Re: [PATCH v4 2/2] phy: qcom-mipi-csi2: Add a CSI2 MIPI DPHY driver

[Bryan O'Donoghue]

On 18/03/2026 10:15, Neil Armstrong wrote:
> > +    /*
> > +     * phy_configure_opts_mipi_dphy.lanes starts from zero to
> > +     * the maximum number of enabled lanes.
> > +     *
> > +     * TODO: add support for bitmask of enabled lanes and polarities
> > +     * of those lanes to the phy_configure_opts_mipi_dphy struct.
> > +     * For now take the polarities as zero and the position as fixed
> > +     * this is fine as no current upstream implementation maps 
> > otherwise.
> > +     */
>
> This is wrong since you loose the lanes mapping defined in DT, which is 
> still in CAMSS
> but is a PHY property. The lanes layout is not a property of the CSI 
> controller,
> CSI controller only need to know the lanes count, and not the layout.

Lane layout is a PHY concern but, the PHY API gives us 
phy_configure_opts_mipi_dphy which should be extended to provide layout 
and polarity. This would then be of benefit to more than just qcom/camss.

Right now none of the CAMSS users for this driver depend on any other 
mapping and I propose a separate series to fix 
phy_configure_opts_mipi_dphy rather than introduce data-lanes to DPHY.

---
bod