[Bug c/111319] New: Wrong code at -O2 on x86_64-linux-gnu

[Bug c/111319] New: Wrong code at -O2 on x86_64-linux-gnu

[shaohua.li at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111319

            Bug ID: 111319
           Summary: Wrong code at -O2 on x86_64-linux-gnu
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: shaohua.li at inf dot ethz.ch
  Target Milestone: ---

The executable produced by gcc -O2 would crash. Not sure if there is UB in the
code or not as `h` has not been dereferenced after `h--`.

Compiler explorer: https://godbolt.org/z/avssssPze

$ cat a.c
int printf(const char *, ...);
int a, b = 5, c, d;
int e[2] = {0, 0};
int i() {
  int f = 0;
  int g[2] = {1, f};
  int *h = g;
  h--;
  if (!b)
    while (h >= &c == 1)
      h--;
  h++;
  d = *h;
  return d;
}
int main() {
  if (e[i()])
    e[0] = 0;
  printf("%d\n", a);
}
$
$ gcc -O2 a.c && ./a.out
Segmentation fault
$

[Bug c/111319] Wrong code at -O2 on x86_64-linux-gnu

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111319

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---

  int *h = g;
  h--;

is undefined right away ...

[Bug c/111319] Wrong code at -O2 on x86_64-linux-gnu

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111319

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Also comparing unrelated addresses with >= is undefined.

[Bug c/111319] Wrong code at -O2 on x86_64-linux-gnu

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111319

Xi Ruoyao <xry111 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
                 CC|                            |xry111 at gcc dot gnu.org
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #3 from Xi Ruoyao <xry111 at gcc dot gnu.org> ---
Please try not to create more reports with "not sure if there is UB" esp. this
UB should be obvious, and it's also detectable with -fsanitize=undefined:

t.c:13:7: runtime error: load of address 0x7ffd53183020 with insufficient space
for an object of type 'int'
0x7ffd53183020: note: pointer points here
 2a 56 00 00  01 00 00 00 00 00 00 00  00 14 1a ed ca 50 d4 0e  00 00 00 00 00
00 00 00  98 31 18 53
              ^

[Bug c/111319] Wrong code at -O2 on x86_64-linux-gnu

[shaohua.li at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111319

--- Comment #4 from Shaohua Li <shaohua.li at inf dot ethz.ch> ---
(In reply to Xi Ruoyao from comment #3)
> Please try not to create more reports with "not sure if there is UB" esp.
> this UB should be obvious, and it's also detectable with
> -fsanitize=undefined:
> 
> t.c:13:7: runtime error: load of address 0x7ffd53183020 with insufficient
> space for an object of type 'int'
> 0x7ffd53183020: note: pointer points here
>  2a 56 00 00  01 00 00 00 00 00 00 00  00 14 1a ed ca 50 d4 0e  00 00 00 00
> 00 00 00 00  98 31 18 53
>               ^

UBSan only detects it at -O2 level and I used `-O0
-fsanitize=address,undefined` to detect if there is any report.

https://godbolt.org/z/zvcG8znvT