On 05/24/2016 01:09 AM, Uwe Kleine-König wrote: > Hello Peter, > > On Mon, May 23, 2016 at 11:02:53AM -0700, Peter Hurley wrote: >> On 05/23/2016 02:09 AM, Uwe Kleine-König wrote: >>>> gpios->port = port; >>>>> >>>>> @@ -193,7 +202,7 @@ struct mctrl_gpios *mctrl_gpio_init(struct uart_port *port, unsigned int idx) >>>>> dev_err(port->dev, >>>>> "failed to find corresponding irq for %s (idx=%d, err=%d)\n", >>>>> mctrl_gpios_desc[i].name, idx, ret); >>>>> - return ERR_PTR(ret); >>>>> + return NULL; >>> >>> Ditto, if a GPIO is used as handshake line and it doesn't have a >>> corresponding irq, polling must be set up instead of simply ignoring the >>> missing irq. >> >> I agree that none of the "return NULL" changes are necessary. >> >> However, note that it's not really possible to recover here by polling; >> some of the gpios may have been claimed and some not when the irq >> claim failure occurs. >> How to know which gpios were claimed? > > I don't understand this. You cannot recover if a GPIO is used by a > different driver/device. If a given GPIO doesn't support triggering an > irq, polling can help, it must of course be implemented correctly. > And if gpio_mctrl implements this (which is the right thing to do, as it > is independant from the actual driver) all necessary knowledge is there. Ah, sorry. I thought the both the code comment and your review comment above about "polling must be set up" implied the caller could do something about that, which obviously the caller can't. 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
