ALSA sound suspend/resume

[Andy Green]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:
| Am Mi  11. Juni 2008 schrieb Andy Green:
|> Somebody in the thread at some point said:
|>
|> |> to do the same, when setup and sequenced properly (on 8753 there are).
|> |> I'll come back with more, when I read about it in the amp's specs
|> |> eventually. Which audio output is it we talk about (hp, earphone, spk?
|> |> BT even? GTA01/02?)
|>
|> | It is the speaker output I have set when I suspend/resume so I
suspected
|> | the speaker amp.
|> |
|> | Thanks for the info. Any help is greatly appreciated.
|>
|> WM8753 datasheet (p89) explicitly goes through the steps to power stuff
|> down without pops (was reading this yesterday), but it didn't bank on an
|> external amp in the way.  You're meant to disable the output pins and
|> then kill Vref, then pull power.  I guess we should do that and then
|> disable the external amp.  I'll have a go at it.
|
| I think we should shut down 4853 first, and start last. It will do
whatever

Suspend for the codec actually only shuts down the LM4853 already it seems.

But as we are getting used to now, there is an implicit race here with
pcf50633 suspend, which will pull the codec power without any ceremony
when it fires (AMP power is "always on").  So I guess the pop is amp
still up, codec did not suspend yet, but pcf50633 suspended first and so
pulled the plug on codec in an uncontrolled way that was then amplified out.

So I think I will meddle a little bit with pcf50633 suspend only going
once, late similar to Sean's patch yesterday only on the last chance not
the first chance, see what horrible bugs shake out of that.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkhPoMUACgkQOjLpvpq7dMo/XQCfcx8a1vhHSrgt4/G0UwtskI/P
UdMAn3e3LTSJEWWi6K1271BSImVJJFaF
=0q4F
-----END PGP SIGNATURE-----