On Wed, 01 Feb 2017 15:59:56 +0100, Mark Brown wrote:
On Wed, Feb 01, 2017 at 01:52:22PM +0100, Takashi Iwai wrote:
Mark Brown wrote:
No, it's complicated. IIRC it's not *actually* doing GPIO management, it's adding an option to wiggle a pin on the one board you've so far that happens to have this issue but if someone finds a GPIO from somewhere else it'll run into trouble.
How that can happen? It applies only via DMI matching.
Sure, it works on that one machine but if someone else did something similar but using a different GPIO it (from what I remember) might not work.
For other GPIO pin, you'll need to write another similar codes, sure. But why do you need to worry so much about it? It doesn't work even without this patch, and a complete generalization isn't so trivial, I'm afraid, as it's anyway pretty device-specific.
I think I was going to try to figure out if it was really a GPIO or if perhaps the changelog is just not clear.
The changelog looks fairly short and maybe not sufficiently explaining the actual code changes. Basically the patch gives a new DMI-based quirk entry to add a control of GPIO2 pin via DAPM per headset detection.
The single deleted line is the removal of the GPIO_CTRL2 reg write at headset detection, and according to Bard, this was superfluous and should have been removed.
But anyway, content free ping so...
I'm not the author of this patch, either, but just a by-stander casually involved with the development, so I leave the rest to Bard :)
thanks,
Takashi