Hi. 2017-02-17 13:08 GMT+09:00 James <ja...@balean.com.au>: > Hi All, > > I have been following this work. Masahiro, perhaps have another look at > pinctrl_select_state_simple (drivers/pinctrl/pinctrl-uclass.c) as this > could be where the confusion lies. This function calls set_state_simple > operation with the pinctrl node as the first 'dev' parameter, and the > former 'dev' as the second 'periph' parameter.
I know how this function works. As you see "git blame drivers/pinctrl/pinctrl-uclass.c", this part (and most of pinctrl uclass) was written by me. (the idea of the _simple_ pinctrl driver is from Simon, though.) I hope you and Felix will check other .set_state_simple() callbacks of other pinctrl drivers. See ar933x_pinctrl_set_state_simple() qca953x_pinctrl_set_state_simple() pic32_pinctrl_set_state_simple() rk3036_pinctrl_set_state_simple() rk3288_pinctrl_set_state_simple() rk3399_pinctrl_set_state_simple() All of them parse the "interrupts" property of the argument "periph". I hope you will understand the argument "periph" represents a peripheral device (like UART, I2C, etc. that are associated with interrupt lines). > My testing indicates that this patch results in the correct pinmux setup. I > have added support for 8 and 16 bit register widths, and will contribute > this to Felix. For Felix's driver to work, the argument "periph" must be a pin configuration node (= child of pinctrl device node). This comes from the "dev" argument of device_probe(). How can you probe the pin configuration device? -- Best Regards Masahiro Yamada _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
