https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79671
--- Comment #21 from rguenther at suse dot de <rguenther at suse dot de> --- On February 27, 2017 5:46:44 PM GMT+01:00, "jakub at gcc dot gnu.org" <gcc-bugzi...@gcc.gnu.org> wrote: >https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79671 > >Jakub Jelinek <jakub at gcc dot gnu.org> changed: > > What |Removed |Added >---------------------------------------------------------------------------- > CC| |redi at gcc dot gnu.org > >--- Comment #20 from Jakub Jelinek <jakub at gcc dot gnu.org> --- >I believe this all boils down to: >inline void* operator new(__SIZE_TYPE__, void *p) { return p; } >struct A { A (float x) : f (x) {} float f; }; >struct B >{ > int x; > union U > { > int a; > char b[sizeof (float)]; > } u; > int y; >}; > >__attribute__((noinline, noclone)) void >bar (B &x, B &y) >{ > if (x.x != 0 || x.y != 3 || y.x != 0 || y.y != 3) > __builtin_abort (); > float f; > __builtin_memcpy (&f, x.u.b, sizeof (float)); > if (f != 3.5f) > __builtin_abort (); > __builtin_memcpy (&f, y.u.b, sizeof (float)); > if (f != 3.5f) > __builtin_abort (); >} > >__attribute__((noinline, noclone)) >B * >baz (B &x) >{ > return &x; >} > >__attribute__((noinline, noclone)) void >foo (float x) >{ > B b { 0, {}, 3 }, c; > B *p = baz (b); > new (b.u.b) A (x); > c = *p; > bar (*p, c); >} > >int >main () >{ > foo (3.5f); >} > >which fails also on x86_64-linux at -O2. And that testcase regressed >with >r223126. Now whether this is valid C++, no idea, placement new is >messy. We have one or two duplicates that were resolved invalid. We can't reasonably support this in the ME anyway and the lvalue used to access this is not one of those listed compatible with A.
