Thank you for your contribution! Sashiko AI review found 1 potential issue(s) 
to consider:
- [Medium] drm: qms_enabled property exposed to legacy userspace without 
DRM_MODE_PROP_ATOMIC
--

commit d13e141a83779dcc2e18e7f2092c22b8d25dc870
Author: Nicolas Frattaroli <[email protected]>

drm/connector: Add "qms_enabled" drm property

This commit introduces a new DRM property "qms_enabled" to allow userspace 
to request the kernel to use Quick Media Switching (QMS) on compatible HDMI 
connectors. It updates the connector state and adds atomic uAPI handlers for 
the new property.

> diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
> index 5c07bd221977c..7ea235a58781f 100644
> --- a/drivers/gpu/drm/drm_connector.c
> +++ b/drivers/gpu/drm/drm_connector.c
[ ... ]
> +int drm_connector_attach_qms_enabled_property(struct drm_connector 
> *connector)
> +{
> +     connector->qms_enabled_property =
> +             drm_property_create_bool(connector->dev, 0, "qms_enabled");

[Severity: Medium]
Does this code inadvertently expose the "qms_enabled" property to legacy
userspace clients?

When creating the property in drm_connector_attach_qms_enabled_property(),
the flag parameter is passed as 0 instead of DRM_MODE_PROP_ATOMIC.

Without the DRM_MODE_PROP_ATOMIC flag, legacy X11/KMS clients can 
inappropriately
discover this property when iterating over connector properties via 
drmModeGetConnector().

Since this property is designed strictly for atomic modesetting state, should
it be hidden from legacy userspace to avoid ABI pollution?

> +     if (!connector->qms_enabled_property)
> +             return -ENOMEM;
> +
> +     drm_object_attach_property(&connector->base, 
> connector->qms_enabled_property, 0);
> +
> +     return 0;
> +}
> +EXPORT_SYMBOL(drm_connector_attach_qms_enabled_property);

-- 
Sashiko AI review · 
https://sashiko.dev/#/patchset/[email protected]?part=17

Reply via email to