On 12/12/2012 10:13 AM, Florian Fainelli wrote: > Hello Hauke, > > Le 12/12/12 02:10, Hauke Mehrtens a écrit : >> After the new GPIO handling for the bcm47xx target was introduced >> wgt634u.c was not changed. >> This patch fixes the following compile error: >> >> arch/mips/bcm47xx/wgt634u.c: In function ‘gpio_interrupt’: >> arch/mips/bcm47xx/wgt634u.c:119:2: error: implicit declaration of >> function ‘gpio_polarity’ [-Werror=implicit-function-declaration] >> arch/mips/bcm47xx/wgt634u.c: In function ‘wgt634u_init’: >> arch/mips/bcm47xx/wgt634u.c:153:4: error: implicit declaration of >> function ‘gpio_intmask’ [-Werror=implicit-function-declaration] > > Thanks for fixing this. We should probably remove this wgt634u file some > day or the other as it was an ad-hoc hack for this single device while > we actually need a general solution for all BCM47xx boards out there. > Yes I also want to get rid of this file. Currently OpenWrt uses its own detection and gpio/led/button configure code [0], but this doe not look nice to me. I saw Jonas RFC patch for device tree integration into bcm63xx [1] and thought about doing something similar for bcm47xx. Initially device tree would be used for mapping gpios to leds and buttons. I haven't read much about DT, but it looks nice. Is it possible to detect devices by some arbitrary nvram settings like it is done in diag.c? Hauke [0]: https://dev.openwrt.org/browser/trunk/package/broadcom-diag/src/diag.c [1]: http://www.linux-mips.org/archives/linux-mips/2012-11/msg00092.html
