On Monday 26 of December 2011 at 18:23:50, Mark Brown wrote:
On Mon, Dec 26, 2011 at 01:30:22PM +0100, Janusz Krzysztofik wrote:
Why's this code not been dropped and what is it for?
It is, and will be, used by another regulator consumer, the modem. The device is controlled by the generic serial8250 driver, extended with a custom .pm hook which enables/disables the regulator. The above code, temporarily shared with the ASoC device before updated, is now just moved down the board file where the modem related bits are located (see patch 3/4).
Any hints like that for the codec driver?
The main thing that looks bad here is that you're keeping track of the state of the regulator by hand. It may be that the upper layers aren't reference counting in which case you might need this but it certainly looks very suspicious as-is.
OK, I can try to examine how the serial8250 driver tracks the device power state when calling the .pm hook and if it looks safe to drop that regulator state manual tracking.
Thanks, Janusz