test/tint/buffer/uniform/std140/unnested/mat3x3_f32/to_workgroup.wgsl.expected.msl

#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];
};

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

tint_array<tint_packed_vec3_f32_array_element, 3> tint_pack_vec3_in_composite(float3x3 in) {
  tint_array<tint_packed_vec3_f32_array_element, 3> result = tint_array<tint_packed_vec3_f32_array_element, 3>{tint_packed_vec3_f32_array_element{.elements=packed_float3(in[0])}, tint_packed_vec3_f32_array_element{.elements=packed_float3(in[1])}, tint_packed_vec3_f32_array_element{.elements=packed_float3(in[2])}};
  return result;
}

void tint_zero_workgroup_memory(uint local_idx, threadgroup tint_array<tint_packed_vec3_f32_array_element, 3>* const tint_symbol) {
  if ((local_idx < 1u)) {
    *(tint_symbol) = tint_pack_vec3_in_composite(float3x3(float3(0.0f), float3(0.0f), float3(0.0f)));
  }
  threadgroup_barrier(mem_flags::mem_threadgroup);
}

void f_inner(uint local_invocation_index, threadgroup tint_array<tint_packed_vec3_f32_array_element, 3>* const tint_symbol_1, const constant tint_array<tint_packed_vec3_f32_array_element, 3>* const tint_symbol_2) {
  tint_zero_workgroup_memory(local_invocation_index, tint_symbol_1);
  *(tint_symbol_1) = *(tint_symbol_2);
  (*(tint_symbol_1))[1].elements = (*(tint_symbol_2))[0].elements;
  (*(tint_symbol_1))[1].elements = packed_float3(float3((*(tint_symbol_2))[0].elements).zxy);
  (*(tint_symbol_1))[0].elements[1] = (*(tint_symbol_2))[1].elements[0];
}

kernel void f(const constant tint_array<tint_packed_vec3_f32_array_element, 3>* tint_symbol_4 [[buffer(0)]], uint local_invocation_index [[thread_index_in_threadgroup]]) {
  threadgroup tint_array<tint_packed_vec3_f32_array_element, 3> tint_symbol_3;
  f_inner(local_invocation_index, &(tint_symbol_3), tint_symbol_4);
  return;
}