public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: liuzhensong <liuzhensong@loongson.cn>
To: Fangrui Song <maskray@google.com>
Cc: binutils@sourceware.org, i.swmail@xen0n.name, xry111@xry111.site,
	chenhuacai@loongson.cn
Subject: Re: [PATCH] LoongArch:ld: Fix relocation error of pcrel.
Date: Sat, 3 Sep 2022 15:12:50 +0800	[thread overview]
Message-ID: <a2119648-5181-4896-6a3c-b68b62f67bb0@loongson.cn> (raw)
In-Reply-To: <CAFP8O3Jkvt==05z7=gPqLZw1UL6TcQGpcSn9dHOPK7cybnnLwQ@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 5761 bytes --]


在 2022/9/3 下午12:33, Fangrui Song 写道:
> On Fri, Sep 2, 2022 at 6:29 PM liuzhensong <liuzhensong@loongson.cn> wrote:
>>    Patch for branch 2.39.
>>    Need to reduce the address of pc when using
>>    reloction R_LARCH_SOP_PUSH_PLT_PCREL.
>>
>>    bfd/
>>      * elfnn-loongarch.c
>> ---
>>   bfd/elfnn-loongarch.c                         |  3 +-
>>   ld/testsuite/ld-loongarch-elf/pcrel-const.d   | 14 +++++++
>>   ld/testsuite/ld-loongarch-elf/pcrel-const.lds | 14 +++++++
>>   ld/testsuite/ld-loongarch-elf/pcrel-const.s   | 11 ++++++
>>   ld/testsuite/ld-loongarch-elf/pr.exp          | 39 +++++++++++++++++++
>>   5 files changed, 80 insertions(+), 1 deletion(-)
>>   create mode 100644 ld/testsuite/ld-loongarch-elf/pcrel-const.d
>>   create mode 100644 ld/testsuite/ld-loongarch-elf/pcrel-const.lds
>>   create mode 100644 ld/testsuite/ld-loongarch-elf/pcrel-const.s
>>   create mode 100644 ld/testsuite/ld-loongarch-elf/pr.exp
>>
>> diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
>> index 5b44901b9e0..fafdc7c7458 100644
>> --- a/bfd/elfnn-loongarch.c
>> +++ b/bfd/elfnn-loongarch.c
>> @@ -2341,9 +2341,10 @@ loongarch_elf_relocate_section (bfd *output_bfd, struct bfd_link_info *info,
>>          case R_LARCH_SOP_PUSH_PLT_PCREL:
>>            unresolved_reloc = false;
>>
>> -         if (resolved_to_const)
>> +         if (!is_undefweak && resolved_to_const)
>>              {
>>                relocation += rel->r_addend;
>> +             relocation -= pc;
>>                break;
>>              }
>>            else if (is_undefweak)
>> diff --git a/ld/testsuite/ld-loongarch-elf/pcrel-const.d b/ld/testsuite/ld-loongarch-elf/pcrel-const.d
>> new file mode 100644
>> index 00000000000..72abb204122
>> --- /dev/null
>> +++ b/ld/testsuite/ld-loongarch-elf/pcrel-const.d
>> @@ -0,0 +1,14 @@
>> +#as: -mla-global-with-pcrel
>> +#objdump: -Drsz
>> +
>> +.*:[    ]+file format .*
>> +
>> +
>> +Disassembly of section .text:
>> +
>> +.* <foo>:
>> +#...
>> +[      ]+4:[   ]+02c03084[     ]+addi.d[       ]+\$a0,[        ]+\$a0,[        ]+12\(0xc\)
>> +#...
>> +0+10 <__sec_end>:
>> +#pass
>> diff --git a/ld/testsuite/ld-loongarch-elf/pcrel-const.lds b/ld/testsuite/ld-loongarch-elf/pcrel-const.lds
>> new file mode 100644
>> index 00000000000..064a6894b4d
>> --- /dev/null
>> +++ b/ld/testsuite/ld-loongarch-elf/pcrel-const.lds
>> @@ -0,0 +1,14 @@
>> +ENTRY(foo);
>> +SECTIONS
>> +{
>> +       .text : ALIGN(4) {
>> +               *(.text*)
>> +       }
>> +
>> +       .data : ALIGN(4) {
>> +               __sec_start = .;
>> +               *(.gzdata)
>> +               __sec_end = .;
>> +       }
>> +}
>> +PROVIDE(__sec_size = __sec_end - 4);
> I do not know the legacy relocation scheme, but for tests I would try
> removing noise like ALIGN(4). .data output section containing .gzdata
> input sections looks strange too.
> Can .data input sections be used instead?

