1  /* { dg-do compile } */
       2  /* { dg-options "-O2 -fdump-tree-optimized" } */
       3  #include <arm_neon.h>
       4  
       5  void
       6  f (float32x4_t **ptr)
       7  {
       8    float32x4_t res = vsetq_lane_f32 (0.0f, **ptr, 0);
       9    **ptr = res;
      10  }
      11  /* GCC should be able to remove the call to "__builtin_aarch64_im_lane_boundsi"
      12     and optimize out the second load from *ptr.  */
      13  /* { dg-final { scan-tree-dump-times "__builtin_aarch64_im_lane_boundsi" 0 "optimized" } } */
      14  /* { dg-final { scan-tree-dump-times " = \\\*ptr_" 1 "optimized" } } */