[PATCH] clk: at91: modify PMC peripheral clock to deal with newer register layout

[nicolas.ferre@atmel.com (Nicolas Ferre)]

Le 10/06/2015 16:55, Boris Brezillon a ?crit :
> On Wed, 10 Jun 2015 16:39:40 +0200
> Nicolas Ferre <nicolas.ferre@atmel.com> wrote:
> 
>> Le 10/06/2015 15:55, Boris Brezillon a ?crit :
>>> Hi Nicolas,
>>>
>>> On Wed, 10 Jun 2015 15:42:44 +0200
>>> Nicolas Ferre <nicolas.ferre@atmel.com> wrote:
>>>
>>>> As some more information is added to the PCR register, we'd better use
>>>> a copy of its content and modify just the peripheral-related bits.
>>>> Implement a read-modify-write for the enable() and disable() callbacks.
>>>>
>>>> Header file is also modified to have the PCR_DIV mask.
>>>>
>>>> Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
>>>
>>> Apart from the below comment you can add my:
>>>
>>> Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
>>>
>>>> ---
>>>>  drivers/clk/at91/clk-peripheral.c | 19 +++++++++++++------
>>>>  include/linux/clk/at91_pmc.h      |  3 ++-
>>>>  2 files changed, 15 insertions(+), 7 deletions(-)
>>>>
>>>> diff --git a/drivers/clk/at91/clk-peripheral.c b/drivers/clk/at91/clk-peripheral.c
>>>> index 597fed423d7d..37e2fea14890 100644
>>>> --- a/drivers/clk/at91/clk-peripheral.c
>>>> +++ b/drivers/clk/at91/clk-peripheral.c
>>>> @@ -161,14 +161,17 @@ static int clk_sam9x5_peripheral_enable(struct clk_hw *hw)
>>>>  {
>>>>  	struct clk_sam9x5_peripheral *periph = to_clk_sam9x5_peripheral(hw);
>>>>  	struct at91_pmc *pmc = periph->pmc;
>>>> +	u32 tmp;
>>>>  
>>>>  	if (periph->id < PERIPHERAL_ID_MIN)
>>>>  		return 0;
>>>>  
>>>> -	pmc_write(pmc, AT91_PMC_PCR, (periph->id & AT91_PMC_PCR_PID) |
>>>> -				     AT91_PMC_PCR_CMD |
>>>> -				     AT91_PMC_PCR_DIV(periph->div) |
>>>> -				     AT91_PMC_PCR_EN);
>>>> +	pmc_lock(pmc);
>>>> +	pmc_write(pmc, AT91_PMC_PCR, (periph->id & AT91_PMC_PCR_PID));
>>>> +	tmp = pmc_read(pmc, AT91_PMC_PCR) & ~AT91_PMC_PCR_P_DIV;
>>>> +	pmc_write(pmc, AT91_PMC_PCR, tmp | AT91_PMC_PCR_PDIV(periph->div)
>>>> +					 | AT91_PMC_PCR_EN);
>>>> +	pmc_unlock(pmc);
>>>>  	return 0;
>>>>  }
>>>>  
>>>> @@ -176,12 +179,16 @@ static void clk_sam9x5_peripheral_disable(struct clk_hw *hw)
>>>>  {
>>>>  	struct clk_sam9x5_peripheral *periph = to_clk_sam9x5_peripheral(hw);
>>>>  	struct at91_pmc *pmc = periph->pmc;
>>>> +	u32 tmp;
>>>>  
>>>>  	if (periph->id < PERIPHERAL_ID_MIN)
>>>>  		return;
>>>>  
>>>> -	pmc_write(pmc, AT91_PMC_PCR, (periph->id & AT91_PMC_PCR_PID) |
>>>> -				     AT91_PMC_PCR_CMD);
>>>> +	pmc_lock(pmc);
>>>> +	pmc_write(pmc, AT91_PMC_PCR, (periph->id & AT91_PMC_PCR_PID));
>>>> +	tmp = pmc_read(pmc, AT91_PMC_PCR) & ~AT91_PMC_PCR_EN;
>>>> +	pmc_write(pmc, AT91_PMC_PCR, tmp);
>>>> +	pmc_unlock(pmc);
>>>>  }
>>>>  
>>>>  static int clk_sam9x5_peripheral_is_enabled(struct clk_hw *hw)
>>>> diff --git a/include/linux/clk/at91_pmc.h b/include/linux/clk/at91_pmc.h
>>>> index 7669f7618f39..4685c3d62f94 100644
>>>> --- a/include/linux/clk/at91_pmc.h
>>>> +++ b/include/linux/clk/at91_pmc.h
>>>> @@ -184,7 +184,8 @@ extern void __iomem *at91_pmc_base;
>>>>  #define AT91_PMC_PCR		0x10c			/* Peripheral Control Register [some SAM9 and SAMA5] */
>>>>  #define		AT91_PMC_PCR_PID	(0x3f  <<  0)		/* Peripheral ID */
>>>>  #define		AT91_PMC_PCR_CMD	(0x1  <<  12)		/* Command (read=0, write=1) */
>>>> -#define		AT91_PMC_PCR_DIV(n)	((n)  <<  16)		/* Divisor Value */
>>>> +#define		AT91_PMC_PCR_P_DIV	(0x3  <<  16)		/* Divisor mask */
>>>
>>> How about renaming this macro into AT91_PMC_PCR_PDIV_MSK ?
>>> I know the macro names in this file are not consistent, but maybe it's
>>> time to choose appropriate names for new AT91_PMC macros.
>>
>> Well, this is what I tried to find: consistency ;-)
>> It seems that other macros are like I did for this one: the pure name of
>> the field for the mask and some kind of other form of the name for a
>> value macro or a (usually useless) list of macro-per-value things.
>>
>> For this one I added a "P" for peripheral which is not in the real name
>> of the register field. This is to differentiate it from the upcoming
>> GCK_DIV field...
> 
> Yes, but that doesn't help in describing what the macros are really
> representing. My point is that, yes moving to _MSK doesn't add any
> consistency, but there already is no consistency in the existing names,
> so, IMHO, we should at least choose representative names.
> I'm not arguing against the addition of a P before the DIV word, but
> why is P_DIV chosen to reprensent the mask and PDIV used to define the
> macro building the value to be stored in the PCR register ?

Ok, I'll try another attempt with cleaner field descriptions for this
register.

Bye,
-- 
Nicolas Ferre