[PATCH 2/2] phy: qcom: qmp-pcie: Add IPQ9650 PCIe PHY support

[Kathiravan Thirumoorthy]

On 6/12/2026 11:44 AM, Dmitry Baryshkov wrote:
> On Fri, Jun 12, 2026 at 08:22:02AM +0530, Kathiravan Thirumoorthy wrote:
>> On 6/12/2026 1:52 AM, Dmitry Baryshkov wrote:
>>> On Tue, Jun 09, 2026 at 03:46:56PM +0530, Kathiravan Thirumoorthy wrote:
>>>> On 6/8/2026 12:26 PM, Dmitry Baryshkov wrote:
>>>>> On Tue, Jun 02, 2026 at 02:40:18PM +0530, Kathiravan Thirumoorthy wrote:
>>>>>> The IPQ9650 platform has three Gen3 2-lane PCIe controllers and two Gen3
>>>>>> 1-lane PCIe controllers. The PHY instances also require the on-chip refgen
>>>>>> supply.
>>>>>>
>>>>>> Add the IPQ9650 Gen3 x1 and x2 QMP PCIe PHY configurations, including the
>>>>>> refgen regulator supply.
>>>>>>
>>>>>> Signed-off-by: Kathiravan Thirumoorthy <kathiravan.thirumoorthy at oss.qualcomm.com>
>>>>>> ---
>>>>>>     drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 220 +++++++++++++++++++++++++++++++
>>>>>>     1 file changed, 220 insertions(+)
>>>>>>
>>>>>> @@ -3378,6 +3524,10 @@ static const char * const qmp_phy_vreg_l[] = {
>>>>>>     	"vdda-phy", "vdda-pll",
>>>>>>     };
>>>>>> +static const char * const ipq9650_qmp_phy_vreg_l[] = {
>>>>>> +	"refgen",
>>>>>> +};
>>>>> Now vdda-phy / vdda-pll supplies?
>>>> Cross checked with HW team again. Along with refgen, there is a on-chip LDO
>>>> which supplies fixed voltage to the PHYs. It is enabled upon system power on
>>>> and no SW intervention is required.
>>> What is it being powered by? MX? CX?
>> It is driven by CX.
> I assume that there is no CX collapse on IPQ9650? Is CX not scaling on
> this chip. Please provide some details on the commit message.

That's right. No CX collapse on IPQ9650. Let me rewrite the commit 
message as below. Hope its okay.

--

Add support for the IPQ9650 platform, which includes three Gen3 x2 PCIe 
controllers and two Gen3 x1 PCIe controllers. The PHY instances require 
the on-chip refgen supply.

Add the IPQ9650 Gen3 x1 and x2 QMP PCIe PHY configurations along with 
the refgen regulator supply. Note that an on-chip LDO, driven by the SoC 
CX, supplies the PHY voltages without requiring software control. Note 
that CX power collapse is not supported on IPQ9650.

--

>
>>>> regulator-fixed doesn't take the resource 'reg'. May be should I create
>>>> another regulator driver which accepts 'reg', something similar to the
>>>> qcom-refgen-regulator? Please advise.
>>> If it doesn't require control, there is no need for a separate driver or
>>> separate supply. For example, the refgen is being references only by
>>> those devices which require software votes.
>> Thanks. Then let me respin this series on top of phy-next so that Vinod can
>> pick it up.
>>
>>>>>> +
>>>>>>     static const char * const sm8550_qmp_phy_vreg_l[] = {
>>>>>>     	"vdda-phy", "vdda-pll", "vdda-qref",
>>>>>>     };
>> -- 
>> linux-phy mailing list
>> linux-phy at lists.infradead.org
>> https://lists.infradead.org/mailman/listinfo/linux-phy