[Letux-kernel] GTA04 voice routing

[H. Nikolaus Schaller]

Hi,
I am working on fixing the voice routing with DT
based kernels and it starts to work partially.

I want to share some log...

root at letux:~# amixer set 'Voice route' 'Voice to SoC'
[   82.491699] voice ctl route: 0
[   82.499938] devm_pinctrl_get ed460040
[   82.503906] twl4030_voice_set_tristate codec=ed0bea00 1
[   82.513488] pinctrl_lookup_state default error -19
Simple mixer control 'Voice route',0
  Capabilities: enum
  Items: 'Voice to SoC' 'Voice to twl4030'
  Item0: 'Voice to SoC'
root at letux:~# amixer set 'Voice route' 'Voice to SoC'
[   87.971649] voice ctl route: 0
Simple mixer control 'Voice route',0
  Capabilities: enum
  Items: 'Voice to SoC' 'Voice to twl4030'
  Item0: 'Voice to SoC'
root at letux:~# amixer set 'Voice route' 'Voice to twl4030'
[   90.949249] voice ctl route: 1
[   90.953491] devm_pinctrl_get ed460040
[   90.959991] pinctrl_lookup_state tristate error -19
[   90.967468] twl4030_voice_set_tristate codec=ed0bea00 0
Simple mixer control 'Voice route',0
  Capabilities: enum
  Items: 'Voice to SoC' 'Voice to twl4030'
  Item0: 'Voice to twl4030'
root at letux:~#

What does not yet work is looking up the pinctrl_lookup_state
"default" and "tristate".

The reason we need this is that we have a shared bus for the
PCM data.

For incoming voice the gtm601 module is the master and either
the twl4030 and the McBSP4-DR of the OMAP are clients (or even both).

For outgoing voice we have to make either the twl4030 (where
the microphone is connected to) of the McBSP-DX line the master.
Both potential masters are connected in parallel to the single
PCM input of the gtm601 module.

To avoid data mix and corruption, we have to tri-state either
one. This is already done by the twl4030-codec driver, but
tristating the McBSP-DX line is not as easy. The best solution
appears to be to provide two different pinmux states in the DT
and make the twl4030-codec choose either one.

Next I have to dig into the code to find out where to specify
it in the DT so that it is really found...

But the code itself seems to compile. I will push it later and
provide a link so that you can look inside (needs some cleanup
first).

BR,
Nikolaus