Yes, I can set GPIOs flashing the mentioned official image to the board, however I need custom image as my I2S HW is not supported in any precompiled images.
The problem is that compiling the LEDE repo by default set GPIOs 14-17 to MDI_xxx (act as Ethernet ports) instead of GPIOs.
I can set those pins to GPIOs with devmem like stuffs at or after boot but that's far from elegant.
Regarding gpioctl, I did not download it directly, but installed by ./scripts/feeds install gpioctl-sysfs (from the top level directory of the installed SDK). However, it is available as part of libugpio. But it is just an interface to access /sys/class/gpio/*, it simply reads / writes those files.
This is what I have changed in Makefile (for both: libnewgpio and gpioirq)
then just compile with make (libnewgpio and gpioirq consequently)
@luz Thats the thing. Even if I use different GPIOs, I never can create a different bus/deviceID interface. I always get an error upon spi registration: spi_gpio_custom fails. I have that package properly installed, though.
On the other hand, if you are not using our Docker Image, you will have to run make menuconfig in the build system to select libugpio package to be compiled and then compile it by running make.
In General you will need to have the external library, for instance libugpio, to be compiled by the Build system before it can be dynamically linked to code that you are trying to cross compile. Above all, I strongly recommend basing your BUild system on our GitHub source repo since it is preconfigured to include the Onion software repo and to build all the libraries that are included on the Omega.