On Thu, Jun 13, 2013 at 11:53 AM, Tony Lindgren [off-list ref] wrote:

* Linus Walleij [off-list ref] [130613 02:42]:

quoted

On Thu, Jun 6, 2013 at 9:14 PM, Balaji T K [off-list ref] wrote:

quoted

+       /* 100ms delay required for PBIAS configuration */
+       msleep(100);
+       if (!vdd && host->pinctrl && host->pbias_off) {
+               ret = pinctrl_select_state(host->pinctrl, host->pbias_off);
+               if (ret < 0)
+                       dev_err(host->dev, "pinctrl pbias_off select error\n");
+       } else if (((1 << vdd) <= MMC_VDD_165_195) && host->pinctrl &&
+                       host->pbias_1v8) {
+               ret = pinctrl_select_state(host->pinctrl, host->pbias_1v8);
+               if (ret < 0)
+                       dev_err(host->dev, "pinctrl pbias_1v8 select error\n");
+       } else if (((1 << vdd) > MMC_VDD_165_195) && host->pinctrl &&
+                       host->pbias_3v) {
+               ret = pinctrl_select_state(host->pinctrl, host->pbias_3v);
+               if (ret < 0)
+                       dev_err(host->dev, "pinctrl pbias_3v select error\n");
+       }

So why does the pin control API control bias voltage?

I agree, it should be a regulator for the MMC driver and that's what I
already suggested earlier. Having it as a regulator allows us to get
rid of all the non-standard before and after calls in the omap_hsmmc.c.
This way the omap_hsmmc.c code can handle the internal and external
voltages the same way.

So someone is not taking the OMAP maintainers word for it and
instead trying to push this past the pinctrl maintainer?

Grrr.... OK I'll assume good faith and I therefore conclude that
this must have been just some kind of mistake in the act.

Yours,
Linus Walleij