[Bug c/102606] New: miscompilation of a program with large array in a dead-code

[Bug c/102606] New: miscompilation of a program with large array in a dead-code

[k.even-mendoza at imperial dot ac.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

            Bug ID: 102606
           Summary: miscompilation of a program with large array in a
                    dead-code
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: k.even-mendoza at imperial dot ac.uk
  Target Milestone: ---

This program miscompiled with gcc-12 (and also older versions) with -O0:

#include "stdio.h"
void foo(int *p)
{
  if(0) // Dead-code
  {
      int *i[(3500)][(300)];
  }
}
int main()
{
  int i = 1;
  foo(&i);
  printf("1\n");
  return 0;
}

the program terminates with Segmentation fault when compiled with -O0 and GCC,
but 
with -O1,O2,O3,Os and clang (all flags), it prints "1" and return 0. 
I assume the definition of "i" is being pulled up, out side the if(0) block.
I used gcc (GCC) 12.0.0 20210811 (experimental).

[Bug middle-end/102606] large array in a dead-code is not optimized away at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |middle-end
            Summary|miscompilation of a program |large array in a dead-code
                   |with large array in a       |is not optimized away at
                   |dead-code                   |-O0

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I don't think this is miscompiled but rather you just run out of stack space
because gcc at -O0 does not remove the the otherwise unused array.

This is by design.
Yes gcc could be improved to remove it but it is not a miscompile.

[Bug middle-end/102606] large array in a dead-code is not optimized away at -O0

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

Richard Biener <rguenth at gcc dot gnu.org> changed:

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

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Yes, even though the variable is in a local not executed scope the allocation
of stack space happens statically at function scope.  If you want this not
to happen you can use alloca() for allocation.

[Bug middle-end/102606] large array in a dead-code is not optimized away at -O0

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

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

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

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Or VLA.
Anyway, even at -O0 we can optimize away some if (0)s, but we need assurance
that there are no labels in there and that the code really dead.  So, e.g. if
constexpr (0) can be optimized away because there is a language guarantee that
one can't jump into those bodies.

[Bug middle-end/102606] large array in a dead-code is not optimized away at -O0

[k.even-mendoza at imperial dot ac.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

--- Comment #4 from Karine EM <k.even-mendoza at imperial dot ac.uk> ---
That would make sense, and I think it is also what happens when compiling the
example with Clang and -O0 since the compilation takes only a few seconds then.

[Bug middle-end/102606] large array in a dead-code is not optimized away at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102606

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |krzyk240 at gmail dot com

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 102577 has been marked as a duplicate of this bug. ***