Re: [gold] First patch for aarch64 backend

[Jing Yu <jingyu at google dot com>]

Yes, ILP32 support may be interesting for ChromeOS and Android.

Given the limited resources, our team have to focus on 64bit now and
leave ILP32 for other developers who are interested to contribute.
We will try our best to consider ILP32 varieties when implementing LP64.

Thanks,
Jing

On Tue, Jul 1, 2014 at 4:09 PM, Andrew Pinski <pinskia@gmail.com> wrote:
> On Tue, Jul 1, 2014 at 4:06 PM, Jing Yu <jingyu@google.com> wrote:
>> On Tue, Jul 1, 2014 at 3:15 PM, Yufeng Zhang <yufengzhangdev@gmail.com> wrote:
>>> On 26 June 2014 00:46, Jing Yu <jingyu@google.com> wrote:
>>>>
>>>> Hi,
>>>>
>>>> We start to work on gold aarch64 backend. Here is the first patch.
>>>> This patch is just a skeleton which almost does nothing.
>>>> The change that enables gold for aarch64 in top level configure was
>>>> removed from this patch.
>>>>
>>>> Tested:
>>>> 1) On x86_64,
>>>> ${binutils}/configure --build=x86_64-unknown-linux-gnu
>>>> --host=x86_64-unknown-linux-gnu --target=x86_64-unknown-linux-gnu
>>>> --enable-plugins --disable-multilib --disable-nls --enable-threads
>>>> --enable-gold=default --enable-targets=all
>>>> $make all-gold
>>>>
>>>> Also built with -O1 and -O0.
>>>>
>>>> 2) On aarch64,
>>>> gcc version 4.8.2 (Ubuntu/Linaro 4.8.2-19ubuntu1)
>>>> $ ../binutils-gdb/configure --enable-plugins --disable-multilib
>>>> --disable-nls --enable-threads --enable-gold=default
>>>> --enable-targets=all
>>>> $make all-gold
>>>> Also built with -O1 and -O0.
>>>>
>>>> The generated native gold/ld-new is able to read hello.o and libc.so.
>>>>
>>>> Thanks,
>>>> Jing
>>>>
>>>> diff --git a/elfcpp/aarch64.h b/elfcpp/aarch64.h
>>>> new file mode 100644
>>>> index 0000000..09bc498
>>>> --- /dev/null
>>>> +++ b/elfcpp/aarch64.h
>>>> @@ -0,0 +1,183 @@
>>>> +// aarch64.h -- ELF definitions specific to AARCH64  -*- C++ -*-
>>>> +
>>>> +// Copyright (C) 2012-2014 Free Software Foundation, Inc.
>>>> +// Written by Jing Yu (jingyu@google.com)
>>>> +
>>>> +// This file is part of elfcpp.
>>>> +
>>>> +// This program is free software; you can redistribute it and/or
>>>> +// modify it under the terms of the GNU Library General Public License
>>>> +// as published by the Free Software Foundation; either version 2, or
>>>> +// (at your option) any later version.
>>>> +
>>>> +// In addition to the permissions in the GNU Library General Public
>>>> +// License, the Free Software Foundation gives you unlimited
>>>> +// permission to link the compiled version of this file into
>>>> +// combinations with other programs, and to distribute those
>>>> +// combinations without any restriction coming from the use of this
>>>> +// file.  (The Library Public License restrictions do apply in other
>>>> +// respects; for example, they cover modification of the file, and
>>>> +/// distribution when not linked into a combined executable.)
>>>> +
>>>> +// This program is distributed in the hope that it will be useful, but
>>>> +// WITHOUT ANY WARRANTY; without even the implied warranty of
>>>> +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>>>> +// Library General Public License for more details.
>>>> +
>>>> +// You should have received a copy of the GNU Library General Public
>>>> +// License along with this program; if not, write to the Free Software
>>>> +// Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA
>>>> +// 02110-1301, USA.
>>>> +
>>>> +#ifndef ELFCPP_AARCH64_H
>>>> +#define ELFCPP_AARCH64_H
>>>> +
>>>> +namespace elfcpp
>>>> +{
>>>> +
>>>> +// The relocation type information is taken from:
>>>> +//
>>>> +//   ELF for the ARM 64-bit Architecture (AArch64)
>>>> +//   Document number: ARM IHI 0056B, current through AArch64 ABI release 1.0
>>>> +//   Date of Issue: 22nd May, 2013
>>>> +//
>>>
>>> Is there a particular reason that you didnât choose the ABI release
>>> 1.1 which has the support for ELF32, including the ELF32 relocation
>>> codes?
>>>
>> We don't have a plan to support ELF32 in anytime soon.
>> The document that describes ELF32 relocations is a beta proposal.ELF_PLATFORM
>
> Why not?  The Linux kernel support for ILP32 is in review right now.
> And I will be submitting the glibc support right after the kernel
> support goes in.
>
> They are only beta because ARM does not want to commit to them until
> an implementation is upstream.  Yes it is a catch 22 though.
>
> Thanks,
> Andrew Pinski
>
>
>>
>>>> +
>>>> +enum
>>>> +{
>>>> +  // Null relocation codes
>>>> +  R_AARCH64_NONE = 0, // None
>>>> +  withdrawn = 256, // Treat as R_AARCH64_NONE
>>>> +
>>>> +  // Static relocations
>>>> +  R_AARCH64_ABS64 = 257, // S + A
>>>> +  R_AARCH64_ABS32 = 258, // S + A
>>>> +  R_AARCH64_ABS16 = 259, // S + A
>>>> +  R_AARCH64_PREL64 = 260, // S + A - P
>>>> +  R_AARCH64_PREL32 = 261, // S + A - P
>>>> +  R_AARCH64_PREL16 = 262, // S + A - P
>>>> +  R_AARCH64_MOVW_UABS_G0 = 263, // S + A
>>>> +  R_AARCH64_MOVW_UABS_G0_NC = 264, // S + A
>>>> +  R_AARCH64_MOVW_UABS_G1 = 265, // S + A
>>>
>>> [snip]
>>>
>>>> diff --git a/gold/configure.tgt b/gold/configure.tgt
>>>> index 93d622d..eaf164b 100644
>>>> --- a/gold/configure.tgt
>>>> +++ b/gold/configure.tgt
>>>> @@ -144,6 +144,14 @@ arm*-*-*)
>>>>   targ_big_endian=false
>>>>   targ_extra_big_endian=true
>>>>   ;;
>>>> +aarch64-*)
>>>
>>> I think you'll need aarch64*-* in order to support targ_extra_size=32
>>> and targ_extra_big_endian=true.
>>>
>> Will do. Thanks!
>> Jing
>>
>>> Yufeng