https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68782
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2015-12-14
CC| |jakub at gcc dot gnu.org
Ever confirmed|0 |1
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Started with r230365 aka merge of the delayed folding branch.
struct B { int &b; };
constexpr B foo (int &x) { return { x }; }
struct C { int c; };
constexpr B bar (C &y) { return foo (y.c); }
int
main ()
{
C a { 42 };
if (&foo (a.c).b == nullptr) __builtin_abort ();
if (&a.c != &foo (a.c).b) __builtin_abort ();
if (&bar (a).b == nullptr) __builtin_abort ();
if (&a.c != &bar (a).b) __builtin_abort ();
}
shows IMHO better what's going on, already in *.original dump the PARM_DECL y
leaks into the main function (supposedly it should have been replaced by &a.
