meta-facebook: rainiera7: Rework MCTP lifecycle

MCTP-over-I3C setup is currently controlled by SOC POST-complete
and power-good GPIO readiness targets. Software force-warm reset
also bypasses the full host stop and restart sequence, leaving
MCTP recovery dependent on subsequent GPIO events.

Adapt the ARM Phoenix lifecycle for Rainier. Start MCTP setup from
host-started, wait for C0 POST readiness, and clean up on
host-stopped. Use ExecStopPost for cleanup after stop or failed
initialization, and let systemd retry failed starts.

Retain controller discovery for AST2700 and explicit endpoint
removal before link teardown. Refresh an already-bound controller,
select the first PID on I3C bus 3, normalize it to six bytes, and
assign host EID 9 with local EID 8. This also supports link recovery
after a BMC reboot while the host is powered on.

Restore the host stop and reboot dependencies for software
force-warm reset. Order the existing hardware reset after host
cleanup and before the host restart, without changing its GPIO
pulse. Route SOC_RESETREQ through the host state manager.

Remove the legacy MCTP readiness targets and direct MCTP control
from the GPIO handler in the same change. Keep SOC_POST_COMPLETE
monitoring without MCTP actions, and preserve PMBus mux control,
power-good state synchronization, RTC handling, thermal events,
button logging, and the MCTP-I2C setup.

Tested: With the lifecycle integration on Rainier A7:
- Normal host power-on and DC off/on.
- Software force-warm reset with endpoint removal before reset.
- MCTP recovery after a BMC reboot with the host powered on.
- PLDM endpoint rediscovery and changing sensor readings.

Change-Id: I5f2bd3a4e5c0f06c022182448164fbd34a619f0e
Signed-off-by: Alan Tsai <alan-tsai@quantatw.com>

0 files changed, 0 insertions, 0 deletions