[Bug tree-optimization/36054] bad code generation with -ftree-vectorize

[tim at klingt dot org]

------- Comment #18 from tim at klingt dot org  2008-05-05 09:09 -------
hm, if that code is broken, what about the following one on x86_64
(__sync_bool_compare_and_swap_16 requires an alignment of 16 byte)?

struct __attribute__((aligned(16))) foo_t
{
    foo_t(int a = 0, int b = 0): a_(a), b_(b) {}

    void CAS (foo_t const & rhs)
    {
        typedef int TItype __attribute__ ((mode (TI)));

        union cu
        {
            long c[2];
            TItype l;
        };

        cu old;
        old.c[0] = (long)a_;
        old.c[1] = (long)b_;

        cu nw;
        nw.c[0] = (long)rhs.a_;
        nw.c[1] = (long)rhs.b_;


        __sync_bool_compare_and_swap_16(reinterpret_cast<volatile
foo_t*>(this), old.l, nw.l);
    }

    int a_, b_;
};

int main()
{
    foo_t * f = new foo_t();
    f->CAS(foo_t(1,2));
}

thanks, tim


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36054