On Tue, Oct 30, 2018 at 5:08 AM Linus Walleij <linus.walleij@xxxxxxxxxx> wrote: > > On Wed, Oct 17, 2018 at 11:30 PM Kun Yi <kunyi@xxxxxxxxxx> wrote: > > > Indicate that the pins are both controlled by the pinctrl driver and the > > generic GPIO driver, thus GPIO driver should read the register value > > before updating, instead of using the stored shadow register values. > > > > Signed-off-by: Kun Yi <kunyi@xxxxxxxxxx> > > This is quite a rough measure, if we instead use regmap-mmio > we can exercise fine control over what register are volatile and > not instead of saying that all of them or some of them are. Thanks for your review Linus! I don't have time to rewrite using regmap-mmio at the moment though. I have discussed with the driver author and we will first patch the pinctrl driver by making the pinctrl functions use the gpio-mmio accessors instead of directly reg read/writes. When I have time I will look into your suggestion to improve the driver. > > Yours, > Linus Walleij -- Regards, Kun