[Bug gas/31115] [ARM] The minimalistic DWARF DIE for function has wrong address in Thumb mode

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=31115

--- Comment #4 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> 
---
Created attachment 15249
  --> https://sourceware.org/bugzilla/attachment.cgi?id=15249&action=edit
ELF files demonstrating the issue

The testcase I'm using is GDB's pr25124.S, built with an
arm-linux-gnueabihf toolchain:

$ cat ~/src/binutils-gdb/gdb/testsuite/gdb.arch/pr25124.S
/* Test proper disassembling of ARM thumb instructions when reloading a symbol
   file.

   Copyright 2012-2023 Free Software Foundation, Inc.

   This file is part of GDB.

   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, see <http://www.gnu.org/licenses/>.  */

        .syntax unified
        .thumb
        .text
        .p2align 2
        .global main
        .thumb
        .thumb_func
        .type main, %function
main:
        bx      pc
        nop
.code 32
        mov     r0, #0
        bx      lr
        .size   main, .-main
        .section        .note.GNU-stack,"",%progbits
$ gcc -g -o pr25124 ~/src/binutils-gdb/gdb/testsuite/gdb.arch/pr25124.S
$ gdb pr25124
Reading symbols from pr25124...
(gdb) x/i main
   0x103e5 <main>:      bx      pc
(gdb) x/i main+8
   0x103ed <main+7>:    vrhadd.u16      d14, d14, d31
(gdb) quit

Interestingly, as can be seen above "x/i main" actually works fine.
it's "x/i main+8" that breaks. This is a detail that I just noticed.

There's another thing I just discovered : I can reproduce GDB's bad
behavior on an ELF executable (produced by GCC from the .S file), but
not on a .o file produced directly by gas:

$ as -g -o pr25124.o ~/src/binutils-gdb/gdb/testsuite/gdb.arch/pr25124.S
$ gdb pr25124.o
Reading symbols from pr25124.o...
(gdb) x/i main
   0x0 <main>:  bx      pc
(gdb) x/i main+8
   0x8 <main+8>:        bx      lr
(gdb) quit

Both the executable and the .o file have the LSB bit set in main's
DW_AT_low_pc:

Object file:

 <1><26>: Abbrev Number: 2 (DW_TAG_subprogram)
    <27>   DW_AT_name        : (strp) (offset: 0x97): main
    <2b>   DW_AT_external    : (flag) 1
    <2c>   DW_AT_low_pc      : (addr) 0x1
    <30>   DW_AT_high_pc     : (addr) 0xd

Executable:

 <1><24>: Abbrev Number: 2 (DW_TAG_subprogram)
    <25>   DW_AT_name        : (strp) (offset: 0x97): main
    <29>   DW_AT_external    : (flag_present) 1
    <29>   DW_AT_type        : (ref_udata) <0x2f>
    <2a>   DW_AT_low_pc      : (addr) 0x103e5
    <2e>   DW_AT_high_pc     : (udata) 12
 <1><2f>: Abbrev Number: 3 (DW_TAG_unspecified_type)
 <1><30>: Abbrev Number: 0

So for some reason GDB is fine with an object file containing the wrong
DW_AT_low_pc in main's DIE, but not when it's with a "full blown"
executable.

I'm attaching the files I generated with a toolchain that uses binutils
from the commit immediately before the one that introduced this
behavior, and with a toolchain that uses a recent commit from trunk.

Any toolchain from a distro that uses binutils >= 2.39 should be enough
to demonstrate the problem. I was able to check with Ubuntu 23.04's
gcc-arm-linux-gnueabihf for example, which uses binutils 2.40.

-- 
You are receiving this mail because:
You are on the CC list for the bug.