Re: [PATCH RFC 25/25] drm/atomic: Disable VRR in helper_set_config

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] drm: __drm_atomic_helper_set_config unconditionally disables VRR for legacy clients
--

commit 353fccdd37726114a8ead2ac16a0d6cf9e239631
Author: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>

drm/atomic: Disable VRR in helper_set_config

This commit explicitly sets crtc_state->vrr_enabled to false in
__drm_atomic_helper_set_config. The goal is to prevent a non-functional KMS
fbdev console if a DRM client exits without properly cleaning up its VRR
state.

> diff --git a/drivers/gpu/drm/drm_atomic.c b/drivers/gpu/drm/drm_atomic.c
> index 9bb761eda59f4..1d3c35f7788eb 100644
> --- a/drivers/gpu/drm/drm_atomic.c
> +++ b/drivers/gpu/drm/drm_atomic.c
> @@ -2051,6 +2051,8 @@ int __drm_atomic_helper_set_config(struct drm_mode_set *set,
>  
>  	crtc_state->active = true;
>  
> +	crtc_state->vrr_enabled = false;
> +

[Severity: High]
Does this unconditionally disable VRR for legacy clients that use
DRM_IOCTL_MODE_SETCRTC?

If a legacy KMS client (like X.org) enables VRR via
DRM_IOCTL_MODE_OBJ_SETPROPERTY and later calls DRM_IOCTL_MODE_SETCRTC (for
DPMS, panning, or resolution changes), won't this unconditionally clear their
persistent state by setting crtc_state->vrr_enabled to false?

Additionally, can this cause unnecessary full hardware modesets?

If drm_atomic_helper_check_modeset() sees vrr_enabled transition from true to
false, it sets mode_changed = true. This could force a slow, full modeset
(with screen blanking/flicker) during what would normally be a fast, seamless
display update.

>  	ret = drm_atomic_set_crtc_for_plane(primary_state, crtc);
>  	if (ret != 0)
>  		return ret;

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260921-vrr-limiter-uapi-v1-0-2fcd7d011646@collabora.com?part=25