Re: [PATCH v2 1/2] phy/qcom: add regulator_set_load to edp/dp phy

[Kuogee Hsieh <quic_khsieh@quicinc.com>]

On 5/18/2022 10:52 AM, Dmitry Baryshkov wrote:
> On 18/05/2022 20:36, Kuogee Hsieh wrote:
>>
>> On 5/18/2022 10:12 AM, Dmitry Baryshkov wrote:
>>> On Wed, 18 May 2022 at 19:43, Kuogee Hsieh <quic_khsieh@quicinc.com> 
>>> wrote:
>>>> This patch add regulator_set_load() to both eDP and DP phy driver
>>>> to have totally control regulators.
>>>>
>>>> Changes in v2:
>>>> -- no regulator_set_laod() before disable regulator
>>>>
>>>> Signed-off-by: Kuogee Hsieh <quic_khsieh@quicinc.com>
>>>> ---
>>>>   drivers/phy/qualcomm/phy-qcom-edp.c | 25 +++++++++++++++++++++----
>>>>   drivers/phy/qualcomm/phy-qcom-qmp.c | 13 +++++++++++++
>>> Split into -edp and -qmp part.
>>>
>>>>   2 files changed, 34 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/drivers/phy/qualcomm/phy-qcom-edp.c 
>>>> b/drivers/phy/qualcomm/phy-qcom-edp.c
>>>> index cacd32f..9b55095 100644
>>>> --- a/drivers/phy/qualcomm/phy-qcom-edp.c
>>>> +++ b/drivers/phy/qualcomm/phy-qcom-edp.c
>>>> @@ -87,17 +87,24 @@ struct qcom_edp {
>>>>
>>>>          struct clk_bulk_data clks[2];
>>>>          struct regulator_bulk_data supplies[2];
>>>> +       int enable_load[2];
>>>> +       int disable_load[2];
>>> As noticed in the review of the previous patch, disable_load is 
>>> unnecessary.
>>>
>>>>   };
>>>>
>>>>   static int qcom_edp_phy_init(struct phy *phy)
>>>>   {
>>>>          struct qcom_edp *edp = phy_get_drvdata(phy);
>>>>          int ret;
>>>> +       int num_consumers = ARRAY_SIZE(edp->supplies);
>>>> +       int i;
>>>>
>>>> -       ret = regulator_bulk_enable(ARRAY_SIZE(edp->supplies), 
>>>> edp->supplies);
>>>> +       ret = regulator_bulk_enable(num_consumers, edp->supplies);
>>>>          if (ret)
>>>>                  return ret;
>>>>
>>>> +       for (i = num_consumers - 1; i >= 0; --i)
>>>> + regulator_set_load(edp->supplies[i].consumer, edp->enable_load[i]);
>>>> +
>>>>          ret = clk_bulk_prepare_enable(ARRAY_SIZE(edp->clks), 
>>>> edp->clks);
>>>>          if (ret)
>>>>                  goto out_disable_supplies;
>>>> @@ -425,9 +432,15 @@ static int qcom_edp_phy_power_off(struct phy 
>>>> *phy)
>>>>   static int qcom_edp_phy_exit(struct phy *phy)
>>>>   {
>>>>          struct qcom_edp *edp = phy_get_drvdata(phy);
>>>> +       int num_consumers = ARRAY_SIZE(edp->supplies);
>>>> +       int i;
>>>>
>>>> clk_bulk_disable_unprepare(ARRAY_SIZE(edp->clks), edp->clks);
>>>> -       regulator_bulk_disable(ARRAY_SIZE(edp->supplies), 
>>>> edp->supplies);
>>>> +
>>>> +       for (i = num_consumers - 1; i >= 0; --i)
>>>> + regulator_set_load(edp->supplies[i].consumer, edp->disable_load[i]);
>>>> +
>>>> +       regulator_bulk_disable(num_consumers, edp->supplies);
>>>>
>>>>          return 0;
>>>>   }
>>>> @@ -633,8 +646,12 @@ static int qcom_edp_phy_probe(struct 
>>>> platform_device *pdev)
>>>>          if (ret)
>>>>                  return ret;
>>>>
>>>> -       edp->supplies[0].supply = "vdda-phy";
>>>> -       edp->supplies[1].supply = "vdda-pll";
>>>> +       edp->supplies[0].supply = "vdda-1p2";
>>>> +       edp->supplies[1].supply = "vdda-0p9";
>>> NAK, You can not randomly change supply names.
>>
>> if you do no change here, then we have to change dtsi.
>>
>> They are not match.
>
> Where is no match? I don't see any in-kernel dtsi using them.

my mistake, we did not pull in Doug's patch at our internal release 
where i run my test.


