#include <metal_stdlib>

using namespace metal;

template<typename T, size_t N>
struct tint_array {
    const constant T& operator[](size_t i) const constant { return elements[i]; }
    device T& operator[](size_t i) device { return elements[i]; }
    const device T& operator[](size_t i) const device { return elements[i]; }
    thread T& operator[](size_t i) thread { return elements[i]; }
    const thread T& operator[](size_t i) const thread { return elements[i]; }
    threadgroup T& operator[](size_t i) threadgroup { return elements[i]; }
    const threadgroup T& operator[](size_t i) const threadgroup { return elements[i]; }
    T elements[N];
};

#define TINT_ISOLATE_UB(VOLATILE_NAME) \
  volatile bool VOLATILE_NAME = true; \
  if (VOLATILE_NAME)

struct tint_packed_vec3_f32_array_element {
  /* 0x0000 */ packed_float3 elements;
  /* 0x000c */ tint_array<int8_t, 4> tint_pad;
};

float3x3 tint_unpack_vec3_in_composite(tint_array<tint_packed_vec3_f32_array_element, 3> in) {
  float3x3 result = float3x3(float3(in[0].elements), float3(in[1].elements), float3(in[2].elements));
  return result;
}

tint_array<float3x3, 4> tint_unpack_vec3_in_composite_1(tint_array<tint_array<tint_packed_vec3_f32_array_element, 3>, 4> in) {
  tint_array<float3x3, 4> result = tint_array<float3x3, 4>{tint_unpack_vec3_in_composite(in[0]), tint_unpack_vec3_in_composite(in[1]), tint_unpack_vec3_in_composite(in[2]), tint_unpack_vec3_in_composite(in[3])};
  return result;
}

void assign_and_preserve_padding_1(device tint_array<tint_packed_vec3_f32_array_element, 3>* const dest, float3x3 value) {
  (*(dest))[0u].elements = packed_float3(value[0u]);
  (*(dest))[1u].elements = packed_float3(value[1u]);
  (*(dest))[2u].elements = packed_float3(value[2u]);
}

void assign_and_preserve_padding(device tint_array<tint_array<tint_packed_vec3_f32_array_element, 3>, 4>* const dest, tint_array<float3x3, 4> value) {
  TINT_ISOLATE_UB(tint_volatile_true) for(uint i = 0u; (i < 4u); i = (i + 1u)) {
    assign_and_preserve_padding_1(&((*(dest))[i]), value[i]);
  }
}

kernel void f(device tint_array<tint_array<tint_packed_vec3_f32_array_element, 3>, 4>* tint_symbol [[buffer(1)]], const constant tint_array<tint_array<tint_packed_vec3_f32_array_element, 3>, 4>* tint_symbol_1 [[buffer(0)]]) {
  assign_and_preserve_padding(tint_symbol, tint_unpack_vec3_in_composite_1(*(tint_symbol_1)));
  assign_and_preserve_padding_1(&((*(tint_symbol))[1]), tint_unpack_vec3_in_composite((*(tint_symbol_1))[2]));
  (*(tint_symbol))[1][0].elements = packed_float3(float3((*(tint_symbol_1))[0][1].elements).zxy);
  (*(tint_symbol))[1][0].elements[0] = (*(tint_symbol_1))[0][1].elements[0];
  return;
}