On Wed, Jan 31, 2024 at 08:57:21AM +0000, Keith Zhao wrote:
> > > +static const struct vs_dc_info dc8200_info = {
> > > + .name = "DC8200",
> > > + .panel_num = 2,
> > > + .plane_num = 8,
> > > + .planes = dc_hw_planes_rev0,
> > > + .layer_num = 6,
> > > + .max_bpc = 10,
> > > + .color_formats = DRM_COLOR_FORMAT_RGB444 |
> > > + DRM_COLOR_FORMAT_YCBCR444 |
> > > + DRM_COLOR_FORMAT_YCBCR422 |
> > > + DRM_COLOR_FORMAT_YCBCR420,
> > > + .gamma_size = GAMMA_EX_SIZE,
> > > + .gamma_bits = 12,
> > > + .pitch_alignment = 128,
> > > + .pipe_sync = false,
> > > + .background = true,
> > > + .panel_sync = true,
> > > + .cap_dec = true,
> > > +};
> >
> > I really think that entire thing is to workaround a suboptimal device tree binding.
> > You should have two CRTCs in the device tree, you'll probe twice, and you won't
> > get to do that whole dance.
> >
> Hi Maxime:
> I tried to modify it according to this idea Found it too difficult In terms of hardware,
> the two crtc designs are too close to separate, and they are even designed into the same reg with different bits representing crtc0 and crtc1.
> It seems not easy to described the 2 ctrc hardware by 2 device nodes
>
> The idea is to avoid a whole dance
> I don't know if I understand correctly about whole dance.
> Is it means I create 2 ctrc and 8 plane in the dc_bind?
It looks like you just sent the same mail?
Maxime
Attachment:
signature.asc
Description: PGP signature
