[UPSTREAM PATCH] remove_pcf50633_voltage_rails_and_friends.patch

[Werner Almesberger]

Balaji Rao wrote:
> OK. MEMLDO is used to keep the SDRAM powerd when we are in pmu-standby.

Hmm, interesting. That may actually work. I can't figure out how much
the external SDRAM is drawing, but the MCP SDRAM wants only about 0.5mA.
MCP NAND and NOR Flash each only drawn 10uA. I don't know how much WLAN
draws.

IO_1V8 also feeds STBY_1V2 (aka VDDalive) and Samsung don't specify what
current flows on that one.

Would be good if someone could measure the current on IO_1V8 during
standby. If we can indeed stay within the 1mA budget of MEMLDO, that
would be great.

> One question. Which one of the low power states mentioned in the manual
> do we get into when we suspend ?

That would be PMU.Active and CPU.SLEEP

> > | That's brings one question to my mind. Why do we need to save the
> > | registers that don't "survive" standby (a comment in pcf50633_suspend
> > | says so), if we are not going into standby at all ? Please explain.

Perhaps you were not the first who found all those different sleep
states in different chips confusing ;-)

- Werner