On 05/23/2016 02:49 AM, Uwe Kleine-König wrote: > Hello Yegor, > > On Mon, May 23, 2016 at 11:40:44AM +0200, Yegor Yefremov wrote: >>> With the same reasoning that gpiod_get_optional returns -ENOSYS without >>> GPIOLIB, this also should return -ENOSYS. >> >> OK. What do think about such workaround in >> drivers/tty/serial/8250/8250_core.c for not breaking the UART >> initialization, when GPIOLIB is disabled? >> >> if (IS_ERR(uart->gpios) && IS_ENABLED(CONFIG_GPIOLIB)) >> return real error; Don't assign uart->gpios directly. gpios = mctrl_gpio_init(&uart->port, 0); if (IS_ERR(gpios)) { if (PTR_ERR(gpios) != -ENOSYS) return PTR_ERR(gpios); } else mctrl->gpios = gpios; > If your 8250 has a rts gpio specified in the device tree, what do you > expect from a kernel without GPIOLIB support? > > If it's important that RTS is setup correctly, you should return > failure above. If it's not important that RTS is handled correctly, then > think again what this implies for a GPIOLIB=y .config. Can you just drop > RTS handling then? I want to emit a diagnostic for: 1. an actual resource acquisition error 2. a misconfiguration The driver should fail probe for actual resource acquisition error. The driver should continue anyway for misconfiguration with mctrl gpio api calls nop'd. Regards, Peter Hurley -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html