From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:31453 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751659AbaIRMXy (ORCPT ); Thu, 18 Sep 2014 08:23:54 -0400 Date: Thu, 18 Sep 2014 15:23:36 +0300 From: Dan Carpenter To: Guennadi Liakhovetski Cc: Mauro Carvalho Chehab , linux-media@vger.kernel.org, kernel-janitors@vger.kernel.org Subject: [patch] [media] mx2-camera: potential negative underflow bug Message-ID: <20140918122336.GA13147@mwanda> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Sender: linux-media-owner@vger.kernel.org List-ID: My static checker complains: drivers/media/platform/soc_camera/mx2_camera.c:1070 mx2_emmaprp_resize() warn: no lower bound on 'num' The heuristic is that it's looking for values which the user can influence and we put an upper bound on them but we (perhaps accidentally) allow negative numbers. I am not very familiar with this code but I have looked at it and think there might be a bug. Making the variable unsigned seems like a safe option either way and this silences the static checker warning. The call tree is: -> subdev_do_ioctl() -> mx2_camera_set_fmt() -> mx2_emmaprp_resize() The check: if (num > RESIZE_NUM_MAX) can underflow and then we use "num" on the else path. Signed-off-by: Dan Carpenter diff --git a/drivers/media/platform/soc_camera/mx2_camera.c b/drivers/media/platform/soc_camera/mx2_camera.c index b40bc2e..bc27a47 100644 --- a/drivers/media/platform/soc_camera/mx2_camera.c +++ b/drivers/media/platform/soc_camera/mx2_camera.c @@ -1003,7 +1003,7 @@ static int mx2_emmaprp_resize(struct mx2_camera_dev *pcdev, struct v4l2_mbus_framefmt *mf_in, struct v4l2_pix_format *pix_out, bool apply) { - int num, den; + unsigned int num, den; unsigned long m; int i, dir;