>
>
>>>> +       edp->enable_load[0] = 21800;    /* 1.2 V */
>>>> +       edp->enable_load[1] = 36000;    /* 1.2 V */
>>>> +       edp->disable_load[0] = 4;       /* 0.9 V */
>>>> +       edp->disable_load[1] = 4;       /* 10.9V */
>>> Again, 10.9V here. Kuogee. Have you read the review points?
>> I have read it. but forget to make  change at edp file.
>>>
>>>>          ret = devm_regulator_bulk_get(dev, 
>>>> ARRAY_SIZE(edp->supplies), edp->supplies);
>>>>          if (ret)
>>>>                  return ret;
>>>> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp.c 
>>>> b/drivers/phy/qualcomm/phy-qcom-qmp.c
>>>> index b144ae1..0a4c8a8 100644
>>>> --- a/drivers/phy/qualcomm/phy-qcom-qmp.c
>>>> +++ b/drivers/phy/qualcomm/phy-qcom-qmp.c
>>>> @@ -3130,6 +3130,7 @@ struct qmp_phy_cfg {
>>>>          int num_resets;
>>>>          /* regulators to be requested */
>>>>          const char * const *vreg_list;
>>>> +       const unsigned int *vreg_enable_load;
>>>>          int num_vregs;
>>>>
>>>>          /* array of registers with different offsets */
>>>> @@ -3346,6 +3347,10 @@ static const char * const qmp_phy_vreg_l[] = {
>>>>          "vdda-phy", "vdda-pll",
>>>>   };
>>>>
>>>> +static const unsigned int qmp_phy_vreg_enable_load[] = {
>>>> +       21800, 36000
>>>> +};
>>>> +
>>>>   static const struct qmp_phy_cfg ipq8074_usb3phy_cfg = {
>>>>          .type                   = PHY_TYPE_USB3,
>>>>          .nlanes                 = 1,
>>>> @@ -4072,6 +4077,7 @@ static const struct qmp_phy_cfg 
>>>> sm8250_usb3phy_cfg = {
>>>>          .reset_list             = msm8996_usb3phy_reset_l,
>>>>          .num_resets             = 
>>>> ARRAY_SIZE(msm8996_usb3phy_reset_l),
>>>>          .vreg_list              = qmp_phy_vreg_l,
>>>> +       .vreg_enable_load       = qmp_phy_vreg_enable_load,
>>>>          .num_vregs              = ARRAY_SIZE(qmp_phy_vreg_l),
>>>>          .regs                   = qmp_v4_usb3phy_regs_layout,
>>>>
>>>> @@ -4139,6 +4145,7 @@ static const struct qmp_phy_cfg 
>>>> sm8250_dpphy_cfg = {
>>>>          .reset_list             = msm8996_usb3phy_reset_l,
>>>>          .num_resets             = 
>>>> ARRAY_SIZE(msm8996_usb3phy_reset_l),
>>>>          .vreg_list              = qmp_phy_vreg_l,
>>>> +       .vreg_enable_load       = qmp_phy_vreg_enable_load,
>
> So, you apply this change only to the sm8250 (sc7280) config. Are you 
> sure that both of them have the same requirement?
>
> Also there are other DP phy instances (sc8180x, sc7180). Do they have 
> to be extended too?
>
>>>>          .num_vregs              = ARRAY_SIZE(qmp_phy_vreg_l),
>>>>          .regs                   = qmp_v4_usb3phy_regs_layout,
>>>>
>>>> @@ -5008,6 +5015,11 @@ static int qcom_qmp_phy_com_init(struct 
>>>> qmp_phy *qphy)
>>>>                  return 0;
>>>>          }
>>>>
>>>> +       if (cfg->vreg_enable_load) {
>>>> +               for (i = cfg->num_vregs - 1; i >= 0; --i)
>>> What's the point of iterating the list backwards?
>>
>> do no  know,
>>
>> I just follow the order from regulator_bulk_enable()
>
> regulator_bulk_enable() iterates the list in the ascending order.
>
>>
>>>
>>>> + regulator_set_load(qmp->vregs[i].consumer, 
>>>> cfg->vreg_enable_load[i]);
>>>> +       }
>>>> +
>>>>          /* turn on regulator supplies */
>>>>          ret = regulator_bulk_enable(cfg->num_vregs, qmp->vregs);
>>>>          if (ret) {
>>>> @@ -5116,6 +5128,7 @@ static int qcom_qmp_phy_com_exit(struct 
>>>> qmp_phy *qphy)
>>>>
>>>>          clk_bulk_disable_unprepare(cfg->num_clks, qmp->clks);
>>>>
>>>> +       /* no minimum load set required before disable regulator */
>>> No otneed for the comment.
>>>
>>>> regulator_bulk_disable(cfg->num_vregs, qmp->vregs);
>>>>
>>>>          mutex_unlock(&qmp->phy_mutex);
>>>> -- 
>>>> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
>>>> Forum,
>>>> a Linux Foundation Collaborative Project
>>>>
>>>
>
>