In order to pulse RTC reset to the PCH when power sequencing exit fails we
need to watch for SLP_S4 to deassert and if it does not then assert RTCRST
using a board specific method. This is attempted up to 5 times before giving
up and staying in G3.
On skylake the RSMRST passthru needs to be honored when the task is woken up,
so while waiting call handle_rsmrst() if woken up early. This is needed
because it is RSMRST that actually tells the PCH to try and wake.
This is all wrapped in a config option and board specific method because not all
boards have a GPIO to control RTCRST and if they do they may not all use the
same method to assert it.
BUG=chrome-os-partner:49564
BRANCH=glados
TEST=manually tested on chell EVT:
First, ensure board sequences properly if everything is OK for a normal boot.
Next, modify handle_rsmrst() to not pass through the signal in order to
simulate being stuck in S5, and ensure that the EC attempts to assert RTCRST
and power up again 5 times before giving up and staying in G3.
Change-Id: Ia3c13069c92762b51beb682a19e5a074194a3c26
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/322724
Reviewed-by: Shawn N <shawnn@chromium.org>
bbe2d886dc