Re: [PATCH v2 22/37] drm/bridge: ti-tfp410: Switch to atomic bridge callbacks

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] This problem wasn't introduced by this patch, but there is a use-after-free vulnerability and missing cleanup of the HPD registration on the `tfp410_attach` error path.
--

--- Patch [22]: [PATCH v2 22/37] drm/bridge: ti-tfp410: Switch to atomic bridge callbacks ---
commit fb525e8e18293a40cab99e1a43c62c7f5ee18083
Author: Maxime Ripard <mripard@kernel.org>

drm/bridge: ti-tfp410: Switch to atomic bridge callbacks

This commit switches the ti-tfp410 bridge from legacy non-atomic enable and
disable callbacks to their atomic counterparts. It also adds the necessary
atomic state handlers to help remove legacy support from the DRM bridge core.

> diff --git a/drivers/gpu/drm/bridge/ti-tfp410.c b/drivers/gpu/drm/bridge/ti-tfp410.c
> index 6f06f7b4e9926..f0a14f52aaace 100644
> --- a/drivers/gpu/drm/bridge/ti-tfp410.c
> +++ b/drivers/gpu/drm/bridge/ti-tfp410.c
[ ... ]
> @@ -242,8 +244,8 @@ static int tfp410_atomic_check(struct drm_bridge *bridge,
>  static const struct drm_bridge_funcs tfp410_bridge_funcs = {
>  	.attach		= tfp410_attach,
>  	.detach		= tfp410_detach,
> -	.enable		= tfp410_enable,
> -	.disable	= tfp410_disable,

[Severity: High]
This isn't a bug introduced by this patch, but there appears to be a
use-after-free risk and missing cleanup of the HPD registration on the error
path in tfp410_attach().

When tfp410_attach() successfully registers the HPD callback on the next bridge,
but drm_connector_init_with_ddc() subsequently fails, it returns an error
without disabling the HPD callback:

drivers/gpu/drm/bridge/ti-tfp410.c:tfp410_attach() {
    ...
	ret = drm_connector_init_with_ddc(bridge->dev, &dvi->connector,
					  &tfp410_con_funcs,
					  dvi->bridge.next_bridge->type,
					  dvi->bridge.next_bridge->ddc);
	if (ret) {
		dev_err(dvi->dev, "drm_connector_init_with_ddc() failed: %d\n",
			ret);
		return ret;
	}
    ...
}

If the attachment fails, .detach() isn't called, and the next bridge might still
retain the HPD callback pointing to the freed dvi structure. A hardware
hotplug event could then trigger the callback and dereference the freed pointer.

Should this error path deregister the HPD callback before returning?

> +	.atomic_enable = tfp410_enable,
> +	.atomic_disable = tfp410_disable,
>  	.mode_valid	= tfp410_mode_valid,
>  	.atomic_create_state = drm_atomic_helper_bridge_create_state,
>  	.atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state,

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260707-drm-all-atomic-bridges-v2-0-21d03cbca446@kernel.org?part=22