On Wed, Aug 20, 2025 at 05:18:13PM +0800, Damon Ding wrote:
> Hi Dmitry,
>
> On 8/17/2025 12:43 AM, Dmitry Baryshkov wrote:
> > On Thu, Aug 14, 2025 at 06:47:47PM +0800, Damon Ding wrote:
> > > Apply drm_bridge_connector helper for Analogix DP driver.
> > >
> > > The following changes have been made:
> > > - Apply drm_bridge_connector helper to get rid of &drm_connector_funcs
> > > and &drm_connector_helper_funcs.
> > > - Remove unnecessary parameter struct drm_connector* for callback
> > > &analogix_dp_plat_data.attach.
> > > - Remove &analogix_dp_device.connector.
> > > - Convert analogix_dp_atomic_check()/analogix_dp_detect() to
> > > &drm_bridge_funcs.atomic_check()/&drm_bridge_funcs.detect().
> > > - Split analogix_dp_get_modes() into &drm_bridge_funcs.get_modes() and
> > > &drm_bridge_funcs.edid_read().
> > >
> > > Signed-off-by: Damon Ding <damon.ding@xxxxxxxxxxxxxx>
> > >
> > > ------
> > >
> > > Changes in v2:
> > > - For &drm_bridge.ops, remove DRM_BRIDGE_OP_HPD and add
> > > DRM_BRIDGE_OP_EDID.
> > > - Add analogix_dp_bridge_edid_read().
> > > - Move &analogix_dp_plat_data.skip_connector deletion to the previous
> > > patches.
> > >
> > > Changes in v3:
> > > - Rebase with the new devm_drm_bridge_alloc() related commit
> > > 48f05c3b4b70 ("drm/bridge: analogix_dp: Use devm_drm_bridge_alloc()
> > > API").
> > > - Expand the commit message.
> > > - Call drm_bridge_get_modes() in analogix_dp_bridge_get_modes() if the
> > > bridge is available.
> > > - Remove unnecessary parameter struct drm_connector* for callback
> > > &analogix_dp_plat_data.attach.
> > > - In order to decouple the connector driver and the bridge driver, move
> > > the bridge connector initilization to the Rockchip and Exynos sides.
> > >
> > > Changes in v4:
> > > - Expand analogix_dp_bridge_detect() parameters to &drm_bridge and
> > > &drm_connector.
> > > - Rename the &analogix_dp_plat_data.bridge to
> > > &analogix_dp_plat_data.next_bridge.
> > > ---
> > > .../drm/bridge/analogix/analogix_dp_core.c | 145 ++++++++----------
> > > .../drm/bridge/analogix/analogix_dp_core.h | 1 -
> > > drivers/gpu/drm/exynos/exynos_dp.c | 18 ++-
> > > .../gpu/drm/rockchip/analogix_dp-rockchip.c | 11 +-
> > > include/drm/bridge/analogix_dp.h | 3 +-
> > > 5 files changed, 88 insertions(+), 90 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> > > index 7876b310aaed..a8ed44ec8ef5 100644
> > > --- a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> > > +++ b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> > > @@ -947,24 +947,16 @@ static int analogix_dp_disable_psr(struct analogix_dp_device *dp)
> > > return analogix_dp_send_psr_spd(dp, &psr_vsc, true);
> > > }
> > > -static int analogix_dp_get_modes(struct drm_connector *connector)
> > > +static int analogix_dp_bridge_get_modes(struct drm_bridge *bridge, struct drm_connector *connector)
> > > {
> > > - struct analogix_dp_device *dp = to_dp(connector);
> > > - const struct drm_edid *drm_edid;
> > > + struct analogix_dp_device *dp = to_dp(bridge);
> > > int num_modes = 0;
> > > - if (dp->plat_data->panel) {
> > > + if (dp->plat_data->panel)
> > > num_modes += drm_panel_get_modes(dp->plat_data->panel, connector);
> > > - } else {
> > > - drm_edid = drm_edid_read_ddc(connector, &dp->aux.ddc);
> > > - drm_edid_connector_update(&dp->connector, drm_edid);
> > > -
> > > - if (drm_edid) {
> > > - num_modes += drm_edid_connector_add_modes(&dp->connector);
> > > - drm_edid_free(drm_edid);
> > > - }
> > > - }
> > > + if (dp->plat_data->next_bridge)
> > > + num_modes += drm_bridge_get_modes(dp->plat_data->next_bridge, connector);
> >
> > If there is a next bridge which provides OP_MODES, then
> > drm_bridge_connector will use it for get_modes() and skip this one
> > completely. I'm not sure what's the value of this call.
>
> Following your advice, it is really a good idea to distinguish the
> drm_bridge_ops between the panel and the bridge. Will add it in v5.
>
> >
> > > if (dp->plat_data->get_modes)
> > > num_modes += dp->plat_data->get_modes(dp->plat_data, connector);
> > > @@ -972,51 +964,39 @@ static int analogix_dp_get_modes(struct drm_connector *connector)
> > > return num_modes;
> > > }
> > > -static struct drm_encoder *
> > > -analogix_dp_best_encoder(struct drm_connector *connector)
> > > +static const struct drm_edid *analogix_dp_bridge_edid_read(struct drm_bridge *bridge,
> > > + struct drm_connector *connector)
> > > {
> > > - struct analogix_dp_device *dp = to_dp(connector);
> > > + struct analogix_dp_device *dp = to_dp(bridge);
> > > + const struct drm_edid *drm_edid = NULL;
> > > - return dp->encoder;
> > > -}
> > > + drm_edid = drm_edid_read_ddc(connector, &dp->aux.ddc);
> > > + if (dp->plat_data->get_modes)
> > > + dp->plat_data->get_modes(dp->plat_data, connector);
> >
> >
> > So, we have DDC, but we still want to return platform modes? What is the
> > usecase for that?
> >
> > There might be some, but I think it deserves a comment in the source
> > file.
> >
>
> For Rockchip side, since RK3588 and RK3576 can support YUV formats while the
> other can not, the &analogix_dp_plat_data.get_modes() help filter out YUV
> formats for some platforms(The YUV feature support may not be fit for this
> patch series and will come later).
Note, get_modes() here adds modes rather than filtering them. You can
use .mode_valid in order to filter out YUV modes.
>
> For Exynos side, I think &analogix_dp_plat_data.get_modes() can help
> parse the video mode set in the eDP DT node when there is no available panel
> or bridge.
I think this should be handled by a separate bridge. E.g. see how the
imx-legacy-bridge is implemented.
>
> I will add some comments about it in the next version.
>
--
With best wishes
Dmitry
