[PATCH v2 6/6] drm/vs: Add hdmi driver

[Maxime Ripard]

On Thu, Oct 26, 2023 at 11:57:22AM +0300, Dmitry Baryshkov wrote:
> On Thu, 26 Oct 2023 at 11:07, Maxime Ripard <mripard at kernel.org> wrote:
> >
> > On Thu, Oct 26, 2023 at 01:23:53AM +0300, Dmitry Baryshkov wrote:
> > > > +static int starfive_hdmi_register(struct drm_device *drm, struct starfive_hdmi *hdmi)
> > > > +{
> > > > +   struct drm_encoder *encoder = &hdmi->encoder;
> > > > +   struct device *dev = hdmi->dev;
> > > > +
> > > > +   encoder->possible_crtcs = drm_of_find_possible_crtcs(drm, dev->of_node);
> > > > +
> > > > +   /*
> > > > +    * If we failed to find the CRTC(s) which this encoder is
> > > > +    * supposed to be connected to, it's because the CRTC has
> > > > +    * not been registered yet.  Defer probing, and hope that
> > > > +    * the required CRTC is added later.
> > > > +    */
> > > > +   if (encoder->possible_crtcs == 0)
> > > > +           return -EPROBE_DEFER;
> > > > +
> > > > +   drm_encoder_helper_add(encoder, &starfive_hdmi_encoder_helper_funcs);
> > > > +
> > > > +   hdmi->connector.polled = DRM_CONNECTOR_POLL_HPD;
> > > > +
> > > > +   drm_connector_helper_add(&hdmi->connector,
> > > > +                            &starfive_hdmi_connector_helper_funcs);
> > > > +   drmm_connector_init(drm, &hdmi->connector,
> > > > +                       &starfive_hdmi_connector_funcs,
> > > > +                       DRM_MODE_CONNECTOR_HDMIA,
> > >
> > > On an embedded device one can not be so sure. There can be MHL or HDMI
> > > Alternative Mode. Usually we use drm_bridge here and drm_bridge_connector.
> >
> > On an HDMI driver, it's far from being a requirement, especially given
> > the limitations bridges have.
> 
> It's a blessing that things like MHL / HDMI-in-USB-C / HDMI-to-MyDP
> are not widely used in the wild and are mostly non-existing except
> several phones that preate wide DP usage.

And those can be supported without relying on bridges.

> Using drm_connector directly prevents one from handling possible
> modifications on the board level. For example, with the DRM connector
> in place, handling a separate HPD GPIO will result in code duplication
> from the hdmi-connector driver. Handling any other variations in the
> board design (which are pretty common in the embedded world) will also
> require changing the driver itself. drm_bridge / drm_bridge_connector
> save us from those issues.

And we have other solutions there too. Like, EDIDs are pretty much in
the same spot with a lot of device variations, but it also works without
a common driver. I'd really wish we were having less bridges and more
helpers, but here we are.

> BTW: what are the limitations of the drm_bridge wrt. HDMI output? I'm
> asking because we heavily depend on the bridge infrastructure for HDMI
> output. Maybe we are missing something there, which went unnoticed to
> me and my colleagues.

A bridge cannot extend the connector state or use properties, for
example. It works for basic stuff but falls apart as soon as you're
trying to do something slightly advanced.

Maxime
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-riscv/attachments/20231026/3f618210/attachment.sig>