[U-Boot] [PATCH 1/3] arm: atmel: sama5d3: add spi spl boot support

[Andreas Bießmann]

Hi Bo,

On 12/02/2013 04:24 AM, Bo Shen wrote:
> Add SPI SPL boot support for sama5d3xek board.
> 
> Signed-off-by: Bo Shen <voice.shen at atmel.com>
> ---
>  arch/arm/cpu/at91-common/spl.c       |    2 ++
>  arch/arm/include/asm/arch-at91/spl.h |    2 ++
>  board/atmel/sama5d3xek/sama5d3xek.c  |    2 ++
>  include/configs/sama5d3xek.h         |   11 +++++++++++
>  4 files changed, 17 insertions(+)
> 
> diff --git a/arch/arm/cpu/at91-common/spl.c b/arch/arm/cpu/at91-common/spl.c
> index 37c0cc4..2598e09 100644
> --- a/arch/arm/cpu/at91-common/spl.c
> +++ b/arch/arm/cpu/at91-common/spl.c
> @@ -52,6 +52,8 @@ u32 spl_boot_device(void)
>  {
>  #ifdef CONFIG_SYS_USE_MMC
>  	return BOOT_DEVICE_MMC1;
> +#elif CONFIG_SYS_USE_SERIALFLASH
> +	return BOOT_DEVICE_SPI;
>  #endif
>  	return BOOT_DEVICE_NONE;
>  }

I'm not really happy with this solution. We will get a ifdef-nightmare
here for each specific SPL implementation.

Could you please check with your ROM developers if there is some way to
get the boot source? I know Ti guys have implemented some function
parameter to tell that to the next running code in ROM. I can't believe
Atmel has nothing comparable.
The last resort could be to add some proprietary parameter to the SPL
that we could read then and return the correct value in
spl_boot_device() depending on that that stored value ... quite ugly but
I think it is better than clutter this place with a lot of ifdef's.

Another solution could be to move this function to board code.

Any other suggestions?

Best regards

Andreas Bießmann