From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
        id S1753992AbcITUic (ORCPT <rfc822;w@1wt.eu>);
        Tue, 20 Sep 2016 16:38:32 -0400
Received: from zeniv.linux.org.uk ([195.92.253.2]:60180 "EHLO
        ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1751783AbcITUib (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Tue, 20 Sep 2016 16:38:31 -0400
Date: Tue, 20 Sep 2016 21:38:21 +0100
From: Al Viro <viro@ZenIV.linux.org.uk>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>,
        Martin Schwidefsky <schwidefsky@de.ibm.com>,
        Jan Stancek <jstancek@redhat.com>, Arnd Bergmann <arnd@arndb.de>,
        Greg Ungerer <gerg@uclinux.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] fix fault_in_multipages_...() on architectures with
 no-op access_ok()
Message-ID: <20160920203821.GQ2356@ZenIV.linux.org.uk>
References: <57E131E6.1090507@redhat.com>
 <20160920150657.GN2356@ZenIV.linux.org.uk>
 <570490469.234828.1474391501934.JavaMail.zimbra@redhat.com>
 <20160920190742.GP2356@ZenIV.linux.org.uk>
 <CA+55aFx-NuAZS9M0_UUwbMU91VnkG+gxLtqX-uWJQAYrh1Vq6w@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+55aFx-NuAZS9M0_UUwbMU91VnkG+gxLtqX-uWJQAYrh1Vq6w@mail.gmail.com>
User-Agent: Mutt/1.6.1 (2016-04-27)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Tue, Sep 20, 2016 at 01:24:25PM -0700, Linus Torvalds wrote:

> Quite frankly, I think it is access_ok() that should be fixed for s390.
> 
> A wrapping user access is *not* ok, not even if kernel and user memory
> are separate.
> 
> It is insane to make fault_in_multipages..() return EFAULT if a normal
> wrapping user access wouldn't. So the fix is not to change
> fault_in_multipage_xyz, but to make sure any op that tries to wrap
> will properly return EFAULT.

Not the point.  Of course it *would* fail; the problem is that the loop
that would ping each page is never executed.  What happens is
	while (uaddr <= end) 
		touch uaddr
		uaddr += PAGE_SIZE
	if uaddr and end point to different pages
		ping end

What happens if uaddr is greater than end, thanks to wraparound?  Right,
we skip the loop entirely and all we do is one ping of the end.  Which
might very well succeed, leaving us with false positive.