On Thu, Dec 10, 2009 at 06:35:52PM +0900, Kuninori Morimoto wrote:
Because this original driver came from Dialog guy. My work is cleanup only. # Of course we already got acceptance from him.
Could you approve it with /* FIXME */ comment ? If you can not approve it, this time I will remove all of DAPM / widget settings, and use fixed value with "This is very simple settings" comments.
It's a bit tricky, if the code looked incomplete (as the twl4030 driver did originally) it'd not be such an issue but because it's got this odd mix of dynamic and static power management I'm reluctant to accept it as is. The code looks wrong as opposed to incomplete and so I worry that it will be fragile in the face of core development.
What's normally done in situations like this is to hook into either the bias level configuration in set_bias_level() or events on widgets to do any power control that DAPM can't do automatically. More documentation might help as well if it's not actually maintaining the power with these settings.
If it's too difficult to support then removing the DAPM support would certainly be OK for merging - it's obviously far from ideal but it doesn't have the fragility issues.