On Tue, Oct 26, 2021 at 11:55:05AM +0100, Kieran Bingham wrote:

Quoting Johan Hovold (2021-10-26 10:55:11)

...

Add the missing bulk-endpoint max-packet sanity check to probe() to avoid division by zero in uvc_alloc_urb_buffers() in case a malicious device has broken descriptors (or when doing descriptor fuzz testing).

Note that USB core will reject URBs submitted for endpoints with zero wMaxPacketSize but that drivers doing packet-size calculations still need to handle this (cf. commit 2548288b4fb0 ("USB: Fix: Don't skip endpoint descriptors with maxpacket=0")).

Fixes: c0efd232929c ("V4L/DVB (8145a): USB Video Class driver") Cc: stable@vger.kernel.org # 2.6.26 Signed-off-by: Johan Hovold johan@kernel.org

---

drivers/media/usb/uvc/uvc_video.c | 4 ++++ 1 file changed, 4 insertions(+)

diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c index e16464606b14..85ac5c1081b6 100644 --- a/drivers/media/usb/uvc/uvc_video.c +++ b/drivers/media/usb/uvc/uvc_video.c @@ -1958,6 +1958,10 @@ static int uvc_video_start_transfer(struct uvc_streaming *stream, if (ep == NULL) return -EIO;

•           /* Reject broken descriptors. */
  

•           if (usb_endpoint_maxp(&ep->desc) == 0)
  

•                   return -EIO;
  

Is there any value in identifying this with a specific return code like -ENODATA?

Going one step further, wouldn't it be better to fail probe() for those devices ?

But either way, this seems sane.

Reviewed-by: Kieran Bingham kieran.bingham+renesas@ideasonboard.com

...

•           ret = uvc_init_video_bulk(stream, ep, gfp_flags);
    }
  

Hi Johan,

On Tue, Oct 26, 2021 at 02:06:55PM +0200, Johan Hovold wrote:

On Tue, Oct 26, 2021 at 02:15:20PM +0300, Laurent Pinchart wrote:

...

On Tue, Oct 26, 2021 at 11:55:05AM +0100, Kieran Bingham wrote:

...

Quoting Johan Hovold (2021-10-26 10:55:11)

...

Add the missing bulk-endpoint max-packet sanity check to probe() to avoid division by zero in uvc_alloc_urb_buffers() in case a malicious device has broken descriptors (or when doing descriptor fuzz testing).

Note that USB core will reject URBs submitted for endpoints with zero wMaxPacketSize but that drivers doing packet-size calculations still need to handle this (cf. commit 2548288b4fb0 ("USB: Fix: Don't skip endpoint descriptors with maxpacket=0")).

Fixes: c0efd232929c ("V4L/DVB (8145a): USB Video Class driver") Cc: stable@vger.kernel.org # 2.6.26 Signed-off-by: Johan Hovold johan@kernel.org

---

drivers/media/usb/uvc/uvc_video.c | 4 ++++ 1 file changed, 4 insertions(+)

diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c index e16464606b14..85ac5c1081b6 100644 --- a/drivers/media/usb/uvc/uvc_video.c +++ b/drivers/media/usb/uvc/uvc_video.c @@ -1958,6 +1958,10 @@ static int uvc_video_start_transfer(struct uvc_streaming *stream, if (ep == NULL) return -EIO;

•           /* Reject broken descriptors. */
  

•           if (usb_endpoint_maxp(&ep->desc) == 0)
  

•                   return -EIO;
  

Is there any value in identifying this with a specific return code like -ENODATA?

Going one step further, wouldn't it be better to fail probe() for those devices ?

This is not how the driver works today. Look at the "return -EIO" just above in case the expected endpoint is missing. And similarly for the isochronous case for which zero wMaxPacket isn't handled until uvc_video_start_transfer() either (a few lines further up still).

That's a good point, it wouldn't be reasonable to ask you to fix all that :-)

Note however the copy-paste error in the commit message mentioning probe(), which is indeed where this would typically be handled.

Do you want me to resend or can you change

s/probe()/uvc_video_start_transfer()/

in the commit message when applying if you think this is acceptable as is?

I can fix this when applying.

Reviewed-by: Laurent Pinchart laurent.pinchart@ideasonboard.com