ot_lc_ctrl: Set LC_STATE to POST_TRANSITION after a transition

[luismarques]

From https://opentitan.org/book/hw/ip/lc_ctrl/doc/programmers_guide.html:

"Note that all life cycle state transition increments the LC_TRANSITION_CNT and moves the life cycle state into the temporary POST_TRANSITION state - even if the transition was unsuccessful"

This is checked by manufacturing tests (on the host side).

[rivos-eblot]

LGTM, although I have not cross checked with the RTL implementation.

[rivos-eblot]

Have you double checked that LC_STATE_POST_TRANSITION was properly handled in all cases?
For example, LC_ENC_STATE_POST_TRANSITION is defined but not decoded from ot_lc_ctrl_convert_code_to_state(), which would trigger a g_assert() that would be invalid. I'm not sure this case may happen, but it is weird it is defined and not used.

There are other locations where I think LC_STATE_POST_TRANSITION is guarded behind ot_lc_ctrl_safe_convert_code_to_state() but I have not reviewed all the sequences.

We need to be sure that g_assert() cannot be triggered by the guest SW and that LC_STATE_POST_TRANSITION is not stored into the OTP backend.

[luismarques]

Have you double checked that LC_STATE_POST_TRANSITION was properly handled in all cases?

I will further check things.