[Bug c/106853] New: compound literal storage duration

[Bug c/106853] New: compound literal storage duration

[marcelgcc at firemail dot eu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106853

            Bug ID: 106853
           Summary: compound literal storage duration
           Product: gcc
           Version: 11.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marcelgcc at firemail dot eu
  Target Milestone: ---

According to the C Standard a compound literal has automatic storage duration.
Please, consider the following code. Note it's C code NOT C++:

% cat test.c
#include <stdio.h>

struct ab {
  int a, b;
} *p1, *p2;

int set_p1(struct ab *ab) {
  p1 = ab;
  return 0;
}
int set_p2(struct ab *ab) {
  p2 = ab;
  return 0;
}

int main() {

  if (0 != set_p1(&(struct ab){42, 42})) {
    return 1;
  }
  set_p2(&(struct ab){46, 46});

  printf("%d, %d\n", p1->a, p1->b);
  printf("%d, %d\n", p2->a, p2->b);
  return 0;
}
/*EOF*/


# Expected Output:
% clang-15 -O1 test.c && ./a.out
42, 42                                        
46, 46

#####GCC#####

% gcc-11 -O1 test.c && ./a.out 
0, 0
46, 46

% gcc-11 test.c && ./a.out
42, 42
46, 46

% gcc-12 test.c && ./a.out 
46, 46
46, 46

% gcc-12 -O1 test.c && ./a.out
-760366272, 32718
46, 46

% gcc-11 --version
gcc (Debian 11.3.0-5) 11.3.0

% gcc-12 --version         
gcc (GCC) 12.2.0

[Bug c/106853] compound literal storage duration

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106853

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

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

> if (0 != set_p1(&(struct ab){42, 42}))

Yes it is auto storage and the scope is only inside the condition statement as
specified by the standard.

[Bug c/106853] compound literal storage duration

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106853

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
In C89 that would be valid scope-wise (but C89 doesn't have compound literals
on the other side), but in C99 and later compound statements like if introduce
a new block scope.  So, accessing *p1 is valid during evaluation of the if
condition or during evaluation of the substatements, but not after the compound
statement ends.  As the testcase does that, it invokes UB and anything can
happen.

[Bug c/106853] compound literal storage duration

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106853

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
The important part of the standards:
6.5.2.5 Compound literals

5 The value of the compound literal is that of an unnamed object initialized by
the initializer list. If the compound literal occurs outside the body of a
function, the object has static storage duration; otherwise, it has automatic
storage duration associated with the enclosing block.

-----
6.8.4 Selection statements

3 A selection statement is a block whose scope is a strict subset of the scope
of its enclosing block. Each associated substatement is also a block whose
scope is a strict subset of the scope of the selection statement.

[Bug c/106853] compound literal storage duration

[marcelgcc at firemail dot eu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106853

--- Comment #4 from marcelgcc at firemail dot eu ---
Thanks for the explanation.