Flooring a float in C

floor.c

/**
* floating points:
* 32 bit number
*
* 0-22 mantissa
* 23-30 exp
* 31 sign
*
* Theory: You can use the exp potion to determine the offset and null out the bits to produce a floored number.
* However, its just a theory I had. I couldn't 100% determine how many bits I needed to cut off, but I did, however,
* determine that you can get up to 120 percision using an offset of 18.
* ex: floorf32(120.1239, 18) -> 120
*
* Another thing is, the compiler may get in your way with errors.
*/

float floorf32(float in, uint32_t offset) {
  return (float)(*((uint32_t*)&in) >> offset) << offset;
}