mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: xiaolei wang <xiaolei.wang@windriver.com>
To: laurent.pinchart@ideasonboard.com,
	tomi.valkeinen@ideasonboard.com, kernel-list@raspberrypi.com,
	mchehab@kernel.org, florian.fainelli@broadcom.com,
	bcm-kernel-feedback-list@broadcom.com, hverkuil@xs4all.nl,
	sakari.ailus@linux.intel.com
Cc: linux-media@vger.kernel.org,
	linux-rpi-kernel@lists.infradead.org,
	linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] media: rp1-cfe: Fix use-after-free and double initialization on sensor reload
Date: Fri, 13 Mar 2026 17:54:39 +0800	[thread overview]
Message-ID: <f9c2542e-cb04-4109-920a-d7cea7f90d0e@windriver.com> (raw)
In-Reply-To: <20260212085526.1598770-1-xiaolei.wang@windriver.com>

Hi,

Gentle ping on this patch. It fixes use-after-free and double
initialization issues when sensor drivers are reloaded.

Any feedback would be appreciated.

Thanks,
Xiaolei

On 2/12/26 16:55, Xiaolei Wang wrote:
> When a sensor driver is unloaded and reloaded (e.g., rmmod/insmod
> ov5647), the cfe_async_complete callback is invoked again, causing
> multiple issues:
>
> 1. KASAN use-after-free: The callback accesses the sensor subdevice
>     that was freed during driver removal, triggering a use-after-free
>     error
> 2. Double initialization: video_register_device() attempts to
>     re-initialize the video_device kobject that is still registered,
>     causing "tried to init an initialized object" warnings
>
> Fix this by:
> - Moving video node registration from cfe_async_complete() to
>    cfe_probe(), ensuring video devices are registered only once
>    during driver initialization
> - Implementing cfe_async_unbind() callback to clear the sensor
>    pointer when the subdevice is unbound, preventing access to
>    freed memory
>
> Signed-off-by: Xiaolei Wang <xiaolei.wang@windriver.com>
> ---
>
> Changes in V2:
>    Register video nodes at probe time. I've also updated the commit log; some
>    descriptions in version V1 were inaccurate. This patch resolves the following issues.
>
>    1. KASAN use-after-free: The callback accesses the sensor subdevice
>       that was freed during driver removal, triggering a use-after-free
>       error
>
>       KASAN: slab-use-after-free in cfe_async_complete+0x3a8/0x470 [rp1_cfe]
>       Read of size 2 at addr ffff0001182fa0b8 by task insmod/742
>
>       Call trace:
>        cfe_async_complete+0x3a8/0x470 [rp1_cfe]
>        v4l2_async_nf_try_complete+0xec/0x140 [v4l2_async]
>        __v4l2_async_register_subdev+0x13c/0x3d0 [v4l2_async]
>        v4l2_async_register_subdev_sensor+0x144/0x278 [v4l2_fwnode]
>        ov5647_probe+0xa28/0xd80 [ov5647]
>
>        Allocated by task 220:
>        ov5647_probe+0xc4/0xd80 [ov5647]
>
>        Freed by task 737:
>        i2c_device_remove+0x88/0x190
>        ov5647_driver_exit+0x18/0x48 [ov5647]
>
>    2. Double initialization: video_register_device() attempts to
>       re-initialize the video_device kobject that is still registered,
>       causing "tried to init an initialized object" warnings
>
> Link to V1: https://patchwork.linuxtv.org/project/linux-media/patch/20260211034501.1815035-1-xiaolei.wang@windriver.com/
>
>   .../media/platform/raspberrypi/rp1-cfe/cfe.c  | 39 ++++++++++++-------
>   1 file changed, 24 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/media/platform/raspberrypi/rp1-cfe/cfe.c b/drivers/media/platform/raspberrypi/rp1-cfe/cfe.c
> index 62dca76b468d..c6ba71d1f0d7 100644
