Re: [PATCH 4/8] drm/bridge: dw-hdmi: document the output_port field

[Damon Ding]

On 3/27/2026 7:10 PM, Luca Ceresoli wrote:
> Hello Damon,
>
> On Thu Mar 26, 2026 at 10:15 AM CET, Damon Ding wrote:
> > On 3/26/2026 3:25 PM, Liu Ying wrote:
> > > Hi Luca,
> > >
> > > On Fri, Mar 20, 2026 at 11:46:15AM +0100, Luca Ceresoli wrote:
> > > > The meaning of this flag may not be obvious at first sight.
> > > >
> > > > Signed-off-by: Luca Ceresoli <luca.ceresoli@xxxxxxxxxxx>
> > > > ---
> > > >    include/drm/bridge/dw_hdmi.h | 5 +++++
> > > >    1 file changed, 5 insertions(+)
> >
> > First of all, these changes related to the DW HDMI controller work well
> > when tested on RK3399 HDMI.
>
> Great!
>
> You'd be welcome to send your Tested-by: tag if you tested the series on
> hardware, that would be useful.
>
> However at this point I suggest to wait for v2, which I'm sending soon, and
> test that. I added you in Cc for it.
>
> > > > diff --git a/include/drm/bridge/dw_hdmi.h b/include/drm/bridge/dw_hdmi.h
> > > > index 336f062e1f9d..45f6ba1a8ee1 100644
> > > > --- a/include/drm/bridge/dw_hdmi.h
> > > > +++ b/include/drm/bridge/dw_hdmi.h
> > > > @@ -126,6 +126,11 @@ struct dw_hdmi_phy_ops {
> > > >    struct dw_hdmi_plat_data {
> > > >    	struct regmap *regm;
> > > >
> > > > +	/*
> > > > +	 * The HDMI output port number (which must be 1) if it is described
> > >
> > > I'd rephrase:
> > > The HDMI output port number must be 1 ...
> >
> > Yes, the output port number should be 1, but I found that the output
> > port number in the Rockchip-side dw-hdmi driver remains 0.
>
> Really? I checked all the bindings in
> Documentation/devicetree/bindings/display/rockchip/*hdmi* and all mention
> port@1 as the output port number. Can you point to code using port@0 as the
> output port?
>
> Should it be true, that would be unfortunate because the output_port
> variable does not handle this case. It's used as a sort of bool-or-int
> variable:
>
>   * as a bool [0] to find out whether the DT is supposed to describe the
>     output port
>   * as an integer to tell the port number to parse in DT [1]
>
> So saying "please parse port 0" is impossible.
>
> [0] https://elixir.bootlin.com/linux/v7.0-rc5/source/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c#L3310
> [1] https://elixir.bootlin.com/linux/v7.0-rc5/source/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c#L3315

Aha, my description might be a little misleading. The 
&dw_hdmi_plat_data.output_port is 0 on the Rockchip side, so the next 
bridge will not be parsed for it.

Then I think the &dw_hdmi_plat_data.output_port should be 1, as this 
helps support the hdmi-connector and other bridge chips.

> > Therefore, it may be better to adapt the dw-hdmi drivers across all
> > platforms to the bridge-connector architecture simultaneously.
> >
> > This would allow removing &dw_hdmi_plat_data.output_port and unify the
> > setting of DRM_BRIDGE_ATTACH_NO_CONNECTOR during the attach stage.
> > (Just as the Analogix DP driver does [0])
> >
> > [0]
> > https://lore.kernel.org/all/20260319071452.1961274-1-damon.ding@xxxxxxxxxxxxxx/
>
> I agree converting all users is a good goal, but I disagree it should be
> done simultaneously. There are various users of dw-hdmi, and converting one
> having the hardware to test it was painful enough for me. Converting all of
> them without testing on hardware would be a hell.
>
> However I might be wrong. Having a precise list of all users, which ones
> need to be converted, and whether they have any special detail to be taken
> care of would be good to estimate the work to convert all users. Without
> that I'd rather let users convert one by one and hopefully get rid of
> legacy code eventually.

Yes, I've noticed that there are quite a few drivers associated with the 
DW HDMI controller. It would be a better idea to let users handle this 
themselves.

BTW: The Rockchip side dw-hdmi patches for bridge connector support will 
be updated as a follow-up to your patch series. :-)

Best regards,
Damon