[v3,1/8] clk: sunxi-ng: nkm: consider alternative parent rates when determining rate

  In case the CLK_SET_RATE_PARENT flag is set, consider using a different
parent rate when determining a new rate.

To find the best match for the requested rate, perform the following
steps for each NKM combination:
 - calculate the optimal parent rate,
 - find the best parent rate that the parent clock actually supports
 - use that parent rate to calculate the effective rate.

In case the clk does not support setting the parent rate, use the same
algorithm as before.

Signed-off-by: Frank Oltmanns <frank@oltmanns.dev>
---
 drivers/clk/sunxi-ng/ccu_nkm.c | 48 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 45 insertions(+), 3 deletions(-)
  

Hi,

On Sun, Jul 02, 2023 at 07:55:20PM +0200, Frank Oltmanns wrote:
> In case the CLK_SET_RATE_PARENT flag is set, consider using a different
> parent rate when determining a new rate.
> 
> To find the best match for the requested rate, perform the following
> steps for each NKM combination:
>  - calculate the optimal parent rate,
>  - find the best parent rate that the parent clock actually supports
>  - use that parent rate to calculate the effective rate.
> 
> In case the clk does not support setting the parent rate, use the same
> algorithm as before.
> 
> Signed-off-by: Frank Oltmanns <frank@oltmanns.dev>
> ---
>  drivers/clk/sunxi-ng/ccu_nkm.c | 48 +++++++++++++++++++++++++++++++++++++++---
>  1 file changed, 45 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/clk/sunxi-ng/ccu_nkm.c b/drivers/clk/sunxi-ng/ccu_nkm.c
> index a0978a50edae..d83843e69c25 100644
> --- a/drivers/clk/sunxi-ng/ccu_nkm.c
> +++ b/drivers/clk/sunxi-ng/ccu_nkm.c
> @@ -6,6 +6,7 @@
>  
>  #include <linux/clk-provider.h>
>  #include <linux/io.h>
> +#include <linux/math.h>
>  
>  #include "ccu_gate.h"
>  #include "ccu_nkm.h"
> @@ -16,6 +17,44 @@ struct _ccu_nkm {
>  	unsigned long	m, min_m, max_m;
>  };
>
> +static unsigned long ccu_nkm_find_best_with_parent_adj(unsigned long *parent, unsigned long rate,
> +						       struct _ccu_nkm *nkm, struct clk_hw *phw)

The usual order in that driver (and Linux in general) would make the
clk_hw and nkm structure pointers first, and then the parent rate and
rate.

But something looks off to me: ccu_nkm_find_best_with_parent_adj takes a
pointer to the parent rate which makes sense since we're going to modify
it.

> +{
> +	unsigned long best_rate = 0, best_parent_rate = *parent, tmp_parent = *parent;
> +	unsigned long best_n = 0, best_k = 0, best_m = 0;
> +	unsigned long _n, _k, _m;
> +
> +	for (_k = nkm->min_k; _k <= nkm->max_k; _k++) {
> +		for (_n = nkm->min_n; _n <= nkm->max_n; _n++) {
> +			for (_m = nkm->min_m; _m <= nkm->max_m; _m++) {
> +				unsigned long tmp_rate;
> +
> +				tmp_parent = clk_hw_round_rate(phw, rate * _m / (_n * _k));
> +
> +				tmp_rate = tmp_parent * _n * _k / _m;
> +				if (tmp_rate > rate)
> +					continue;
> +
> +				if ((rate - tmp_rate) < (rate - best_rate)) {
> +					best_rate = tmp_rate;
> +					best_parent_rate = tmp_parent;
> +					best_n = _n;
> +					best_k = _k;
> +					best_m = _m;
> +				}
> +			}
> +		}
> +	}
> +
> +	nkm->n = best_n;
> +	nkm->k = best_k;
> +	nkm->m = best_m;
> +
> +	*parent = best_parent_rate;
> +
> +	return best_rate;
> +}
> +
>  static unsigned long ccu_nkm_find_best(unsigned long parent, unsigned long rate,
>  				       struct _ccu_nkm *nkm)

You haven't modified ccu_nkm_find_best though, and it still takes the
parent rate value.

