On Mon, Dec 15, 2014 at 4:15 PM, Florian Fainelli <flor...@openwrt.org> wrote: > 2014-12-12 6:33 GMT-08:00 Jonas Gorski <j...@openwrt.org>: >> On Fri, Dec 12, 2014 at 3:17 PM, John Crispin <blo...@openwrt.org> wrote: >>> >>> >>> On 12/12/2014 15:12, Jonas Gorski wrote: >>>> or the gpio-base problem, we should be able to register appropriate >>>> platform data for it as OF_DEV_AUXDATA() in of_platform_populate. >>>> >>>> e.g. >>>> >>>> struct bgpio_pdata gpio0_pdata = { >>>> .base = 0, >>>> }; >>>> >>>> struct bgpio_pdata gpio1_pdata = { >>>> .base = 32, >>>> }; >>> >>> tried it for ralink and lantiq and it got nak'ed by LinusW >>> >>> i keep that part of the code as a small patch inside owrt on top of the >>> stuff i sent upstream. >> >> Wasn't the objection on putting the gpio base into the dts(i) file >> itself? I explicitly try to avoid that here, as gpio-base is something >> linux internal. > > Right that is typically the objection, but with the new gpio > descriptor based API, we should not have that problem anymore, right?
But we would first have to convert the consumers and the gpio driver to use/provide it. I think it's easier to leave the gpio base set for now, and remove it later once we have converted all users to either get their gpios through DT (leds, buttons, ...) or being able to use descriptors (phy reset gpio for ethernet). Jonas _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel