[Bug c/49779] New: Wrong code generated for while loop with guard containing continue

[Bug c/49779] New: Wrong code generated for while loop with guard containing continue

[bagnara at cs dot unipr.it]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

           Summary: Wrong code generated for while loop with guard
                    containing continue
           Product: gcc
           Version: 4.6.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: bagnara@cs.unipr.it


GCC turns a program that should loop forever into one that terminates:

$ cat q.c
int main() {
  do {
    while (({ continue; 1; })) {
      return 1;
    }
  } while (0);
  return 0;
}
$ gcc q.c
$ a.out && echo yes
yes

Apparently, GCC erroneously associates the `continue' to the outer `do' loop.

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-07-18 17:36:38 UTC ---
Hmm, statement expressions are an extension to the C language.

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[bagnara at cs dot unipr.it]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

--- Comment #2 from bagnara at cs dot unipr.it 2011-07-18 17:42:39 UTC ---
An extension that is still supported by GCC, right?

Here is another testcase showing the same phenomenon with `break':

int main() {
  do {
    while (({ break; 1; })) {
      ;
    }
    return 0;
  } while (0);
  return 1;
}

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-07-18 17:47:49 UTC ---
(In reply to comment #2)
> An extension that is still supported by GCC, right?

Yes but the semantics are not well defined in this area IIRC.

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[bagnara at cs dot unipr.it]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

--- Comment #4 from bagnara at cs dot unipr.it 2011-07-20 06:44:25 UTC ---
Well...  do you agree that the condition of a while loop has to be
evaluated at each iteration?  If so, then we agree that the condition
is inside the while loop, not outside.  Hence, if the condition
contains unnested breaks or continues these have to refer to the while
loop in question, not to the while/for/do/switch construction that may
(or may not) contain it.

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

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

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

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-07-20 07:09:56 UTC ---
The break/continue/goto out of the condition in the do/while/for condition (or
for increment or for init) might very well be considered undefined behavior,
the statement expression extension isn't specified precisely enough to say what
exactly happens.

[Bug c/49779] Wrong code generated for while loop with guard containing continue

[joseph at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49779

--- Comment #6 from joseph at codesourcery dot com <joseph at codesourcery dot com> 2011-07-21 14:26:05 UTC ---
See PR 44715 for previous discussion of issues with loop control 
statements in statement expressions.