Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <1a6d46e4-7ec4-262c-dc3b-fc9c988f979e@quicinc.com>
Date:   Fri, 9 Jun 2023 17:20:18 +0530
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.1
Subject: Re: [PATCH V2 3/6] clk: qcom: clk-alpha-pll: Remove explicit CAL_L
 configuration for EVO PLL
To:     Dmitry Baryshkov <dmitry.baryshkov@linaro.org>,
        Konrad Dybcio <konrad.dybcio@linaro.org>,
        Andy Gross <agross@kernel.org>,
        Michael Turquette <mturquette@baylibre.com>,
        Stephen Boyd <sboyd@kernel.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>
CC:     Bjorn Andersson <andersson@kernel.org>,
        Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>,
        Vinod Koul <vkoul@kernel.org>, <linux-arm-msm@vger.kernel.org>,
        <linux-clk@vger.kernel.org>, <devicetree@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, Taniya Das <quic_tdas@quicinc.com>,
        Satya Priya Kakitapalli <quic_skakitap@quicinc.com>,
        Imran Shaik <quic_imrashai@quicinc.com>,
        Ajit Pandey <quic_ajipan@quicinc.com>
References: <20230525172142.9039-1-quic_jkona@quicinc.com>
 <20230525172142.9039-4-quic_jkona@quicinc.com>
 <6e1d098d-03b9-aa63-a0bf-6cf748a0db0d@linaro.org>
 <7074f718-a3d5-8a03-3830-77a5a0b15500@linaro.org>
 <df7ab6f7-6c5e-9a7d-8d9b-09ff32da34d6@quicinc.com>
 <c60bb4d9-1b53-6c60-8b9d-13069bdff882@linaro.org>
From:   Jagadeesh Kona <quic_jkona@quicinc.com>
In-Reply-To: <c60bb4d9-1b53-6c60-8b9d-13069bdff882@linaro.org>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi Dmitry,

Thanks for your review!

On 6/1/2023 8:13 PM, Dmitry Baryshkov wrote:
> On 01/06/2023 17:33, Jagadeesh Kona wrote:
>> Hi Dmitry, Konrad,
>>
>> On 5/26/2023 9:23 PM, Dmitry Baryshkov wrote:
>>> On 26/05/2023 12:33, Konrad Dybcio wrote:
>>>>
>>>>
>>>> On 25.05.2023 19:21, Jagadeesh Kona wrote:
>>>>> In lucid evo pll, the CAL_L field is part of L value register 
>>>>> itself, and
>>>>> the l value configuration passed from clock controller driver includes
>>>>> CAL_L and L values as well. Hence remove explicit configuration of 
>>>>> CAL_L
>>>>> for evo pll.
>>>>>
>>>>> Fixes: 260e36606a03 ("clk: qcom: clk-alpha-pll: add Lucid EVO PLL 
>>>>> configuration interfaces")
>>>>> Signed-off-by: Taniya Das <quic_tdas@quicinc.com>
>>>>> Signed-off-by: Jagadeesh Kona <quic_jkona@quicinc.com>
>>>>> ---
>>>> Oh that isn't obvious at first sight, nice find!
>>>>
>>>> I'd suggest a different solution though:
>>>>
>>>> #define LUCID_EVO_PLL_L_LVAL    GENMASK(..
>>>> #define LUCID_EVO_PLL_L_CAL_L    GENMASK(..
>>>>
>>>> lval = FIELD_PREP(LUCID_EVO_PLL_L_LVAL, config->l) |
>>>>         FIELD_PREP(LUCID_EVO_PLL_L_CAL_L, config->cal_l);
>>>>
>>>> This would make the separation between the two parts more explicit
>>>>
>>>> however
>>>>
>>>> config->l would then represent the L value and not the end value
>>>> written to the L register
>>>
>>> Yes. I think there should be separate config->l and config->cal_l 
>>> values (and probably ringosc_cal_l, basing on the comment in the 
>>> source).
>>> Thanks for your suggestions. In all recent chipsets, L & CAL_L fields 
>> are encapsulated in the same register, so we feel it is better to 
>> directly pass the combined configuration value in config->l itself and 
>> program it directly into register without any additional handling 
>> required in pll driver code.
> 
> My feeling is that it is better to split it, since these are the 
> different fields. The value .l = 0x4444003e doesn't mean anything per se.
> 
> Three values are much more meaningful:
> .l = 0x3e,
> .cal_l = 0x44,
> .ringosc_cal_l = 0x44,
> 
> Not to mention that this way you don't have to touch pll configuration 
> for the existing Lucid EVO PLL. Not to mention that for the Lucid ole 
> PLLs the cal_l and ringosc_cal_l values seem to be static (0x44), so 
> there is no need to put them to the variable data.
> 

Sure, will keep the existing code as is and will remove this patch in 
the next series.

>>
>> Also the evo pll code is currently reused for both lucid evo and ole 
>> pll's. Lucid ole PLL has an additional RINGOSC_CAL_L field along with 
>> L, CAL_L fields in the same L register. By passing combined 
>> configuration value in config->l itself, we feel we can avoid all the 
>> additional handling required in PLL code.
>>
>>> Just a question: is camcc-sm8550 using the same PLL type or is it 
>>> some kind of subtype of lucid_evo PLL?
>>>
>> No, it is not the same lucid evo PLL. It uses lucid ole PLL.
> 
> Then please don't reuse the clk_lucid_evo_pll_configure() call.
> You can add a new one, which will handle L/CAL_L/RINGOSC_CAL_L differences.
> 

The only difference between evo and ole pll configure is extra 
RINGOSC_CAL_L programming needed only for ole pll. We can achieve the 
same with clk_lucid_evo_pll_configure() itself by directly including 
RINGOSC_CAL_L field in L configuration for OLE PLL's.

Thanks & Regards,
Jagadeesh