Re: [PATCH v2] drm/bridge: adv7511: Clear HPD IRQ during atomic_enable()
From: Tommaso Merciai
Date: Wed Apr 22 2026 - 09:42:30 EST
Hi Biju,
Thank you for your patch.
On Wed, Apr 22, 2026 at 01:14:54PM +0100, Biju wrote:
> From: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
>
> On RZ/G3E SMARC EVK, suspend-to-RAM via PSCI powers down the ADV7535
> chip entirely, causing the HPD status bit to be in a stale state on
> resume. When the display controller driver's system PM resume callback
> invokes drm_mode_config_helper_resume(), it calls the bridge's
> atomic_enable(), but the stale HPD IRQ is never cleared, leading to
> incorrect behaviour.
>
> Clear the HPD status bit in adv7511_bridge_atomic_enable() after
> powering on, so that any HPD interrupt latched before or during
> power-loss is dismissed before normal operation resumes.
>
> Signed-off-by: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
Patch LGTM.
Tested-by: Tommaso Merciai <tommaso.merciai.xr@xxxxxxxxxxxxxx>
Reviewed-by: Tommaso Merciai <tommaso.merciai.xr@xxxxxxxxxxxxxx>
Kind Regards,
Tommaso
> ---
> Tested HPD, s2idle and s2ram on RZ/G3L SMARC EVK connected to ADV7535
> on both polling and IRQ mode.
> v1->v2:
> * Dropped PM support instead clearing latched HPD status bit in
> adv7511_bridge_atomic_enable()
> * Dropped suspended variable from struct adv7511.
> * Updated comment in adv7511_bridge_atomic_enable().
> * Clearing the HPD status bit unconditionally as there is no harm.
> * Updated commit header and description.
> * Dropped the tags.
> ---
> drivers/gpu/drm/bridge/adv7511/adv7511_drv.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/drivers/gpu/drm/bridge/adv7511/adv7511_drv.c b/drivers/gpu/drm/bridge/adv7511/adv7511_drv.c
> index 6bd76c1fb007..7663814b4032 100644
> --- a/drivers/gpu/drm/bridge/adv7511/adv7511_drv.c
> +++ b/drivers/gpu/drm/bridge/adv7511/adv7511_drv.c
> @@ -792,6 +792,13 @@ static void adv7511_bridge_atomic_enable(struct drm_bridge *bridge,
>
> adv7511_power_on(adv);
>
> + /*
> + * Clear the HPD status bit (ADV7511_INT0_HPD), so that any HPD
> + * interrupt latched before or during power loss is dismissed before
> + * normal operation resumes.
> + */
> + regmap_write(adv->regmap, ADV7511_REG_INT(0), ADV7511_INT0_HPD);
> +
> connector = drm_atomic_get_new_connector_for_encoder(state, bridge->encoder);
> if (WARN_ON(!connector))
> return;
> --
> 2.43.0
>