This is a target bug as it does not effect any reasonable processor.
With -mfpmath=sse -msse2 I get:
.L2:
        decl    %eax
        addsd   %xmm1, %xmm0
        jne     .L2

-- Pinski



Reply via email to