From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S942931AbdEYRFB (ORCPT ); Thu, 25 May 2017 13:05:01 -0400 Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:42124 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S936662AbdEYRFA (ORCPT ); Thu, 25 May 2017 13:05:00 -0400 Date: Thu, 25 May 2017 19:04:57 +0200 From: Pavel Machek To: Al Viro , mtk.manpages@gmail.com Cc: Dmitry Vyukov , Alexander Potapenko , Kostya Serebryany , Eric Dumazet , LKML Subject: Re: [PATCH] [iov_iter] use memmove() when copying to/from user page Message-ID: <20170525170457.GA3085@amd> References: <20170516122734.56760-1-glider@google.com> <20170516184821.GQ390@ZenIV.linux.org.uk> <20170516193750.GR390@ZenIV.linux.org.uk> <20170516205212.GS390@ZenIV.linux.org.uk> <20170516213347.GT390@ZenIV.linux.org.uk> <20170516224854.GU390@ZenIV.linux.org.uk> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="sdtB3X0nJg68CQEu" Content-Disposition: inline In-Reply-To: <20170516224854.GU390@ZenIV.linux.org.uk> User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org --sdtB3X0nJg68CQEu Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Hi! On Tue 2017-05-16 23:48:54, Al Viro wrote: > On Tue, May 16, 2017 at 03:15:16PM -0700, Dmitry Vyukov wrote: > > > Because it's not going to be *one* call of memcpy() or memmove(). It= 's > > > one per page. > >=20 > >=20 > > I missed that. > >=20 > > I assumed that in the case of sendfile from memfd to memfd data will > > be copied directly. But it goes through a pipe with multiple buffers. > > Does not look easily fixable. >=20 > Which leaves us only with "will nasal demons really fly there?". It seems so: Date: Tue, 16 May 2017 14:27:34 +0200 =46rom: Alexander Potapenko Subject: [PATCH] [iov_iter] use memmove() when copying to/from user page BUG: memcpy-param-overlap in generic_perform_write+0x551/0xa20 __msan_memcpy(ffff88013c6e3001, ffff88013c6e3000, 105) CPU: 0 PID: 1040 Comm: probe Not tainted 4.11.0-rc5+ #2562 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 At the very least, we do not want userland to trigger kernel BUG()s... so this needs fixes beyond documentation. > Consider the case of write() from an mmapped piece of file to overlapping > piece of the same file. It is possible and not hard to trigger; all we > can guarantee is the lack of infoleaks, filesystem corruption or memory > corruption. File *contents* in the affected area can't be sanely relied > upon. >=20 > This case is not different. BTW, neither SUS, nor our manpages for > write(2) mention these issues with mmap()-created aliases between the > source and destination. Added people doing documentation; seems like we have some updates to do there, too. Pavel --=20 (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blo= g.html --sdtB3X0nJg68CQEu Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iEYEARECAAYFAlknDrkACgkQMOfwapXb+vJ5ogCgn55812t+CW690AQzVk+ZgHo3 PZsAn27a6tNvDgv0K5zAl1ao6UoVDGyG =5Gsl -----END PGP SIGNATURE----- --sdtB3X0nJg68CQEu--