aux button interrupt to pmu?

Carsten Haitzler (The Rasterman) raster at openmoko.org
Fri Jun 6 17:09:42 CEST 2008


On Fri, 06 Jun 2008 11:57:06 +0100 Andy Green <andy at openmoko.com> babbled:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Somebody in the thread at some point said:
> |
> |> | Yes, I think this will have compatible issue if user don't have debug
> |> | board for re-flash the NOR.
> |>
> |> Can be, but we can maybe work around it.  For example we can look in
> |> resume function of "keyboard" that deals with buttons to see if AUX is
> |> still down when it resumes, and infer from this that we woke from AUX,
> |> set the AUX resume reason bit right then.
> |
> | Difficulty is user usually don't count how many second they press on
> | key, so the result come from user behavior will be un-predictable during
> | resume. Like press too long/twice or more/too short/repeately suspend.
> | User behavior is hard to predict even we only got 2 key in GTA02 :)
> 
> Well as Werner described, if AUX is only pressed for a few ms, something
> BAD can happen and we crash.  But if it is pressed long enough to get
> through steppingstone, then it is not critical if it is still held down
> when Linux comes.  All that will happen then is we are not sure about
> wake source, but we are still properly resumed and doing whatever it is
> we were doing again.

makes sense - and to me that would say it rules it out as a wake source... :(

> |> Carsten mentioned it doesn't wake at the moment, I think it's probably
> |> because it isn't enabled as wake source.  If it does wake through to
> |> Linux OK from AUX -> NOR U-Boot -> Linux then the workaround above will
> |> probably work for even current NOR U-Boot.
> |
> | I don't know if NAND and NOR have different init code, does this affect?
> 
> No need for it to affect anything, whether it does or not Werner can know.
> 
> Wake starts in U-Boot again like a cold reset, but very early if it sees
> it is waking from suspend, it jumps to "magic address" of Linux resume
> stuff in RAM that Linux stored into a special CPU register that survives
> no core power.  So not much happens to be different during wake in
> U-Boot path.  Still Werner is right logically there are two diverged
> chunks of code there and logically they can differ for whatever reason.
> 
> |> Hardware design is a little bit older than this :-)
> |
> | Yup, we are doing software-able hardware ;)
> 
> Nobody told this to the guy that did the AUX key -> NOR design ;-)
> 
> - -Andy
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
> 
> iEYEARECAAYFAkhJGAIACgkQOjLpvpq7dMp6aACglZtjCpe18XBm3g44nxCGhBKQ
> kAoAniFnakPeHyuwmJ1isW8GVFgtnn+8
> =Eh9i
> -----END PGP SIGNATURE-----


-- 
Carsten Haitzler (The Rasterman) <raster at openmoko.org>




More information about the openmoko-kernel mailing list