Re: [PATCH 2/3] drm/rockchip: add ability to handle external dphys in mipi-dsi

[Heiko Stuebner]

Hi Laurent,

Am Mittwoch, 6. November 2019, 14:05:57 CET schrieb Laurent Pinchart:
> On Wed, Nov 06, 2019 at 12:26:49PM +0100, Heiko Stuebner wrote:
> > While the common case is that the dsi controller uses an internal dphy,
> > accessed through the phy registers inside the dsi controller, there is
> > also the possibility to use a separate dphy from a different vendor.
> > 
> > One such case is the Rockchip px30 that uses a Innosilicon Mipi dphy,
> > so add the support for handling such a constellation, including the pll
> > also getting generated inside that external phy.
> > 
> > Signed-off-by: Heiko Stuebner <heiko.stuebner@xxxxxxxxxxxxxxxxxxxxx>
> > ---
> >  .../display/rockchip/dw_mipi_dsi_rockchip.txt |  7 ++-
> >  .../gpu/drm/rockchip/dw-mipi-dsi-rockchip.c   | 54 ++++++++++++++++++-
> >  2 files changed, 57 insertions(+), 4 deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/display/rockchip/dw_mipi_dsi_rockchip.txt b/Documentation/devicetree/bindings/display/rockchip/dw_mipi_dsi_rockchip.txt
> > index ce4c1fc9116c..8b25156a9dcf 100644
> > --- a/Documentation/devicetree/bindings/display/rockchip/dw_mipi_dsi_rockchip.txt
> > +++ b/Documentation/devicetree/bindings/display/rockchip/dw_mipi_dsi_rockchip.txt
> > @@ -8,8 +8,9 @@ Required properties:
> >  	      "rockchip,rk3399-mipi-dsi", "snps,dw-mipi-dsi".
> >  - reg: Represent the physical address range of the controller.
> >  - interrupts: Represent the controller's interrupt to the CPU(s).
> > -- clocks, clock-names: Phandles to the controller's pll reference
> > -  clock(ref) and APB clock(pclk). For RK3399, a phy config clock
> > +- clocks, clock-names: Phandles to the controller's and APB clock(pclk)
> > +  and either a pll reference clock(ref) (internal dphy) or pll clock(pll)
> > +  (when connected to an external phy). For RK3399, a phy config clock
> 
> Why does external PHY clock need to be specified here ? Shouldn't it be
> handled by the PHY instead ?

You're completely right and it seems I didn't "see the forest  for the trees",
as there actually exists the phy_configure_* structs to transfer parameters
to an external phy in a correct way.

I'll revise my approach (and the phy driver) accordingly.

Thanks for the push in the right direction :-)
Heiko