The __sec_end segment cannot be dumped using .data.

t:     file format elf64-loongarch


Disassembly of section .text:

0000000000000000 <foo>:
    0:    03400000     andi            $zero, $zero, 0x0
    4:    1c000004     pcaddu12i       $a0, 0
    8:    02c06084     addi.d          $a0, $a0, 24(0x18)
    c:    24000084     ldptr.w         $a0, $a0, 0
   10:    4c000020     jirl            $zero, $ra, 0

Disassembly of section .data:

0000000000000014 <__sec_start>:
   14:    00000001     0x00000001
>
> `__sec_size = __sec_end - 4` is unusual. Is `- 4` necessary to
> reproduce the bug? If necessary, add a comment.
Unnecessary, using "nop" before pcaddu12i instead.
>
>> diff --git a/ld/testsuite/ld-loongarch-elf/pcrel-const.s b/ld/testsuite/ld-loongarch-elf/pcrel-const.s
>> new file mode 100644
>> index 00000000000..bc4fb979a80
>> --- /dev/null
>> +++ b/ld/testsuite/ld-loongarch-elf/pcrel-const.s
>> @@ -0,0 +1,11 @@
>> +       .text
>> +       .align  2
>> +       .globl  foo
>> +       .type   foo, @function
>> +foo:
>> +       la.global       $r4,__sec_size
>> +       ldptr.w $r4,$r4,0
>> +       jr      $r1
>> +       .size   foo, .-foo
>> +       .data
>> +       .word 1
>> diff --git a/ld/testsuite/ld-loongarch-elf/pr.exp b/ld/testsuite/ld-loongarch-elf/pr.exp
>> new file mode 100644
>> index 00000000000..c0cc2fc30d2
>> --- /dev/null
>> +++ b/ld/testsuite/ld-loongarch-elf/pr.exp
>> @@ -0,0 +1,39 @@
>> +# Expect script for LoongArch ELF linker tests
>> +#   Copyright (C) 2022 Free Software Foundation, Inc.
>> +#
>> +# This file is part of the GNU Binutils.
>> +#
>> +# This program is free software; you can redistribute it and/or modify
>> +# it under the terms of the GNU General Public License as published by
>> +# the Free Software Foundation; either version 3 of the License, or
>> +# (at your option) any later version.
>> +#
>> +# 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 General Public License for more details.
>> +#
>> +# You should have received a copy of the GNU 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.
>> +#
>> +
>> +if ![istarget loongarch64-*-*] {
>> +  return
>> +}
>> +
>> +set link_tests [list \
>> +  [list \
>> +    "pcrel const" \
>> +    "-T pcrel-const.lds" "" \
>> +    "-mla-global-with-pcrel" \
>> +    { pcrel-const.s } \
>> +    [list \
>> +      [list objdump -D pcrel-const.d] \
>> +    ] \
>> +    "pcrel-const" \
>> +  ] \
>> +]
>> +
>> +run_ld_link_tests $link_tests
>> --
>> 2.31.1
>>

      reply	other threads:[~2022-09-03  7:12 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-03  1:28 liuzhensong
2022-09-03  4:33 ` Fangrui Song
2022-09-03  7:12   ` liuzhensong [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=a2119648-5181-4896-6a3c-b68b62f67bb0@loongson.cn \
    --to=liuzhensong@loongson.cn \
    --cc=binutils@sourceware.org \
    --cc=chenhuacai@loongson.cn \
    --cc=i.swmail@xen0n.name \
    --cc=maskray@google.com \
    --cc=xry111@xry111.site \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).