Re: [PATCH v2] clk: qcom: clk-rcg2: Add support for duty-cycle for RCG

[Taniya Das <tdas@codeaurora.org>]

Hello Stephen,

Thanks for your review.

On 3/14/2021 4:29 AM, Stephen Boyd wrote:
> Quoting Taniya Das (2021-03-11 04:51:32)
>> The root clock generators with MND divider has the capability to support
>> change in duty-cycle by updating the 'D'. Add the clock ops which would
>> check all the boundary conditions and enable setting the desired duty-cycle
>> as per the consumer.
>>
>> Signed-off-by: Taniya Das <tdas@codeaurora.org>
>> ---
>>   drivers/clk/qcom/clk-rcg2.c | 42 ++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 42 insertions(+)
>>
>> diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
>> index 42f13a2..aac6893 100644
>> --- a/drivers/clk/qcom/clk-rcg2.c
>> +++ b/drivers/clk/qcom/clk-rcg2.c
>> @@ -357,6 +357,46 @@ static int clk_rcg2_set_floor_rate_and_parent(struct clk_hw *hw,
>>          return __clk_rcg2_set_rate(hw, rate, FLOOR);
>>   }
>>
>> +static int clk_rcg2_set_duty_cycle(struct clk_hw *hw, struct clk_duty *duty)
>> +{
>> +       struct clk_rcg2 *rcg = to_clk_rcg2(hw);
>> +       u32 notn_m_val, n_val, m_val, d_val, not2d_val, mask, duty_per;
>> +       int ret;
>> +
>> +       if (!rcg->mnd_width)
>> +               return 0;
> 
> Shouldn't we fail this call if the duty cycle can't be changed? Or have
> another set of clk ops that doesn't support this clk op if the mnd
> isn't present.
> 

Sure, will fail the call for non-MND(HID)RCGs, will be part of my next 
patch.

>> +
>> +       mask = BIT(rcg->mnd_width) - 1;
>> +
>> +       regmap_read(rcg->clkr.regmap, RCG_N_OFFSET(rcg), &notn_m_val);
>> +       regmap_read(rcg->clkr.regmap, RCG_M_OFFSET(rcg), &m_val);
>> +
>> +       n_val = (~(notn_m_val) + m_val) & mask;
>> +
>> +       duty_per = (duty->num * 100) / duty->den;
>> +
>> +       /* Calculate 2d value */
>> +       d_val = DIV_ROUND_CLOSEST(n_val * duty_per * 2, 100);
>> +
>> +        /* Check BIT WIDTHS OF 2d. If D is too big reduce Duty cycle. */
> 
> Why is BIT WIDTHS capitalized? And Duty?
> 

Will take care of it in the next patch.


>> +       if (d_val > mask)
>> +               d_val = mask;
>> +
>> +       if ((d_val / 2) > (n_val - m_val))
>> +               d_val = (n_val - m_val) * 2;
>> +       else if ((d_val / 2) < (m_val / 2))
>> +               d_val = m_val;
>> +
>> +       not2d_val = ~d_val & mask;
>> +
>> +       ret = regmap_update_bits(rcg->clkr.regmap, RCG_D_OFFSET(rcg), mask,
>> +                                not2d_val);
>> +       if (ret)
>> +               return ret;
>> +
>> +       return update_config(rcg);
>> +}
>> +
>>   const struct clk_ops clk_rcg2_ops = {
>>          .is_enabled = clk_rcg2_is_enabled,
>>          .get_parent = clk_rcg2_get_parent,
>> @@ -365,6 +405,7 @@ const struct clk_ops clk_rcg2_ops = {
>>          .determine_rate = clk_rcg2_determine_rate,
>>          .set_rate = clk_rcg2_set_rate,
>>          .set_rate_and_parent = clk_rcg2_set_rate_and_parent,
>> +       .set_duty_cycle = clk_rcg2_set_duty_cycle,
> 
> Can you also implement get_duty_cycle?
> 

Sure, will implement the same.

>>   };
>>   EXPORT_SYMBOL_GPL(clk_rcg2_ops);
>>

-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation.

--