Buffer overflow in uvc-video.

Buffer overflow in uvc-video.

[Ralph Loader]

Hi,

There appears to be a buffer overflow in
drivers/media/video/uvc/uvc_ctrl.c:

Sep 20 17:53:12 i kernel: BUG kmalloc-8 (Not tainted): Redzone
overwritten Sep 20 17:53:12 i kernel:
--------------------------------------------------------------------
--------- Sep 20 17:53:12 i kernel:
Sep 20 17:53:12 i kernel: INFO: 0xf2c5ce08-0xf2c5ce0b. First byte 0xa1
instead of 0xcc 
Sep 20 17:53:12 i kernel: INFO: Allocated in
uvc_query_v4l2_ctrl+0x3c/0x239 [uvcvideo] age=13 cpu=1 pid=4975
...


A fixed size 8-byte buffer is allocated, and a variable size field is
read into it; there is no particular bound on the size of the field
(it is dependent on hardware and configuration) and it can overflow
[also verified by inserting printk's.]

The patch below attempts to size the buffer to the correctly; I do not
know if sufficient validation of that size is carried out.

Cheers,
Ralph Loader.

diff --git a/drivers/media/video/uvc/uvc_ctrl.c
b/drivers/media/video/uvc/uvc_ct rl.c
index 6ef3e52..feab12a 100644
--- a/drivers/media/video/uvc/uvc_ctrl.c
+++ b/drivers/media/video/uvc/uvc_ctrl.c
@@ -592,7 +592,7 @@ int uvc_query_v4l2_ctrl(struct uvc_video_device
*video, if (ctrl == NULL)
                return -EINVAL;
 
-       data = kmalloc(8, GFP_KERNEL);
+       data = kmalloc(ctrl->info->size, GFP_KERNEL);
        if (data == NULL)
                return -ENOMEM;
 

Re: [Linux-uvc-devel] Buffer overflow in uvc-video.

[Laurent Pinchart]

Hi Ralph,

On Saturday 20 September 2008, Ralph Loader wrote:
> Hi,
>
> There appears to be a buffer overflow in
> drivers/media/video/uvc/uvc_ctrl.c:
>
> Sep 20 17:53:12 i kernel: BUG kmalloc-8 (Not tainted): Redzone
> overwritten Sep 20 17:53:12 i kernel:
> --------------------------------------------------------------------
> --------- Sep 20 17:53:12 i kernel:
> Sep 20 17:53:12 i kernel: INFO: 0xf2c5ce08-0xf2c5ce0b. First byte 0xa1
> instead of 0xcc
> Sep 20 17:53:12 i kernel: INFO: Allocated in
> uvc_query_v4l2_ctrl+0x3c/0x239 [uvcvideo] age=13 cpu=1 pid=4975
> ...
>
>
> A fixed size 8-byte buffer is allocated, and a variable size field is
> read into it; there is no particular bound on the size of the field
> (it is dependent on hardware and configuration) and it can overflow
> [also verified by inserting printk's.]

Thanks for catching this.

> The patch below attempts to size the buffer to the correctly; I do not
> know if sufficient validation of that size is carried out.

UVC control sizes come either from the predefined controls (uvc_ctrls in 
uvc_ctrl.c) or from user-defined controls.

Predefined controls size doesn't need any validation. User-defined controls 
size should be validated, but defining a new UVC control from userspace 
requires root privileges. This currently mitigates the issue. I'll submit a 
patch with an upper bound check on the size.

> Cheers,
> Ralph Loader.
>
> diff --git a/drivers/media/video/uvc/uvc_ctrl.c
> b/drivers/media/video/uvc/uvc_ct rl.c
> index 6ef3e52..feab12a 100644
> --- a/drivers/media/video/uvc/uvc_ctrl.c
> +++ b/drivers/media/video/uvc/uvc_ctrl.c
> @@ -592,7 +592,7 @@ int uvc_query_v4l2_ctrl(struct uvc_video_device
> *video, if (ctrl == NULL)
>                 return -EINVAL;
>
> -       data = kmalloc(8, GFP_KERNEL);
> +       data = kmalloc(ctrl->info->size, GFP_KERNEL);
>         if (data == NULL)
>                 return -ENOMEM;

Laurent Pinchart