[Bug c++/33887] New: Reference to bitfield gets wrong value when optimizing

["ian at airs dot com" <gcc-bugzilla at gcc dot gnu dot org>]

This C++ test case passes a bitfield value to a function which expects a const
reference.  When optimizing, the function gets the wrong value: it gets the
value 0x1000000, which is an acceptable value for a 24-bit bitfield, but is the
wrong value for extending the bitfield out to a 32-bit integer.  This failure
happens on i686-pc-linux-gnu with -O2.  The test passes without optimization.

extern "C" void abort() __attribute__ ((noreturn));

template<typename t1, typename t2>
void fn(const t1&, const t2&) __attribute__ ((noinline));

template<typename t1, typename t2>
void fn(const t1& v1, const t2& v2)
{
  if (v1 != v2)
    abort();
}

struct s
{
  unsigned long long f1 : 40;
  unsigned int f2 : 24;
};

s sv;

int main()
{
  sv.f1 = 0;
  sv.f2 = (1 << 24) - 1;
  fn(sv.f1, 0);
  fn(sv.f2, (1 << 24) - 1);
  ++sv.f2;
  fn(sv.f2, 0);
  return 0;
}


-- 
           Summary: Reference to bitfield gets wrong value when optimizing
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: ian at airs dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33887