Re: [RFC PATCH] AT91: manage clock by functionality instead of CPUs

[Nicolas Ferre <nicolas.ferre@atmel.com>]

Hi Ville,

Ville Syrjälä :
> On Fri, Mar 20, 2009 at 11:56:55AM +0100, Nicolas Ferre wrote:
>> --- a/arch/arm/mach-at91/clock.c
>> +++ b/arch/arm/mach-at91/clock.c
>> @@ -45,6 +45,25 @@
>>  #define clk_is_sys(x)		((x)->type & CLK_TYPE_SYSTEM)
>>  
>>  
>> +/*
>> + * Chips have some kind of clocks : group them by functionality
>> + */
>> +#define cpu_has_utmi()		(  cpu_is_at91cap9() \
>> +				|| cpu_is_at91sam9rl())
>> +
>> +#define cpu_has_800M_plla()	(cpu_is_at91sam9g20())
>> +
>> +#define cpu_has_pllb()		(!(cpu_is_at91sam9rl()))
> 
> Useless parentheses.

Ok.

>> +
>> +#define cpu_has_upll()		(0)
> 
> Is this for some future chip?

Definitely, yes !

>> +
>> +/* USB host HS & FS */
>> +#define cpu_has_uhp()		(!cpu_is_at91sam9rl())
>> +
>> +/* USB device FS only */
>> +#define cpu_has_udpfs()		(!(cpu_is_at91sam9rl()))
> 
> Useless parentheses.

Ok.


>> @@ -175,7 +194,12 @@ static struct clk __init *at91_css_to_cl
>>  		case AT91_PMC_CSS_PLLA:
>>  			return &plla;
>>  		case AT91_PMC_CSS_PLLB:
>> -			return &pllb;
>> +			if (cpu_has_upll())
>> +				/* CSS_PLLB == CSS_UPLL */
>> +				return &utmi_clk;
>> +			else
>> +				if (cpu_has_pllb())
>> +					return &pllb;
> 
> Put "else if" on one line?

Ok.


Thanks for your review, cheers,
-- 
Nicolas Ferre