>  {
> @@ -106,7 +145,7 @@ static unsigned long ccu_nkm_recalc_rate(struct clk_hw *hw,
>  }
>  
>  static unsigned long ccu_nkm_round_rate(struct ccu_mux_internal *mux,
> -					struct clk_hw *hw,
> +					struct clk_hw *parent_hw,

(This should be another patch)

>  					unsigned long *parent_rate,
>  					unsigned long rate,
>  					void *data)
> @@ -124,7 +163,10 @@ static unsigned long ccu_nkm_round_rate(struct ccu_mux_internal *mux,
>  	if (nkm->common.features & CCU_FEATURE_FIXED_POSTDIV)
>  		rate *= nkm->fixed_post_div;
>  
> -	rate = ccu_nkm_find_best(*parent_rate, rate, &_nkm);

parent_rate is a pointer, we were dereferencing it to pass its value to
ccu_nkm_find_best. All good so far.

> +	if (!clk_hw_can_set_rate_parent(&nkm->common.hw))
> +		rate = ccu_nkm_find_best(*parent_rate, rate, &_nkm);

Still passing by value

> +	else
> +		rate = ccu_nkm_find_best_with_parent_adj(parent_rate, rate, &_nkm, parent_hw);

And passing the pointer there since it takes a pointer. Still good.

>  
>  	if (nkm->common.features & CCU_FEATURE_FIXED_POSTDIV)
>  		rate /= nkm->fixed_post_div;
> @@ -159,7 +201,7 @@ static int ccu_nkm_set_rate(struct clk_hw *hw, unsigned long rate,
>  	_nkm.min_m = 1;
>  	_nkm.max_m = nkm->m.max ?: 1 << nkm->m.width;
>  
> -	ccu_nkm_find_best(parent_rate, rate, &_nkm);
> +	ccu_nkm_find_best(&parent_rate, rate, &_nkm);

But here, we're passing a pointer to parent_rate to ccu_nkm_find_best,
while it's still supposed to take it by value?

Maxime
  

On 2023-07-03 at 08:47:43 +0200, Maxime Ripard <maxime@cerno.tech> wrote:
> [[PGP Signed Part:Undecided]]
> Hi,
>
> On Sun, Jul 02, 2023 at 07:55:20PM +0200, Frank Oltmanns wrote:
>> In case the CLK_SET_RATE_PARENT flag is set, consider using a different
>> parent rate when determining a new rate.
>>
>> To find the best match for the requested rate, perform the following
>> steps for each NKM combination:
>>  - calculate the optimal parent rate,
>>  - find the best parent rate that the parent clock actually supports
>>  - use that parent rate to calculate the effective rate.
>>
>> In case the clk does not support setting the parent rate, use the same
>> algorithm as before.
>>
>> Signed-off-by: Frank Oltmanns <frank@oltmanns.dev>
>> ---
>>  drivers/clk/sunxi-ng/ccu_nkm.c | 48 +++++++++++++++++++++++++++++++++++++++---
>>  1 file changed, 45 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/clk/sunxi-ng/ccu_nkm.c b/drivers/clk/sunxi-ng/ccu_nkm.c
>> index a0978a50edae..d83843e69c25 100644
>> --- a/drivers/clk/sunxi-ng/ccu_nkm.c
>> +++ b/drivers/clk/sunxi-ng/ccu_nkm.c
>> @@ -6,6 +6,7 @@
>>
>>  #include <linux/clk-provider.h>
>>  #include <linux/io.h>
>> +#include <linux/math.h>
>>
>>  #include "ccu_gate.h"
>>  #include "ccu_nkm.h"
>> @@ -16,6 +17,44 @@ struct _ccu_nkm {
>>  	unsigned long	m, min_m, max_m;
>>  };
>>
>> +static unsigned long ccu_nkm_find_best_with_parent_adj(unsigned long *parent, unsigned long rate,
>> +						       struct _ccu_nkm *nkm, struct clk_hw *phw)
>
> The usual order in that driver (and Linux in general) would make the
> clk_hw and nkm structure pointers first, and then the parent rate and
> rate.

I'll address that in v4.

