Re: [PATCH 2/2] media: uvcvideo: Remove dangling pointers

From: Laurent Pinchart
Date: Wed Nov 27 2024 - 04:35:30 EST


Hi Ricardo,

Thank you for the patch.

On Wed, Nov 27, 2024 at 07:46:11AM +0000, Ricardo Ribalda wrote:
> When an async control is written, we copy a pointer to the file handle
> that started the operation. That pointer will be used when the device is
> done. Which could be anytime in the future.
>
> If the user closes that file descriptor, its structure will be freed,
> and there will be one dangling pointer per pending async control, that
> the driver will try to use.
>
> Clean all the dangling pointers during release().
>
> To avoid adding a performance penalty in the most common case (no async
> operation). A counter has been introduced with some logic to make sure
> that it is properly handled.
>
> Cc: stable@xxxxxxxxxxxxxxx
> Fixes: e5225c820c05 ("media: uvcvideo: Send a control event when a Control Change interrupt arrives")
> Signed-off-by: Ricardo Ribalda <ribalda@xxxxxxxxxxxx>
> ---
> drivers/media/usb/uvc/uvc_ctrl.c | 40 ++++++++++++++++++++++++++++++++++++++--
> drivers/media/usb/uvc/uvc_v4l2.c | 2 ++
> drivers/media/usb/uvc/uvcvideo.h | 3 +++
> 3 files changed, 43 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/media/usb/uvc/uvc_ctrl.c b/drivers/media/usb/uvc/uvc_ctrl.c
> index 5d3a28edf7f0..51a53ad25e9c 100644
> --- a/drivers/media/usb/uvc/uvc_ctrl.c
> +++ b/drivers/media/usb/uvc/uvc_ctrl.c
> @@ -1589,7 +1589,12 @@ void uvc_ctrl_status_event(struct uvc_video_chain *chain,
> mutex_lock(&chain->ctrl_mutex);
>
> handle = ctrl->handle;
> - ctrl->handle = NULL;
> + if (handle) {
> + ctrl->handle = NULL;
> + WARN_ON(!handle->pending_async_ctrls);
> + if (handle->pending_async_ctrls)
> + handle->pending_async_ctrls--;
> + }

There's at least one other location where ctrl->handle is reset to NULL.

>
> list_for_each_entry(mapping, &ctrl->info.mappings, list) {
> s32 value = __uvc_ctrl_get_value(mapping, data);
> @@ -2050,8 +2055,11 @@ int uvc_ctrl_set(struct uvc_fh *handle,
> mapping->set(mapping, value,
> uvc_ctrl_data(ctrl, UVC_CTRL_DATA_CURRENT));
>
> - if (ctrl->info.flags & UVC_CTRL_FLAG_ASYNCHRONOUS)
> + if (ctrl->info.flags & UVC_CTRL_FLAG_ASYNCHRONOUS) {
> + if (!ctrl->handle)
> + handle->pending_async_ctrls++;
> ctrl->handle = handle;

Is this protected by ctrl_mutex ?

Please be careful about locking and race conditions, taking the time to
double check will help getting your patches merged faster.

> + }
>
> ctrl->dirty = 1;
> ctrl->modified = 1;
> @@ -2774,6 +2782,34 @@ int uvc_ctrl_init_device(struct uvc_device *dev)
> return 0;
> }
>
> +void uvc_ctrl_cleanup_fh(struct uvc_fh *handle)
> +{
> + struct uvc_entity *entity;
> +
> + guard(mutex)(&handle->chain->ctrl_mutex);
> +
> + if (!handle->pending_async_ctrls)
> + return;
> +
> + list_for_each_entry(entity, &handle->chain->dev->entities, list) {
> + int i;

unsigned int

I wonder if these days you could event write

for (unsigned int i = 0; i < entity->ncontrols; ++i) {

> +
> + for (i = 0; i < entity->ncontrols; ++i) {
> + struct uvc_control *ctrl = &entity->controls[i];
> +
> + if (!ctrl->handle || ctrl->handle != handle)

Given that handle can't be null, you can write

if (ctrl->handle != handle)

> + continue;
> +
> + ctrl->handle = NULL;
> + if (WARN_ON(!handle->pending_async_ctrls))
> + continue;

Is this needed ? If we find more controls for this handle than
pending_async_ctrls, decrementing it below 0 will case the WARN_ON() at
the end of this function to trigger, isn't that enough ?

> + handle->pending_async_ctrls--;
> + }
> + }
> +
> + WARN_ON(handle->pending_async_ctrls);
> +}
> +
> /*
> * Cleanup device controls.
> */
> diff --git a/drivers/media/usb/uvc/uvc_v4l2.c b/drivers/media/usb/uvc/uvc_v4l2.c
> index 97c5407f6603..b425306a3b8c 100644
> --- a/drivers/media/usb/uvc/uvc_v4l2.c
> +++ b/drivers/media/usb/uvc/uvc_v4l2.c
> @@ -652,6 +652,8 @@ static int uvc_v4l2_release(struct file *file)
>
> uvc_dbg(stream->dev, CALLS, "%s\n", __func__);
>
> + uvc_ctrl_cleanup_fh(handle);
> +
> /* Only free resources if this is a privileged handle. */
> if (uvc_has_privileges(handle))
> uvc_queue_release(&stream->queue);
> diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
> index 07f9921d83f2..2f8a9c48e32a 100644
> --- a/drivers/media/usb/uvc/uvcvideo.h
> +++ b/drivers/media/usb/uvc/uvcvideo.h
> @@ -612,6 +612,7 @@ struct uvc_fh {
> struct uvc_video_chain *chain;
> struct uvc_streaming *stream;
> enum uvc_handle_state state;
> + unsigned int pending_async_ctrls; /* Protected by ctrl_mutex. */

The kernel does it the other way around, it lists in the documentation
of the lock what data it protects.

> };
>
> struct uvc_driver {
> @@ -797,6 +798,8 @@ int uvc_ctrl_is_accessible(struct uvc_video_chain *chain, u32 v4l2_id,
> int uvc_xu_ctrl_query(struct uvc_video_chain *chain,
> struct uvc_xu_control_query *xqry);
>
> +void uvc_ctrl_cleanup_fh(struct uvc_fh *handle);
> +
> /* Utility functions */
> struct usb_host_endpoint *uvc_find_endpoint(struct usb_host_interface *alts,
> u8 epaddr);

--
Regards,

Laurent Pinchart