[En-Nut-Discussion] Where to define PIN mapping

[Harald Kipp]

On 19.02.2013 14:09, Uwe Bonnes wrote:

>     Harald> I'm not sure if I got you. Even early AT91 devices allow to have
>     Harald> certain device at alternate pins, this is nothing new.
> 
> So probably arch/arm/dev/atmel/i2cbus_at91.c was a bad example I looked at?
> The datasheet has no alternate pins for I2C, at least I see no such pins,
> and I also see no way to choose the pins.

Well, OK. I do not know any member of the AT91 family with alternate
I2C/UART/etc. pins. But almost all AT91 devices have configurable SPI
_hardware: chip selects. Quite often clock and PWM outputs are provided
at alternative pins.


>     Harald> The board.conf files are responsible to define the pins actually
>     Harald> used on a specific board layout.
> 
> Probably we don't have a common understanding. I mean pins like the UART TX
> pin. For the STM32 you may use up to at least 3 different pins for this
> function . The usart driver needs to know which pin to connect internally to
> the uart device. So if board.h would define the usage, we would again need
> board.h at library compile time. I am stuck.

To better understand the problem: Why can't this be defined in the
board.conf file?

Regards,

Harald