>
> But something looks off to me: ccu_nkm_find_best_with_parent_adj takes a
> pointer to the parent rate which makes sense since we're going to modify
> it.
>
>> +{
>> +	unsigned long best_rate = 0, best_parent_rate = *parent, tmp_parent = *parent;
>> +	unsigned long best_n = 0, best_k = 0, best_m = 0;
>> +	unsigned long _n, _k, _m;
>> +
>> +	for (_k = nkm->min_k; _k <= nkm->max_k; _k++) {
>> +		for (_n = nkm->min_n; _n <= nkm->max_n; _n++) {
>> +			for (_m = nkm->min_m; _m <= nkm->max_m; _m++) {
>> +				unsigned long tmp_rate;
>> +
>> +				tmp_parent = clk_hw_round_rate(phw, rate * _m / (_n * _k));
>> +
>> +				tmp_rate = tmp_parent * _n * _k / _m;
>> +				if (tmp_rate > rate)
>> +					continue;
>> +
>> +				if ((rate - tmp_rate) < (rate - best_rate)) {
>> +					best_rate = tmp_rate;
>> +					best_parent_rate = tmp_parent;
>> +					best_n = _n;
>> +					best_k = _k;
>> +					best_m = _m;
>> +				}
>> +			}
>> +		}
>> +	}
>> +
>> +	nkm->n = best_n;
>> +	nkm->k = best_k;
>> +	nkm->m = best_m;
>> +
>> +	*parent = best_parent_rate;
>> +
>> +	return best_rate;
>> +}
>> +
>>  static unsigned long ccu_nkm_find_best(unsigned long parent, unsigned long rate,
>>  				       struct _ccu_nkm *nkm)
>
> You haven't modified ccu_nkm_find_best though, and it still takes the
> parent rate value.
>
>>  {
>> @@ -106,7 +145,7 @@ static unsigned long ccu_nkm_recalc_rate(struct clk_hw *hw,
>>  }
>>
>>  static unsigned long ccu_nkm_round_rate(struct ccu_mux_internal *mux,
>> -					struct clk_hw *hw,
>> +					struct clk_hw *parent_hw,
>
> (This should be another patch)

Ok, will do in v4.

>
>>  					unsigned long *parent_rate,
>>  					unsigned long rate,
>>  					void *data)
>> @@ -124,7 +163,10 @@ static unsigned long ccu_nkm_round_rate(struct ccu_mux_internal *mux,
>>  	if (nkm->common.features & CCU_FEATURE_FIXED_POSTDIV)
>>  		rate *= nkm->fixed_post_div;
>>
>> -	rate = ccu_nkm_find_best(*parent_rate, rate, &_nkm);
>
> parent_rate is a pointer, we were dereferencing it to pass its value to
> ccu_nkm_find_best. All good so far.
>
>> +	if (!clk_hw_can_set_rate_parent(&nkm->common.hw))
>> +		rate = ccu_nkm_find_best(*parent_rate, rate, &_nkm);
>
> Still passing by value
>
>> +	else
>> +		rate = ccu_nkm_find_best_with_parent_adj(parent_rate, rate, &_nkm, parent_hw);
>
> And passing the pointer there since it takes a pointer. Still good.
>
>>
>>  	if (nkm->common.features & CCU_FEATURE_FIXED_POSTDIV)
>>  		rate /= nkm->fixed_post_div;
>> @@ -159,7 +201,7 @@ static int ccu_nkm_set_rate(struct clk_hw *hw, unsigned long rate,
>>  	_nkm.min_m = 1;
>>  	_nkm.max_m = nkm->m.max ?: 1 << nkm->m.width;
>>
>> -	ccu_nkm_find_best(parent_rate, rate, &_nkm);
>> +	ccu_nkm_find_best(&parent_rate, rate, &_nkm);
>
> But here, we're passing a pointer to parent_rate to ccu_nkm_find_best,
> while it's still supposed to take it by value?

Ugh. Yeah, sorry. I had that error in V2 but squashed the correction
into patch 5 instead of patch 1. I'll fix that in v4.

Thanks,
  Frank

>
> Maxime
>
> [[End of PGP Signed Part]]