> --- a/drivers/media/platform/raspberrypi/rp1-cfe/cfe.c
> +++ b/drivers/media/platform/raspberrypi/rp1-cfe/cfe.c
> @@ -2151,31 +2151,19 @@ static int cfe_probe_complete(struct cfe_device *cfe)
>   
>   	cfe->v4l2_dev.notify = cfe_notify;
>   
> -	for (unsigned int i = 0; i < NUM_NODES; i++) {
> -		ret = cfe_register_node(cfe, i);
> -		if (ret) {
> -			cfe_err(cfe, "Unable to register video node %u.\n", i);
> -			goto unregister;
> -		}
> -	}
> -
>   	ret = cfe_link_node_pads(cfe);
>   	if (ret) {
>   		cfe_err(cfe, "Unable to link node pads.\n");
> -		goto unregister;
> +		return ret;
>   	}
>   
>   	ret = v4l2_device_register_subdev_nodes(&cfe->v4l2_dev);
>   	if (ret) {
>   		cfe_err(cfe, "Unable to register subdev nodes.\n");
> -		goto unregister;
> +		return ret;
>   	}
>   
>   	return 0;
> -
> -unregister:
> -	cfe_unregister_nodes(cfe);
> -	return ret;
>   }
>   
>   static int cfe_async_bound(struct v4l2_async_notifier *notifier,
> @@ -2204,8 +2192,19 @@ static int cfe_async_complete(struct v4l2_async_notifier *notifier)
>   	return cfe_probe_complete(cfe);
>   }
>   
> +static void cfe_async_unbind(struct v4l2_async_notifier *notifier,
> +			     struct v4l2_subdev *subdev,
> +			     struct v4l2_async_connection *asd)
> +{
> +	struct cfe_device *cfe = to_cfe_device(notifier->v4l2_dev);
> +
> +	cfe->source_sd = NULL;
> +	cfe_info(cfe, "Unbinding subdev %s\n", subdev->name);
> +}
> +
>   static const struct v4l2_async_notifier_operations cfe_async_ops = {
>   	.bound = cfe_async_bound,
> +	.unbind = cfe_async_unbind,
>   	.complete = cfe_async_complete,
>   };
>   
> @@ -2243,6 +2242,14 @@ static int cfe_register_async_nf(struct cfe_device *cfe)
>   	cfe->csi2.dphy.max_lanes = ep.bus.mipi_csi2.num_data_lanes;
>   	cfe->csi2.bus_flags = ep.bus.mipi_csi2.flags;
>   
> +	for (unsigned int i = 0; i < NUM_NODES; i++) {
> +		ret = cfe_register_node(cfe, i);
> +		if (ret) {
> +			cfe_err(cfe, "Unable to register video node %u.\n", i);
> +			goto err_unregister;
> +		}
> +	}
> +
>   	/* Initialize and register the async notifier. */
>   	v4l2_async_nf_init(&cfe->notifier, &cfe->v4l2_dev);
>   	cfe->notifier.ops = &cfe_async_ops;
> @@ -2252,7 +2259,7 @@ static int cfe_register_async_nf(struct cfe_device *cfe)
>   	if (IS_ERR(asd)) {
>   		ret = PTR_ERR(asd);
>   		cfe_err(cfe, "Error adding subdevice: %d\n", ret);
> -		goto err_put_local_fwnode;
> +		goto err_unregister;
>   	}
>   
>   	ret = v4l2_async_nf_register(&cfe->notifier);
> @@ -2267,6 +2274,8 @@ static int cfe_register_async_nf(struct cfe_device *cfe)
>   
>   err_nf_cleanup:
>   	v4l2_async_nf_cleanup(&cfe->notifier);
> +err_unregister:
> +	cfe_unregister_nodes(cfe);
>   err_put_local_fwnode:
>   	fwnode_handle_put(local_ep_fwnode);
>   

      reply	other threads:[~2026-03-13  9:55 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-12  8:55 Xiaolei Wang
2026-03-13  9:54 ` xiaolei wang [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f9c2542e-cb04-4109-920a-d7cea7f90d0e@windriver.com \
    --to=xiaolei.wang@windriver.com \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=florian.fainelli@broadcom.com \
    --cc=hverkuil@xs4all.nl \
    --cc=kernel-list@raspberrypi.com \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-rpi-kernel@lists.infradead.org \
    --cc=mchehab@kernel.org \
    --cc=sakari.ailus@linux.intel.com \
    --cc=tomi.valkeinen@ideasonboard.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®