Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751293AbaGZKpH (ORCPT ); Sat, 26 Jul 2014 06:45:07 -0400 Received: from mail-wg0-f48.google.com ([74.125.82.48]:53403 "EHLO mail-wg0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750940AbaGZKpG (ORCPT ); Sat, 26 Jul 2014 06:45:06 -0400 Message-ID: <53D386A9.9040001@gmail.com> Date: Sat, 26 Jul 2014 11:44:57 +0100 From: Malcolm Priestley User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.6.0 MIME-Version: 1.0 To: =?ISO-8859-1?Q?Guillaume_CL=C9MENT?= CC: Forest Bond , Greg Kroah-Hartman , devel@driverdev.osuosl.org, linux-kernel@vger.kernel.org, Dan Carpenter Subject: Re: [PATCH] staging: vt6655: fix direct dereferencing of user pointer References: <20140725123301.GZ13737@mwanda> <1406292443-11734-1-git-send-email-gclement@baobob.org> <53D2E3BD.1000208@gmail.com> <20140726082429.GA3121@pleinair.baobob.org> <20140726091843.GA5956@pleinair.baobob.org> In-Reply-To: <20140726091843.GA5956@pleinair.baobob.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 26/07/14 10:18, Guillaume CL?MENT wrote: > On Sat, Jul 26, 2014 at 10:24:30AM +0200, Guillaume CL?MENT wrote: >> Hi Malcolm, >> >> On Sat, Jul 26, 2014 at 12:09:49AM +0100, Malcolm Priestley wrote: >>> Hi Guillaume >>> >>> On 25/07/14 13:47, Guillaume Clement wrote: >>>> Sparse reported that the data from tagSCmdRequest is given by >>>> userspace, so it should be tagged as such. >>> extra is not in user space >>> >> >> All right. >> >> This is still confusing to me because, taking the SIOCSIWGENIE ioctl as >> an example, in device_main.c, we have this code: >> >> rc = iwctl_siwgenie(dev, NULL, &(wrq->u.data), wrq->u.data.pointer); >> >> Here the extra parameter is the last one, wrq->u.data.pointer. >> >> I was led to believe that wrq->u.data.pointer is in userspace (this was >> reported by sparse actually) because the pointer field in data is >> actually defined as __user. >> >> > By the way, the original code (before my patch) reads: > > if ((wrq->length < 2) || (extra[1]+2 != wrq->length)) { > ret = -EINVAL; > goto out; > } > if (wrq->length > MAX_WPA_IE_LEN) { > ret = -ENOMEM; > goto out; > } > memset(pMgmt->abyWPAIE, 0, MAX_WPA_IE_LEN); > if (copy_from_user(pMgmt->abyWPAIE, extra, wrq->length)) { > ret = -EFAULT; > goto out; > } > > Note extra[1] and later copy_from_user(x, extra, y). > > If extra is not in userspace, we should not call copy_from_user, and if > it is, we should not dereference it. There is definitely something fishy > here. > I got it wrong when the iw_handler is not present a standard ioctl is called extra is in userspace. Sorry for the noise. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/