[U-Boot] [PATCH v4] at91: Add support for taskit AT91SAM9G20 boards.

[Andreas Bießmann]

Dear Markus Hubig,

On 06.08.2012 18:49, Markus Hubig wrote:
> On Mon, Aug 06, 2012 at 03:01:40PM +0200, Andreas Bießmann wrote:
>> On 06.08.2012 11:11, Markus Hubig wrote:
>>>
>>> +static unsigned int saved_state[3] = {STATUS_LED_OFF,
>>> +	STATUS_LED_OFF, STATUS_LED_OFF};
>>> +
>>> +void coloured_LED_init(void)
>>> +{
>>> +	/* Clock is enabled in board_early_init_f() */
>>
>> But coloured_LED_init() is called way before!
>>
>> Beware, this coloured LED stuff is sometimes used for very early debug
>> purposes. To get this working you will need to switch the clocks here.
>>
>> If you not insist on this colour LED framework I would appreciate to not
>> add this coloured_LED_init for the stamp9g20 but use the status LED
>> framework -> README.LED
> 
> Are you shure about this?

No, I've not worked with that status LED thing. Contrastingly I thought
that LED stuff is at91 specific. But there are some other ARM SoC using
this colored LED and status LED thing. So this appears to be a wrong
decision.

> Is I just remove this and put the 
> 
>>> +	at91_set_gpio_output(CONFIG_RED_LED, 1);
>>> +	at91_set_gpio_output(CONFIG_GREEN_LED, 1);
>>> +	at91_set_gpio_output(CONFIG_YELLOW_LED, 1);
> 
> in __led_init() I have no LED's at all and the commandlie tool 'led'
> stopps working... At all the Status LED stuff is a bit confusing ...
> 
> - It's not clear to me if the STATUS_LED_BIT should be e.g. 1 or
>   e.g. AT91_PIN_PC5.
> 
> - It's not clear what the benefit of using this is.
> 
> - It's not clear if the coloured_LED stuff is part of it or something
>   extra.

Sorry, I can not help you here.

> Maybe it's better is I just switch on the green LED in board_init?

but you will loose the led cmd which may be useful.

> For now I just switched the clock on in coloured_LED_init. At least I
> have some LED now.

Thats ok with me.

Best regards

Andreas Bießmann