Hi Maxime,
Thank you for the patch.
On Thu, Jul 30, 2020 at 11:35:03AM +0200, Maxime Ripard wrote:
> The A20 can use its second TCON as the secondary LVDS link in a dual-link
> setup, with the TCON0 being the main link. Extend a bit the parsing code to
> leverage the DRM dual-link code, register only the LVDS output on the
> primary TCON, and add the needed bits to setup the TCON properly.
>
> Signed-off-by: Maxime Ripard <maxime@xxxxxxxxxx>
> ---
> drivers/gpu/drm/sun4i/sun4i_tcon.c | 36 +++++++++++++++++++++++++++++++-
> drivers/gpu/drm/sun4i/sun4i_tcon.h | 4 +++-
> 2 files changed, 40 insertions(+)
>
> diff --git a/drivers/gpu/drm/sun4i/sun4i_tcon.c b/drivers/gpu/drm/sun4i/sun4i_tcon.c
> index d03ad75f9900..ed2abf6eb18b 100644
> --- a/drivers/gpu/drm/sun4i/sun4i_tcon.c
> +++ b/drivers/gpu/drm/sun4i/sun4i_tcon.c
> @@ -487,6 +487,9 @@ static void sun4i_tcon0_mode_set_lvds(struct sun4i_tcon *tcon,
> else
> reg |= SUN4I_TCON0_LVDS_IF_DATA_POL_NORMAL;
>
> + if (tcon->lvds_dual_link)
> + reg |= SUN4I_TCON0_LVDS_IF_DUAL_LINK;
> +
> if (sun4i_tcon_get_pixel_depth(encoder) == 24)
> reg |= SUN4I_TCON0_LVDS_IF_BITWIDTH_24BITS;
> else
> @@ -896,6 +899,16 @@ static int sun4i_tcon_register_panel(struct drm_device *drm,
> return sun4i_rgb_init(drm, tcon);
>
> /*
> + * Only the TCON0 will be relevant for the LVDS output, so if
> + * our ID is something else, let's prevent our TCON from
> + * registering its own LVDS output
> + */
> + if (tcon->id) {
> + dev_info(dev, "Secondary TCON, disabling panel output");
This may worry the user unnecessarily. I'd make it a debug message, or
drop it completely, and like reword it a bit as pointed out by Chen-Yu.
> + return 0;
> + }
> +
> + /*
> * This can only be made optional since we've had DT
> * nodes without the LVDS reset properties.
> *
> @@ -941,6 +954,28 @@ static int sun4i_tcon_register_panel(struct drm_device *drm,
> return -ENODEV;
> }
>
> + /*
> + * If we don't have a second TCON, we will never be able to do
> + * dual-link LVDS, so we don't have much more to do.
> + */
> + companion = of_parse_phandle(dev->of_node, "allwinner,lvds-companion", 0);
Should there be a patch to add this property to the DT bindings ?
> + if (!companion)
> + return 0;
> +
> + /*
> + * Let's do a sanity check on the dual-link setup to make sure
> + * everything is properly described.
> + */
> + ret = drm_of_lvds_get_dual_link_pixel_order(dev->of_node, 1, 0,
> + companion, 1, 0);
> + if (ret < 0) {
> + dev_err(dev, "Invalid Dual-Link Configuration.\n");
> + return ret;
> + }
> +
> + dev_info(dev, "Primary TCON, enabling LVDS Dual-Link");
> + tcon->lvds_dual_link = true;
> +
> return sun4i_lvds_init(drm, tcon);
> }
>
> @@ -1500,6 +1535,7 @@ static const struct sun4i_tcon_quirks sun7i_a20_tcon0_quirks = {
> };
>
> static const struct sun4i_tcon_quirks sun7i_a20_quirks = {
> + .supports_lvds = true,
Should this be split to a separate patch, or at least mentioned in the
commit message ?
> .has_channel_0 = true,
> .has_channel_1 = true,
> .dclk_min_div = 4,
> diff --git a/drivers/gpu/drm/sun4i/sun4i_tcon.h b/drivers/gpu/drm/sun4i/sun4i_tcon.h
> index cfbf4e6c1679..51c4e09cdd13 100644
> --- a/drivers/gpu/drm/sun4i/sun4i_tcon.h
> +++ b/drivers/gpu/drm/sun4i/sun4i_tcon.h
> @@ -98,6 +98,7 @@
>
> #define SUN4I_TCON0_LVDS_IF_REG 0x84
> #define SUN4I_TCON0_LVDS_IF_EN BIT(31)
> +#define SUN4I_TCON0_LVDS_IF_DUAL_LINK BIT(30)
> #define SUN4I_TCON0_LVDS_IF_BITWIDTH_MASK BIT(26)
> #define SUN4I_TCON0_LVDS_IF_BITWIDTH_18BITS (1 << 26)
> #define SUN4I_TCON0_LVDS_IF_BITWIDTH_24BITS (0 << 26)
> @@ -274,6 +275,9 @@ struct sun4i_tcon {
> /* Associated crtc */
> struct sun4i_crtc *crtc;
>
> + /* Is the LVDS link a dual-channel link? */
> + bool lvds_dual_link;
> +
> int id;
>
> /* TCON list management */
--
Regards,
Laurent Pinchart
