Bug 29450 - run-low_high_pc.sh fails on i386 against binutils-2.39
Summary: run-low_high_pc.sh fails on i386 against binutils-2.39
Status: RESOLVED MOVED
Alias: None
Product: elfutils
Classification: Unclassified
Component: libdw (show other bugs)
Version: unspecified
: P2 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-08-06 06:34 UTC by Sergei Trofimovich
Modified: 2022-08-11 09:17 UTC (History)
4 users (show)

See Also:
Host:
Target:
Build:
Last reconfirmed:


Attachments
low_high_pc (22.60 KB, application/x-executable)
2022-08-06 06:47 UTC, Sergei Trofimovich
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Sergei Trofimovich 2022-08-06 06:34:23 UTC
binutils-2.39 is out: https://sourceware.org/pipermail/binutils/2022-August/122246.html

I tried to package it in nixpkgs and found a few build or test failures.

elfutils-0.187 and elfutils from git both fail run-low_high_pc.sh as:

FAIL: run-low_high_pc.sh
========================

[b] main.c
[2d] main

[b] ../sysdeps/i386/start.S
[26] _start
[40] ../sysdeps/x86/abi-note.c
[b52] init.c
[b8e] static-reloc.c
[2dba] _dl_relocate_static_pie
[2dd8] ../sysdeps/i386/crti.S
[2def] _init
lowpc: 8049000, highpc: 8049000lx
../sysdeps/i386/crti.S: [2def] '_init' highpc <= lowpc
FAIL run-low_high_pc.sh (exit status: 255)

Extra `lx` prefix looks minor. lowpc == highpc looks real. gdb somehow manages to detect the PC range for _init. Not sure which of tested files is that.
Comment 1 Sergei Trofimovich 2022-08-06 06:47:04 UTC
Created attachment 14257 [details]
low_high_pc

Attaching the binary itself. It was a failing selftest:

$ LD_LIBRARY_PATH=$PWD/libdw:$PWD/libelf tests/low_high_pc -e tests/low_high_pc
[b] ../sysdeps/i386/start.S
[26] _start
[40] ../sysdeps/x86/abi-note.c
[b52] init.c
[b8e] static-reloc.c
[2dba] _dl_relocate_static_pie
[2dd8] ../sysdeps/i386/crti.S
[2def] _init
lowpc: 8049000, highpc: 8049000
../sysdeps/i386/crti.S: [2def] '_init' highpc <= lowpc

gdb somehow knows the size of the _init:

$ gdb /tmp/low_high_pc
Reading symbols from /tmp/low_high_pc...
(gdb) disassemble _init
Dump of assembler code for function _init:
   0x08049000 <+0>:     endbr32
   0x08049004 <+4>:     push   %ebx
   0x08049005 <+5>:     sub    $0x8,%esp
   0x08049008 <+8>:     call   0x8049400 <__x86.get_pc_thunk.bx>
   0x0804900d <+13>:    add    $0x2f93,%ebx
   0x08049013 <+19>:    mov    0x5c(%ebx),%eax
   0x08049019 <+25>:    test   %eax,%eax
   0x0804901b <+27>:    je     0x804901f <_init+31>
   0x0804901d <+29>:    call   *%eax
   0x0804901f <+31>:    add    $0x8,%esp
   0x08049022 <+34>:    pop    %ebx
   0x08049023 <+35>:    ret
End of assembler dump.
Comment 2 Sergei Trofimovich 2022-08-06 06:58:03 UTC
Could it be that it's a mild `crti.o` corruption by new binutils? "MIPS assembler" does not sound right.

```
$ readelf -aW crti.o --debug-dump
...
Contents of the .debug_info section:

  Compilation Unit @ offset 0x0:
   Length:        0x49 (32-bit)
   Version:       2
   Abbrev Offset: 0x0
   Pointer Size:  4
 <0><b>: Abbrev Number: 1 (DW_TAG_compile_unit)
    <c>   DW_AT_stmt_list   : (data4) 0x0
    <10>   DW_AT_ranges      : (data4) 0x0
    <14>   DW_AT_name        : (strp) (offset: 0x0): ../sysdeps/i386/crti.S
    <18>   DW_AT_comp_dir    : (strp) (offset: 0x17): /build/glibc-2.35/csu
    <1c>   DW_AT_producer    : (strp) (offset: 0x2d): GNU AS 2.39
    <20>   DW_AT_language    : (data2) 32769    (MIPS assembler)
 <1><22>: Abbrev Number: 2 (DW_TAG_subprogram)
    <23>   DW_AT_name        : (strp) (offset: 0x39): _init
    <27>   DW_AT_external    : (flag) 1
    <28>   DW_AT_low_pc      : (addr) 0x0
    <2c>   DW_AT_high_pc     : (addr) 0x0
 <1><30>: Abbrev Number: 2 (DW_TAG_subprogram)
    <31>   DW_AT_name        : (strp) (offset: 0x3f): __x86.get_pc_thunk.bx
    <35>   DW_AT_external    : (flag) 1
    <36>   DW_AT_low_pc      : (addr) 0x0
    <3a>   DW_AT_high_pc     : (addr) 0x4
 <1><3e>: Abbrev Number: 2 (DW_TAG_subprogram)
    <3f>   DW_AT_name        : (strp) (offset: 0x55): _fini
    <43>   DW_AT_external    : (flag) 1
    <44>   DW_AT_low_pc      : (addr) 0x0
    <48>   DW_AT_high_pc     : (addr) 0x0
 <1><4c>: Abbrev Number: 0
...

