This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug middle-end/69258] Flexible arrays break TBAA
- From: "rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Thu, 14 Jan 2016 08:14:30 +0000
- Subject: [Bug middle-end/69258] Flexible arrays break TBAA
- Auto-submitted: auto-generated
- References: <bug-69258-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69258
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution|--- |INVALID
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to joseph@codesourcery.com from comment #5)
> I don't see a need for these different structures to be able to alias.
> (Flexible array members do need to be able to alias static storage, in the
> case where that static storage was declared with the structure type with
> the flexible array member and the GNU C extension of initializing such
> flexible array members was used.)
Ok, that works. Phew. Not sure if it is a solution for the Fortran FE
issue which wants to construct Xflex with a specific size on the stack
(it's the array descriptor with a known dimensionality). I remember that
GNU C extension poses some interesting effects on the ME (inconsistent
TYPE_SIZE / DECL_SIZE at least).
extern void abort (void);
struct Xflex { int n; int a[]; };
struct Xflex x = { 1, { 1, 2, 3, 4, 5, 6, 7 } };
int __attribute__((noinline,noclone))
foo (struct Xflex *f)
{
x.a[6] = 1;
f->a[6] = 2;
return x.a[6];
}
int __attribute__((noinline,noclone))
bar (struct Xflex *f)
{
f->a[6] = 2;
x.a[6] = 1;
return f->a[6];
}
int main()
{
if (foo ((struct Xflex *)&x) != 2)
abort ();
if (bar ((struct Xflex *)&x) != 1)
abort ();
return 0;
}