[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <b918cfa9-1f40-4aee-ad91-d6425798bd07@oss.qualcomm.com>
Date: Fri, 30 Jan 2026 22:24:06 +0530
From: Praveen Talari <praveen.talari@....qualcomm.com>
To: Konrad Dybcio <konrad.dybcio@....qualcomm.com>,
Andi Shyti <andi.shyti@...nel.org>, Rob Herring <robh@...nel.org>,
Krzysztof Kozlowski <krzk+dt@...nel.org>,
Conor Dooley
<conor+dt@...nel.org>,
Mukesh Kumar Savaliya <mukesh.savaliya@....qualcomm.com>,
Viken Dadhaniya <viken.dadhaniya@....qualcomm.com>,
Bjorn Andersson <andersson@...nel.org>,
Konrad Dybcio <konradybcio@...nel.org>, linux-arm-msm@...r.kernel.org,
linux-i2c@...r.kernel.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org, bryan.odonoghue@...aro.org,
dmitry.baryshkov@....qualcomm.com, bjorn.andersson@....qualcomm.com
Cc: prasad.sodagudi@....qualcomm.com, quic_vtanuku@...cinc.com,
aniket.randive@....qualcomm.com, chandana.chiluveru@....qualcomm.com
Subject: Re: [PATCH v3 06/12] soc: qcom: geni-se: Introduce helper APIs for
performance control
Hi Konrad
On 1/30/2026 5:53 PM, Konrad Dybcio wrote:
> On 1/12/26 11:47 AM, Praveen Talari wrote:
>> The GENI Serial Engine (SE) drivers (I2C, SPI, and SERIAL) currently
>> manage performance levels and operating points directly. This resulting
>> in code duplication across drivers. such as configuring a specific level
>> or find and apply an OPP based on a clock frequency.
>>
>> Introduce two new helper APIs, geni_se_set_perf_level() and
>> geni_se_set_perf_opp(), addresses this issue by providing a streamlined
>> method for the GENI Serial Engine (SE) drivers to find and set the OPP
>> based on the desired performance level, thereby eliminating redundancy.
>>
>> Signed-off-by: Praveen Talari <praveen.talari@....qualcomm.com>
>> ---
>
> [...]
>
>> +/**
>> + * geni_se_set_perf_level() - Set performance level for GENI SE.
>> + * @se: Pointer to the struct geni_se instance.
>> + * @level: The desired performance level.
>> + *
>> + * Sets the performance level by directly calling dev_pm_opp_set_level
>> + * on the performance device associated with the SE.
>> + *
>> + * Return: 0 on success, or a negative error code on failure.
>> + */
>> +int geni_se_set_perf_level(struct geni_se *se, unsigned long level)
>> +{
>> + return dev_pm_opp_set_level(se->pd_list->pd_devs[DOMAIN_IDX_PERF], level);
>> +}
>> +EXPORT_SYMBOL_GPL(geni_se_set_perf_level);
>
> This function is never used
it will be used by UART driver, not for I2C/SPI.
>
>> +
>> +/**
>> + * geni_se_set_perf_opp() - Set performance OPP for GENI SE by frequency.
>> + * @se: Pointer to the struct geni_se instance.
>> + * @clk_freq: The requested clock frequency.
>> + *
>> + * Finds the nearest operating performance point (OPP) for the given
>> + * clock frequency and applies it to the SE's performance device.
>> + *
>> + * Return: 0 on success, or a negative error code on failure.
>> + */
>> +int geni_se_set_perf_opp(struct geni_se *se, unsigned long clk_freq)
>
> I think with the SPI driver in mind (which seems to do a simple rateset
APIs were added as generic interfaces shared across I²C/SPI which is
specific to firmware control, not Linux control.
> for both backends) we could do:
>
>> +{
>> + struct device *perf_dev = se->pd_list->pd_devs[DOMAIN_IDX_PERF];
>
> Then, we can do struct device * perf_dev = se->dev;
I don't think, it is needed since this is specific to firmware control,
not Linux control.
Thanks,
Praveen Talari
>
> if (se->pd_list && se->pd_list->pd_devs[DOMAIN_IDX_PERF])
> perf_dev = se->pd_list->pd_devs[DOMAIN_IDX_PERF];
>
> and reuse it in both cases, completely transparently to the caller
>
> Konrad
Powered by blists - more mailing lists