I try to write this simple C function for AVX2 horizontal sum in assembly.

static inline float hsum256_ps(__m256 v) {
    __m128 lo = _mm256_castps256_ps128(v);
    __m128 hi = _mm256_extractf128_ps(v, 1);
    __m128 sum128 = _mm_add_ps(lo, hi);
    __m128 shuf = _mm_movehdup_ps(sum128);
    __m128 sums = _mm_add_ps(sum128, shuf);
    shuf = _mm_movehl_ps(shuf, sums);
    sums = _mm_add_ss(sums, shuf);
    return _mm_cvtss_f32(sums);
}

I came up with

static inline float hsum256_ps(__m256 v){
    float r;

    __asm__ __volatile__ (
        "vextractf128  $0,  %1, %%xmm0           \n\t"
        "vextractf128  $1,  %1, %%xmm1           \n\t"
        "vaddps        %%xmm0, %%xmm0, %%xmm1    \n\t"
        "movshdup      %%xmm0, %%xmm1            \n\t"
        "vaddps        %%xmm0, %%xmm0, %%xmm1    \n\t"
        "movhlps       %%xmm0, %%xmm1            \n\t"
        "vaddss        %%xmm0, %%xmm0, %%xmm1    \n\t"
        "vmovss        %%xmm0, %0                \n\t"

        : "=m"(r) 
        : "x"(v) 
        : "xmm0", "xmm1"
    );

    return r;
}

but it does not seem to work. Where did I do wrong?