```
Comment 3 Sergei Trofimovich 2022-08-06 08:06:02 UTC
gas-2.39 started adding 0-sized DIE for function without a .size marking:

$ cat crti.S.S

 .section .init,"ax",@progbits
 .p2align 2
 .globl _init
 .hidden _init
 .type _init, @function
_init:
  .section .text.__x86.get_pc_thunk.bx,"axG",@progbits,__x86.get_pc_thunk.bx,comdat
    .globl __x86.get_pc_thunk.bx
    .hidden __x86.get_pc_thunk.bx
    .p2align 4
    .type __x86.get_pc_thunk.bx,@function
    __x86.get_pc_thunk.bx:
      ud2
    .size __x86.get_pc_thunk.bx, . - __x86.get_pc_thunk.bx
    .previous
  ud2

$ as-2.39 --gdwarf2 --32 -o crti.o crti.S.S
$ readelf -aW --debug-dump crti.o

 <0><b>: Abbrev Number: 1 (DW_TAG_compile_unit)
    <c>   DW_AT_stmt_list   : (data4) 0x0
    <10>   DW_AT_ranges      : (data4) 0x0
    <14>   DW_AT_name        : (strp) (offset: 0x0): crti.S.S
    <18>   DW_AT_comp_dir    : (strp) (offset: 0x9): /home/slyfox/dev/bugs/binutils-2.39-crti-size/delta
    <1c>   DW_AT_producer    : (strp) (offset: 0x3d): GNU AS 2.39
    <20>   DW_AT_language    : (data2) 32769    (MIPS assembler)
 <1><22>: Abbrev Number: 2 (DW_TAG_subprogram)
    <23>   DW_AT_name        : (strp) (offset: 0x49): _init
    <27>   DW_AT_external    : (flag) 1
    <28>   DW_AT_low_pc      : (addr) 0x0
    <2c>   DW_AT_high_pc     : (addr) 0x0
 <1><30>: Abbrev Number: 2 (DW_TAG_subprogram)
    <31>   DW_AT_name        : (strp) (offset: 0x4f): __x86.get_pc_thunk.bx
    <35>   DW_AT_external    : (flag) 1
    <36>   DW_AT_low_pc      : (addr) 0x0
    <3a>   DW_AT_high_pc     : (addr) 0x2
 <1><3e>: Abbrev Number: 0

$ as-2.38 --gdwarf2 --32 -o crti.o crti.S.S
$ readelf -aW --debug-dump crti.o

 <0><b>: Abbrev Number: 1 (DW_TAG_compile_unit)
    <c>   DW_AT_stmt_list   : (data4) 0x0
    <10>   DW_AT_ranges      : (data4) 0x0
    <14>   DW_AT_name        : (strp) (offset: 0x0): crti.S.S
    <18>   DW_AT_comp_dir    : (strp) (offset: 0x9): /home/slyfox/dev/bugs/binutils-2.39-crti-size/delta
    <1c>   DW_AT_producer    : (strp) (offset: 0x3d): GNU AS 2.38
    <20>   DW_AT_language    : (data2) 32769    (MIPS assembler)

Not sure if it's a discrepancy in binutils.
Comment 4 Sergei Trofimovich 2022-08-06 08:10:37 UTC
Filed https://sourceware.org/PR29451 against gas.
Comment 5 Mark Wielaard 2022-08-07 23:30:04 UTC
(In reply to Sergei Trofimovich from comment #4)
> Filed https://sourceware.org/PR29451 against gas.

Right. I added a comment there. Hopefully it gets fixed in binutils. Because I believe the elfutils testcase is correct. Adding a low_pc/high_pc attribute means you are describing a single continuous address range. You cannot do that by having low_pc == high_pc.
Comment 6 Mark Wielaard 2022-08-10 11:13:29 UTC
Did the binutils 2.39 commit:

commit e8cf73215187b0c08679d726a5cc7c019fa3ea2e
Author: Jan Beulich <jbeulich@suse.com>
Date:   Wed Aug 10 10:34:22 2022 +0200

    gas/Dwarf: properly skip zero-size functions

Fix this?

In that case I suggest we close this bug and recommend people upgrade to binutils 2.39.1 when that gets released
Comment 7 Martin Liska 2022-08-10 11:15:05 UTC
(In reply to Mark Wielaard from comment #6)
> Did the binutils 2.39 commit:
> 
> commit e8cf73215187b0c08679d726a5cc7c019fa3ea2e
> Author: Jan Beulich <jbeulich@suse.com>
> Date:   Wed Aug 10 10:34:22 2022 +0200
> 
>     gas/Dwarf: properly skip zero-size functions
> 
> Fix this?

Yes.
Comment 8 Sergei Trofimovich 2022-08-11 09:17:08 UTC
Yeah. The gas fix fixed elfutils test suite for me. Let's mark it RESOLVED/MOVED.