Re: [PATCH v11 2/7] dt-bindings: media: qcom,x1e80100-camss: Add support for combo-mode endpoints
From: Konrad Dybcio
Date: Fri Mar 27 2026 - 06:42:30 EST
On 3/26/26 3:08 AM, Bryan O'Donoghue wrote:
> On 26/03/2026 01:51, Vladimir Zapolskiy wrote:
>> On 3/26/26 03:28, Bryan O'Donoghue wrote:
>>> Qualcomm CSI2 PHYs support a mode where two sensors may be attached to the
>>> one CSIPHY.
>>>
>>> When we have one endpoint we may have
>>> - DPHY 1, 2 or 4 data lanes + 1 clock lane
>>> - CPHY 3 wire data lane
>>>
>>> When we have two endpoints this indicates the special fixed combo-mode.
>>> - DPHY endpoint0 => 2+1 and endpoint1 => 1+1 data-lane/clock-lane combination.
>>>
>>> Reviewed-by: Christopher Obbard <christopher.obbard@xxxxxxxxxx>
>>> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@xxxxxxxxxx>
>>> ---
[...]
>>> + bus-type:
>>> + const: 4 # Combo is D-PHY specific
>>> +
>>
>> It's unclear why both 'bus-type' and 'phys' cell argument are needed
>> at the same time, they are equal and thus one of two is redundant.
>>
>
> bus-type lives on the CAMSS controller endpoint. It tells the V4L2 fwnode parser (v4l2_fwnode_endpoint_parse) how to interpret the endpoint properties — DPHY has data-lanes + clock-lanes, CPHY has trios.
>
> PHY phandle cell lives on the phys reference. It tells the PHY driver which electrical mode to configure
But we don't need that second part, no?
If it's strictly required that we keep the bus-type in DT, we already
store that information once and can translate MEDIA_BUS_TYPE_CSI2_DPHY
to PHY_MODE_MIPI_CSI or whatever before we power on the PHY (which we
wouldn't do without first setting up other bits of the topology anyway)
Konrad