Re: [PATCH] pwm: atmel-pwm: fix calculation of prescale value

[Thierry Reding]

On Tue, Sep 23, 2014 at 03:30:21PM +0200, Nikolaus Voss wrote:
> The prescale value used for calculating the period was incremented
> afterwards, thus the resulting prescale value is by one too high.
> This resulted in a pwm frequency only half as high as requested.
> 
> This patch moves the 64 bit division out of the prescale loop to
> correct the above issue and make the calculation more efficient.
> 
> Signed-off-by: Nikolaus Voss <n.voss@xxxxxxxxxxxxxxx>
> ---
>  drivers/pwm/pwm-atmel.c | 24 +++++++++++-------------
>  1 file changed, 11 insertions(+), 13 deletions(-)

Hi Nikolaus,

Please Cc the linux-pwm@xxxxxxxxxxxxxxx mailing list for PWM-related
patches in the future.

Also a couple more comments:

In the patch description: "pwm frequency" should be "PWM frequency".

> diff --git a/drivers/pwm/pwm-atmel.c b/drivers/pwm/pwm-atmel.c
> index 6e700a5..ff17b5d 100644
> --- a/drivers/pwm/pwm-atmel.c
> +++ b/drivers/pwm/pwm-atmel.c
> @@ -102,7 +102,7 @@ static int atmel_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
>  			    int duty_ns, int period_ns)
>  {
>  	struct atmel_pwm_chip *atmel_pwm = to_atmel_pwm_chip(chip);
> -	unsigned long clk_rate, prd, dty;
> +	unsigned long prd, dty;
>  	unsigned long long div;
>  	unsigned int pres = 0;
>  	u32 val;
> @@ -113,20 +113,18 @@ static int atmel_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
>  		return -EBUSY;
>  	}
>  
> -	clk_rate = clk_get_rate(atmel_pwm->clk);
> -	div = clk_rate;
> +	/* Calculate the period cycles and prescale value */
> +	div = (unsigned long long)clk_get_rate(atmel_pwm->clk) * period_ns;
> +	do_div(div, (int)1e9);

1e9 should be NSEC_PER_SEC.

> -	/* Calculate the period cycles */
>  	while (div > PWM_MAX_PRD) {
> -		div = clk_rate / (1 << pres);
> -		div = div * period_ns;
> -		/* 1/Hz = 100000000 ns */
> -		do_div(div, 1000000000);
> -
> -		if (pres++ > PRD_MAX_PRES) {
> -			dev_err(chip->dev, "pres exceeds the maximum value\n");
> -			return -EINVAL;
> -		}
> +		div >>= 1;
> +		++pres;

Unless you really need the prefix increment behaviour (you don't in this
case) I prefer using the postfix operator because it is slightly more
idiomatic.

No need for you to respin the patch, I've fixed up the above when
applying.

Thierry

Attachment: pgpKdZjG1MI6r.pgp
Description: PGP signature