Hurricos
commented
9 months ago Another WIP issue: I don't check clutch state. It's possible that on the MT, with the clutch closed and car in gear, this would attempt to move the car -- but in the CVT, while stopped and with engine off in D, the ATPNP signal value is misleading -- the start clutch is depressurized.
Getting around that MT/CVT logic mismatch might be doable if we abstract the car's neutral state behind a function. We can ifdef to separate MT from CVT, then take the clutch-state for the MT directly to determine whether to inhibit start; for the CVT, a motionless engine and wheels guarantees a neutral position -- only if those aren't both true should we check ATPNP.
This is a WIP because this isn't the ideal way to write this code.
Before this commit, after using regen FAS in mode 2, the ECM will spend some time attempting to restart the engine (i.e. switching to mode 0 would give us an auto-restart), but eventually give up.
Now, when the engine is considered "force-stopped", translate the momentary button to MAMODE1_STATE_IS_START. Pressing the momentary button will mark mode-START; if immediately followed by a return to mode ASSIST plus some CMDPWR, the engine will start -- this is what you get when you press momentary + flick the joystick.
(I assume the MAMODE1 START state tells the MCM it's OK to be torquey, or that it needs to do some special magic to resolve the motor position.)
Note that this code is not tested against actual auto-stop. I have tested it against a key-off at neutral, and the engine DOES start ;^)
WIP because this isn't the ideal way to do this: we should really be instead generating a state and intentionally walking a bit of assist into the engine to re-start it.
No s-o-b yet, let's fix this WIP.