[U-Boot] [PATCH v2 3/4] arm, at91: add Siemens board taurus and axm

[Andreas Bießmann]

Hello Heiko,

On 11/04/2013 10:09 AM, Heiko Schocher wrote:
> Am 04.11.2013 09:43, schrieb Andreas Bießmann:
>> On 11/04/2013 07:40 AM, Heiko Schocher wrote:

<snip>

>>> +#ifdef CONFIG_MACB
>>> +static void taurus_macb_hw_init(void)
>>> +{
>>> +    struct at91_pmc *pmc = (struct at91_pmc *)ATMEL_BASE_PMC;
>>> +    struct at91_rstc *rstc = (struct at91_rstc *)ATMEL_BASE_RSTC;
>>> +    unsigned long erstl;
>>> +
>>> +    /* Enable EMAC clock */
>>> +    writel(1<<  ATMEL_ID_EMAC0,&pmc->pcer);
>>
>> Is it required to enable the MACB clock before PHY reset? It will be
>> done later on in at91_macb_hw_init() in any case.
> 
> Hmm.. this is done in more at91 boards ... Hmm.. seems we should
> look, if we can make this function common? If I see this correct,
> only the pin setup is board specific, or?

That's true.

How could we provide the gpio setup to a common function then?

I think we should go with some at91_phy_reset() which should preferably
run before enabling the (e)macb (would that be true for
CONFIG_RESET_PHY_R?).
I haven't checked it but think it would be Ok to enable the (e)macb
peripheral clock when enabling the peripheral. Therefore the 'Enable
EMAC clock' above could be removed. _All_ at91_macb_hw_init() should
therefore enable the peripheral clock ... unfortunately some do not at
the moment.

<snip>

>>> +/* USB */
>>> +#if defined(CONFIG_BOARD_TAURUS)
>>> +#define CONFIG_USB_ATMEL
>>> +#define CONFIG_USB_OHCI_NEW
>>> +#define CONFIG_SYS_USB_OHCI_CPU_INIT
>>> +#define CONFIG_SYS_USB_OHCI_REGS_BASE        0x00500000
>>> +#define CONFIG_SYS_USB_OHCI_SLOT_NAME        "at91sam9260"
>>> +#define CONFIG_SYS_USB_OHCI_MAX_ROOT_PORTS    2LEGA
>>
>> What does '2LEGA' mean here?
> 
> Hups... This should be:
> 
> #define CONFIG_SYS_USB_OHCI_MAX_ROOT_PORTS    2
> 
> Hmm... I do not find this in my patch file I send with git-sendmail
> 
> Also in the mailarchive:
> http://lists.denx.de/pipermail/u-boot/2013-November/166040.html
> 
> I could not find "2LEGA" ...
> 
> But in Patchwork, there is a "2LEGA" ...
> 
> http://patchwork.ozlabs.org/patch/288117/

I can't find it in the patch ...

> 
> It seems, thats something introduced from you?

Yea, maybe some cursor jump around in MUA while writing 'ATMEL_LEGACY'.
Sorry for the trouble.

Best regards

Andreas Bießmann