Re: [PATCH] drm/bridge: dw-mipi-dsi: fix cmd mode transition when disabling

From: Icenowy Zheng

Date: Thu Nov 19 2026 - 09:38:10 EST


在 2026-10-30五的 13:58 +0800,Icenowy Zheng写道:
> With pre_enable_prev_first flag set, the atomic_post_disable hook of the
> next bridge (which should be the DSI panel bridge and calls panel
> unprepare hook) is called before the hook of the DSI controller bridge,
> which leads to the switch to command mode happening after the panel
> unprepare. In this case some panels' unprepare hooks sliently fails
> because they expect the DSI controller to be in command mode.
>
> As the DSI controller atomic_disable hook happens after panel bridge
> atomic_disable but before panel bridge atomic_post_disable, it seems to
> be a good position for the transition to the command mode.
>
> Move the mode transition from the atomic_post_disable hook to a newly
> created atomic_disable hook.
>
> Fixes: 05aa61334592 ("drm: bridge: dw-mipi-dsi: Fix enable/disable of DSI controller")
> Signed-off-by: Icenowy Zheng <zhengxingda@xxxxxxxxxxx>

Ping, would anyone take a look at this patch?

> ---
> Tested on Lichee Console 4A with a pending glue driver for T-Head TH1520
> DW MIPI DSI. This device has a always-powered panel (no switchable
> supply and no reset pin), so correct execution of unprepare commands is
> critical.
>
> drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi.c | 23 ++++++++++++-------
> 1 file changed, 15 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi.c b/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi.c
> index 5dd475fcaacf8..49b6228ca69e5 100644
> --- a/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi.c
> +++ b/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi.c
> @@ -944,14 +944,6 @@ static void dw_mipi_dsi_bridge_post_atomic_disable(struct drm_bridge *bridge,
> struct dw_mipi_dsi *dsi = bridge_to_dsi(bridge);
> const struct dw_mipi_dsi_phy_ops *phy_ops = dsi->plat_data->phy_ops;
>
> - /*
> - * Switch to command mode before panel-bridge post_disable &
> - * panel unprepare.
> - * Note: panel-bridge disable & panel disable has been called
> - * before by the drm framework.
> - */
> - dw_mipi_dsi_set_mode(dsi, 0);
> -
> if (phy_ops->power_off)
> phy_ops->power_off(dsi->plat_data->priv_data);
>
> @@ -1058,6 +1050,20 @@ static void dw_mipi_dsi_bridge_atomic_enable(struct drm_bridge *bridge,
> dw_mipi_dsi_set_mode(dsi->slave, MIPI_DSI_MODE_VIDEO);
> }
>
> +static void dw_mipi_dsi_bridge_atomic_disable(struct drm_bridge *bridge,
> + struct drm_atomic_commit *state)
> +{
> + struct dw_mipi_dsi *dsi = bridge_to_dsi(bridge);
> +
> + /*
> + * Switch to command mode before panel-bridge post_disable &
> + * panel unprepare.
> + * Note: panel-bridge disable & panel disable has been called
> + * before by the drm framework.
> + */
> + dw_mipi_dsi_set_mode(dsi, 0);

Sashiko says that the secondary controller needs to be put into command
mode too, is it the case?

If it is, I would add it to this hunk (although it's previously not
present).

Thanks,
Icenowy

> +}
> +
> static enum drm_mode_status
> dw_mipi_dsi_bridge_mode_valid(struct drm_bridge *bridge,
> const struct drm_display_info *info,
> @@ -1098,6 +1104,7 @@ static const struct drm_bridge_funcs dw_mipi_dsi_bridge_funcs = {
> .atomic_create_state = drm_atomic_helper_bridge_create_state,
> .atomic_pre_enable = dw_mipi_dsi_bridge_atomic_pre_enable,
> .atomic_enable = dw_mipi_dsi_bridge_atomic_enable,
> + .atomic_disable = dw_mipi_dsi_bridge_atomic_disable,
> .atomic_post_disable = dw_mipi_dsi_bridge_post_atomic_disable,
> .mode_set = dw_mipi_dsi_bridge_mode_set,
> .mode_valid = dw_mipi_dsi_bridge_mode_valid,