From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <binutils-return-70209-listarch-binutils=sources.redhat.com@sourceware.org>
Received: (qmail 658 invoked by alias); 18 Feb 2011 08:49:41 -0000
Received: (qmail 536 invoked by uid 22791); 18 Feb 2011 08:49:40 -0000
X-SWARE-Spam-Status: No, hits=-1.8 required=5.0	tests=AWL,BAYES_00,T_RP_MATCHES_RCVD
X-Spam-Check-By: sourceware.org
Received: from vpn.id2.novell.com (HELO vpn.id2.novell.com) (195.33.99.129)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Fri, 18 Feb 2011 08:49:38 +0000
Received: from EMEA1-MTA by vpn.id2.novell.com	with Novell_GroupWise; Fri, 18 Feb 2011 08:49:24 +0000
Message-Id: <4D5E40D80200007800032A0B@vpn.id2.novell.com>
Date: Fri, 18 Feb 2011 08:49:00 -0000
From: "Jan Beulich" <JBeulich@novell.com>
To: "Jakub Jelinek" <jakub@redhat.com>,"Jan Hubicka" <hubicka@ucw.cz>
Cc: "GCC Development" <gcc@gcc.gnu.org>,	 "H.J. Lu" <hjl.tools@gmail.com>,<x32-abi@googlegroups.com>,	 "Binutils" <binutils@sourceware.org>,	 "GNU C Library" <libc-alpha@sourceware.org>,	 "H. Peter Anvin" <hpa@zytor.com>
Subject: Re: x32 psABI draft version 0.2
References: <AANLkTindyH6koj0dCn1nxU9Cu3XUtg3XfvTnt_EWD13Q@mail.gmail.com> <4D5C2DD2.10608@zytor.com> <4D5CEBDE02000078000325A2@vpn.id2.novell.com> <20110217142916.GI13037@sunsite.ms.mff.cuni.cz> <20110217152233.GB11346@atrey.karlin.mff.cuni.cz> <AANLkTin=LRV7F88w=7cKHovpOh_BnEJE9QqA_bWtOJN2@mail.gmail.com> <20110217154452.GA18799@kam.mff.cuni.cz> <20110217180646.GB30899@tyan-ft48-01.lab.bos.redhat.com> <20110217224956.GA20055@kam.mff.cuni.cz> <20110217230727.GK30899@tyan-ft48-01.lab.bos.redhat.com>
In-Reply-To: <20110217230727.GK30899@tyan-ft48-01.lab.bos.redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
X-IsSubscribed: yes
Mailing-List: contact binutils-help@sourceware.org; run by ezmlm
Precedence: bulk
List-Id: <binutils.sourceware.org>
List-Subscribe: <mailto:binutils-subscribe@sourceware.org>
List-Archive: <http://sourceware.org/ml/binutils/>
List-Post: <mailto:binutils@sourceware.org>
List-Help: <mailto:binutils-help@sourceware.org>, <http://sourceware.org/ml/#faqs>
Sender: binutils-owner@sourceware.org
X-SW-Source: 2011-02/txt/msg00240.txt.bz2

>>> On 18.02.11 at 00:07, Jakub Jelinek <jakub@redhat.com> wrote:
> So one way to cut down the size of .rela.dyn section would be a relocation
> like
> R_X86_64_RELATIVE_BLOCK where applying such a relocation with r_offset O =
and
> r_addend N would be:
> uint64_t *ptr =3D O;
> for (i =3D 0; i < N; i++)
>   ptr[i] +=3D bias;
> Then e.g.
> 0000003ec6d86008  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5aef3f3
> 0000003ec6d86010  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5af92f6
> 0000003ec6d86018  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b06d17
> 0000003ec6d86020  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b1dc5f
> 0000003ec6d86028  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b1edaf
> 0000003ec6d86030  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b27358
> 0000003ec6d86038  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b30f9f
> 0000003ec6d86040  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b3317d
> 0000003ec6d86048  0000000000000008 R_X86_64_RELATIVE=20=20=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000003ec5b34479
> could be represented as:
> 0000003ec6d86008  00000000000000MN R_X86_64_RELATIVE_BLOCK=20=20=20=20=20=
=20=20=20=20=20=20=20=20=20=20=20=20=20=20
>       0000000000000009
> I see many hundreds of consecutive R_X86_64_RELATIVE relocs in libxul.so,=
=20
> though
> of course it would need much better analysis over larger body of code.
>=20
> In most programs if the library is prelinked all relative relocs are skip=
ped
> and .rela.dyn for them doesn't need to be even paged in, but Mozilla is=20
> quite
> special in that it one of the most common security relevant packages and=
=20
> thus
> wants randomization, but is linked against huge libraries, so the questio=
n=20
> is
> if Mozilla is the right candidate to drive our decisions on.
>=20
> Another alternative to compress relative relocations would be an indirect
> relative relocation, which would give you in r_offset address of a block =
of=20
> addresses
> and r_addend the size of that block, and the block would just contain=20
> offsets
> on which words need to be +=3D bias.  Then, instead of changing RELA to R=
EL to
> save 8 bytes from 24 you'd save 16 bytes from those 24 (well, for x32 hal=
f=20
> of that).

For relocations where the relocated field is large enough, considering
chained relocations (as seen in NetWare NLMs) would also be a
possibility, i.e. r_offset specifies just the first relocation that all need
the same addend (and eventual other properties), and the relocated
field holds the r_offset of the next field to be relocated.

Jan