On 12/10/19 4:15 AM, Chuhong Yuan wrote:
> allegro_open() misses a check for v4l2_m2m_ctx_init().
> Add a check and error handling code to fix it.
> 
> Fixes: f20387dfd065 ("media: allegro: add Allegro DVT video IP core driver")
> Signed-off-by: Chuhong Yuan <hsleste...@gmail.com>
> ---
> Changes in v2:
>   - Fix the use-after-free in v1.
> 
>  drivers/staging/media/allegro-dvt/allegro-core.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/staging/media/allegro-dvt/allegro-core.c 
> b/drivers/staging/media/allegro-dvt/allegro-core.c
> index 6f0cd0784786..66736beb67af 100644
> --- a/drivers/staging/media/allegro-dvt/allegro-core.c
> +++ b/drivers/staging/media/allegro-dvt/allegro-core.c
> @@ -2270,6 +2270,7 @@ static int allegro_open(struct file *file)
>       struct allegro_channel *channel = NULL;
>       struct v4l2_ctrl_handler *handler;
>       u64 mask;
> +     int ret;
>  
>       channel = kzalloc(sizeof(*channel), GFP_KERNEL);
>       if (!channel)
> @@ -2341,6 +2342,14 @@ static int allegro_open(struct file *file)
>       channel->fh.m2m_ctx = v4l2_m2m_ctx_init(dev->m2m_dev, channel,
>                                               allegro_queue_init);
>  
> +     if (IS_ERR(channel->fh.m2m_ctx)) {
> +             ret = PTR_ERR(channel->fh.m2m_ctx);
> +             v4l2_fh_del(&channel->fh);
> +             v4l2_fh_exit(&channel->fh);

Just move the v4l2_fh_init/add calls to just before the return 0, i.e. when 
everything
is right. Then you don't need to call del/exit on error.

Also, I see that the result of all the v4l2_ctrl_new* calls isn't checked.

Just after the last v4l2_ctrl_new_std() call you need to check handler->error:
if not 0, then there was an error and you need to call v4l2_ctrl_handler_free
to clean it up.

Regards,

        Hans

> +             kfree(channel);
> +             return ret;
> +     }
> +
>       return 0;
>  }
>  
> 

_______________________________________________
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Reply via email to