Unable to stop at breakpoint in 32-bit executable

[James Foster]

I believe I'm having the problem described in #1846557 and #1848200. Those bugs were reported fixed in 8.1.1 but I'm observing them in 9.2 so I suspect a regression (in fact, if I read the notes correctly, the earlier reports were from another regression!).

I create a new EC2 server on AWS with Ubuntu Server 20.04 LTS (64-bit Arm) then perform the following installs (not sure what is needed or if something is missing):

```
sudo dpkg --add-architecture armhf
sudo apt update
sudo apt upgrade -y
sudo apt install -y build-essential vim binutils-arm-linux-gnueabi \
  gdb libc6:armhf libstdc++6:armhf
```

Then I create hello.s:

```
        .text
        .global _start
_start:
        MOV R7, #4
        MOV R0, #1
        LDR R1, =hello
        LDR R2, =hello_size
        SWI 0
        MOV R7, #1
        SWI 0

        .data
hello: .asciz "Hello world!\n"
        .equ hello_size, (.-hello)
```

Then I assemble, link, and run, all with success. Finally, I try to debug:

```
$ arm-linux-gnueabi-as -ggdb -mcpu=cortex-a57 -o out.o hello.s
$ arm-linux-gnueabi-ld out.o -dynamic-linker=/usr/lib/ld-linux-armhf.so.3 -o out
$ rm out.o
$ ./out
Hello world!
$ gdb out
GNU gdb (Ubuntu 9.2-0ubuntu1~20.04) 9.2
...
Reading symbols from out...
(gdb) b _start
Breakpoint 1 at 0x10074: file hello.s, line 4.
(gdb) run
Starting program: /home/ubuntu/280/out
```

At this point everything hangs until I press ^C:

```
^C
Program received signal SIGINT, Interrupt.
0x0000aaaaba072284 in ?? ()
(gdb) bt
#0 0x0000aaaaba072284 in ?? ()
#1 0x0000000000007232 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) l
1 .text
2 .global _start
3 _start:
4 MOV R7, #4
5 MOV R0, #1
6 LDR R1, =hello
7 LDR R2, =hello_size
8 SWI 0
9 MOV R7, #1
10 SWI 0
(gdb) q
A debugging session is active.

 Inferior 1 [process 29234] will be killed.

Quit anyway? (y or n) y
$
```

It appears that gdb generally works and recognizes the executable, but breakpoints don't work. I'm new at Arm and don't have a lot of experience with assembly (at least not since the early 1970s), so it could easily be my problem but it seems suspiciously like an earlier problem and I think what I've done should work.

[Luis (freezehawk)]

This doesn't seem to be related to the previous bugs.

I tried to reproduce this on 18.04.4 (I don't have 20.04) and a fresh build of GDB master, and I couldn't.

(gdb) disass _start
Dump of assembler code for function _start:
   0x00010074 <+0>: mov r7, #4
   0x00010078 <+4>: mov r0, #1
   0x0001007c <+8>: ldr r1, [pc, #12] ; 0x10090 <_start+28>
   0x00010080 <+12>: ldr r2, [pc, #12] ; 0x10094 <_start+32>
   0x00010084 <+16>: svc 0x00000000
   0x00010088 <+20>: mov r7, #1
   0x0001008c <+24>: svc 0x00000000
   0x00010090 <+28>: muleq r2, r8, r0
   0x00010094 <+32>: andeq r0, r0, lr
End of assembler dump.
(gdb) b _start
Breakpoint 1 at 0x10074: file /tmp/hello.S, line 4.
(gdb) r
Starting program: /tmp/hello

Breakpoint 1, _start () at /tmp/hello.S:4
4 MOV R7, #4

With Ubuntu 18.04.4's GDB it also works.

Can you please turn on debugging and paste the log?

Use "set debug infrun 1" before the run command.

[James Foster (ubuntu-a6a)]

I've been able to reproduce the problem on 32-bit Ubuntu 20.04 for RPi4. The problem reproduces when the following is linked:

ld-linux-armhf.so.3 -> ld-2.31.so*

fostja@cptr280-arm:~/code/asm$ gdb -q hello
GEF for linux ready, type `gef' to start, `gef config' to configure
75 commands loaded for GDB 9.2 using Python engine 3.8
[*] 5 commands could not be loaded, run `gef missing` to know why.
Reading symbols from hello...
gef➤ set debug infrun 1
gef➤ b _start
Breakpoint 1 at 0x1016c: file hello.s, line 5.
gef➤ start
infrun: infrun_async(1)
infrun: proceed (addr=0xb6fd5a80, signal=GDB_SIGNAL_0)
infrun: proceed: resuming process 10100
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread [process 10100] at 0xb6fd5a80
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = ignore
infrun: handle_inferior_event status->kind = ignore
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = ignore
infrun: handle_inferior_event status->kind = ignore
infrun: prepare_to_wait

When I revert to the following then things work properly:

ld-linux-armhf.so.3 -> ld-2.28.so*

So I have a work-around. It appears that the problem was introduced between 2.28 and 2.31.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Hyun (hyunwoo-park)]

I have same issue as James described. A static link of execution file, "-static" option in gcc, works in gdb while shared is not.
Any updates on this issue?

bulssi@bulssi:~/work$ cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=20.04
DISTRIB_CODENAME=focal
DISTRIB_DESCRIPTION="Ubuntu 20.04 LTS"
bulssi@bulssi:~/work$ lscpu
Architecture: aarch64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 4
On-line CPU(s) list: 0,1
Off-line CPU(s) list: 2,3
Thread(s) per core: 1
Core(s) per socket: 2
Socket(s): 1
Vendor ID: ARM
Model: 0
Model name: Cortex-A55
Stepping: r2p0
BogoMIPS: 100.00
L2 cache: 256 KiB
L3 cache: 512 KiB
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Spec store bypass: Not affected
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm lrcpc dcpop asimddp
bulssi@bulssi:~/work$ uname -a
Linux bulssi 5.4.96+ #4 SMP PREEMPT Tue Sep 28 02:25:02 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux
bulssi@bulssi:~/work$ cat test.c

#include <stdio.h>

int main (int argc, char **argv)
{
        printf ("hello\n");
        return 0;
}
bulssi@bulssi:~/work$ arm-linux-gnueabihf-gcc -o a32.out test.c
bulssi@bulssi:~/work$ ./a32.out
hello
bulssi@bulssi:~/work$ gdb ./a32.out
GNU gdb (Ubuntu 9.1-0ubuntu1) 9.1
Copyright (C) 2020 Free Software Foundation, Inc.

...

Reading symbols from ./a32.out...
(No debugging symbols found in ./a32.out)
(gdb) set architecture arm
The target architecture is assumed to be arm
(gdb) r
Starting program: /home/bulssi/work/a32.out
^C
Program received signal SIGINT, Interrupt.
0xf7fd12c4 in ?? () from /lib/ld-linux-armhf.so.3
(gdb) bt
#0 0xf7fd12c4 in ?? () from /lib/ld-linux-armhf.so.3
#1 0xf7fc81e4 in ?? () from /lib/ld-linux-armhf.so.3
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb)
(gdb) set debug infrun 1
(gdb) run
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /home/bulssi/work/a32.out
infrun: infrun_async(1)
infrun: proceed (addr=0xf7fc5a80, signal=GDB_SIGNAL_0)
infrun: proceed: resuming process 19335
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread [process 19335] at 0xf7fc5a80
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = ignore
infrun: handle_inferior_event status->kind = ignore
infrun: prepare_to_wait

^Cinfrun: target_wait (-1.0.0, status) =
infrun: 19335.19335.0 [process 193...