin holding, or is it just long-term pin-holding (through deep sleep) enough? If glitch-free pin state is needed, it's perhaps better done in hardware. ... I see now that #6732 has already raised this issue. I will close this issue in favor of that.

I think it would be more efficient to invest time and effort instead into the "third sleep-mode", be it a deep-sleep without reset or a light-sleep with aggressive power-saving. Preserving dios is just one aspect of keeping state.

I agree with this. I am not sure the current light sleep has any utility, given its "no state change" promise. time.sleep() in many ports already essentially does this, if possible. #9521 is relevant.

[tannewt]

Maybe the GPIO registers aren't reset either?