Re: [PATCH v3 01/19] media: dt-bindings: media: camss: Add qcom,sm8250-camss binding

[Bryan O'Donoghue <bryan.odonoghue@linaro.org>]

On 04/01/2022 15:28, Bryan O'Donoghue wrote:
> On 04/01/2022 13:42, Vladimir Zapolskiy wrote:
>> Hi Bryan, Robert,
>>
>> On 12/22/21 2:37 AM, Bryan O'Donoghue wrote:
>>> From: Jonathan Marek <jonathan@marek.ca>
>>>
>>> Add bindings for qcom,sm8250-camss in order to support the camera
>>> subsystem for SM8250.
>>>
>>> Cc: devicetree@vger.kernel.org
>>> Signed-off-by: Jonathan Marek <jonathan@marek.ca>
>>> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
>>> Reviewed-by: Rob Herring <robh@kernel.org>
>>> ---
>>>   .../bindings/media/qcom,sm8250-camss.yaml     | 450 ++++++++++++++++++
>>>   1 file changed, 450 insertions(+)
>>>   create mode 100644 
>>> Documentation/devicetree/bindings/media/qcom,sm8250-camss.yaml
>>>
>>> diff --git 
>>> a/Documentation/devicetree/bindings/media/qcom,sm8250-camss.yaml 
>>> b/Documentation/devicetree/bindings/media/qcom,sm8250-camss.yaml
>>
>> <snip>
>>
>>> +required:
>>> +  - clock-names
>>> +  - clocks
>>> +  - compatible
>>> +  - interconnects
>>> +  - interconnect-names
>>> +  - interrupts
>>> +  - interrupt-names
>>> +  - iommus
>>> +  - power-domains
>>> +  - reg
>>> +  - reg-names
>>> +
>>> +additionalProperties: false
>>
>> I've discovered that there is a noticeable difference between this 
>> bindings
>> and all the previous ones, for instance see qcom,sdm845-camss.yaml
>>
>> There is no required 'vdda-supply' property on the list, and fwiw I 
>> believe
>> there should be two supply properties for 0p9 and 1p2 supplies in fact.
>> Similarly, two separate supplies should be present in sdm845 camss 
>> bindings.
> 
> The 1p2 supply is defined in the camera sensor section as dvdd-supply
> 
> https://git.linaro.org/people/bryan.odonoghue/kernel.git/tree/arch/arm64/boot/dts/qcom/qrb5165-rb5.dts?h=v5.16-rc6-sm8250-camss-imx577-only-v3 
> 
> 
> 1p2 connects to dvdd-supply as per Miura(865-RB5)-Camera_NAV_Mezz page 
> 10 - thundercomm registration required to view
> 
> arch/arm64/boot/dts/qcom/qrb5165-rb5.dts::&cci_i2c2::camera@1a
> {
>      dovdd-supply  = <&vreg_s4a_1p8>;
>      avdd-supply = <&vreg_l7f_1p8>;
>      dvdd-supply = <&vreg_l9a_1p2>;
> }
> 
> similarly to
> 
> arch/arm64/boot/dts/qcom/sdm845-db845c.dts::&cci_i2c0::camera@10 {
>      dovdd-supply = <&vreg_lvs1a_1p8>;
>      avdd-supply = <&cam0_avdd_2v8>;
>      dvdd-supply = <&cam0_dvdd_1v2>;
> }
> 
> and
> 
> arch/arm64/boot/dts/qcom/apq8016-sbc.dts::&cci_i2c0::camera_rear@3b
> {
>      vdddo-supply = <&camera_vdddo_1v8>;
>      vdda-supply = <&camera_vdda_2v8>;
>      vddd-supply = <&camera_vddd_1v5>; /* bod: here */
> }
> 
> The IMX sensor needs to have the regulator_bulk_enable() stuff added, 
> I'll post this patch it works standalone
> 
> https://git.linaro.org/people/bryan.odonoghue/kernel.git/commit/?h=v5.16-rc6-sm8250-camss-imx577-only-v3&id=e82fc1b29d9227cad3ad7dcab362c39dd4a63bdb 
> 
> 
> Simiar to 0c2c7a1e0d69 ("media: ov8856: Add devicetree support")
> 
> Downstream points the CPA to "camss-vdd-supply = <&titan_top_gdsc>;" 
> which is covered by our TITAN_TOP_GDSC power-domain and 
> "mipi-csi-vdd-supply = <&pm8150_l9>;"
> 
> regulator-pm8150-l9 == rpmh-regulator-ldoa9 == ldoa9 == pmic5-ldo
> 
> aka l9a upstream
> 
> "dvdd-supply = <&vreg_l9a_1p2>;"
> 
> vreg_l9a_1p2: ldo9 {
>      regulator-name = "vreg_l9a_1p2";
>      regulator-min-microvolt = <1200000>;
>      regulator-max-microvolt = <1200000>;
>      regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
> };
> 
> Not sure I see on the schematic or in the downstream dts how 0p9 is 
> connected to the camera - seems to be a pcie and or mdss-dsi regulator.
> 
> If vdda-supply is a required property of the camera and not the sensor 
> then what regulator do you think it should point to ?
> 
>> At the moment the driver operates with 'vdda' supply only, commit 
>> 9e5d1581
>> introduced undocumented 'vdd_sec' for sdm660, but, if I'm not mistaken,
>> it's unused.
> 
> I agree with you there - vdd_sec is either unspecified in the 
> Documentation or not required.
> 
> ---
> bod

Keep guessing there BOD.

Having a deep dive on the schematics of the RB3 and the RB5 there are a 
number of broken assumptions on regulators that "just happen to work"

I've corrected these for RB5 and will presently validate on RB3, I've 
added in regulator_bulk_enable and regulator_bulk_disable which should 
capture what we need to do for sm8250 and 630 - once we fixup the DTS 
for 630

Note: I don't have schematics for 630 or to my knowledge DTS either - I 
haven't looked all that hard, perhaps its easy to find.

The CSI vdda enable is still required in the sensor but @

dovdd-supply = <&vreg_l7f_1p8>; not dovdd-supply  = <&vreg_s4a_1p8>;

The sensor stuff hasn't been submitted upstream but still my v3 tree is 
wrong.

On RB5 if we switched off USB, UFS and PCIe there's no way the camera 
could come up.

The patch in the series I picked up from @Jonathan to drop regulator 
enable for rb3 is also wrong.

I've got the RB5 part working a bit better - need to fixup for RB3 and 
will resend

git.linaro.org/people/bryan.odonoghue/kernel.git / 
v5.16-rc6-sm8250-camss-imx577-only-v4


Thanks for the ping @Vladimir.

Please don't apply V3 for now

---
bod