On Thu, Jul 23, 2020 at 04:11:51PM +0530, Vinod Koul wrote:
> Hi Sam, Laurent,
>
> On 22-07-20, 16:14, Laurent Pinchart wrote:
> > > > +static int lt9611_bridge_attach(struct drm_bridge *bridge,
> > > > + enum drm_bridge_attach_flags flags)
> > > > +{
> > > > + struct lt9611 *lt9611 = bridge_to_lt9611(bridge);
> > > > + int ret;
> > > > +
> > > > + if (!(flags & DRM_BRIDGE_ATTACH_NO_CONNECTOR)) {
> > > > + dev_err(lt9611->dev, "Fix bridge driver to make connector optional!");
> > > > + return -EINVAL;
> > > > + }
> > >
> > > This should say that the display driver should be fixed.
> > > If a display driver expects this bridge to create the connector
> > > it would not work.
> >
> > Actually, for new bridge drivers, connector creation should be optional
> > from the start. We don't want a failure in that case, the feature should
> > be implemented.
>
> Yes this is causing issues for me now !. The patch 4/4 adds support in
> msm/dsi but causes regression on qualcomm laptops with ti-sn65dsi86 eDP
> bridge. I tried to fix that up with changes like Laurent has done for
> adv7511, but it hasnt worked yet for me (remote debug of this is bit
> painful)
>
> So I am going to drop patch 4 from this series and add support for both
> DRM_BRIDGE_ATTACH_NO_CONNECTOR set and cleared (like we have in adv7511)
> so that it can work in both cases, while I fix all bridge uses of
> msm/dsi and then we can drop these. Does that sound okay to you folks?
Yes, sounds like a good plan.
Only when all display drivers are migrated over can we drop all the
workarounds in the bridge drivers.
I had hoped all users of this bridge was converted - alas that was not
the case.
Sam
