Use of SSE4.2 to compute crc32c

crc32c.c

#include <nmmintrin.h>
#include <stdint.h>
#include <string.h>

uint32_t crc32(uint32_t len, uint8_t *msg) {
	uint32_t pos = 0, hsh = -1;
	while (1) {
		uint32_t tmp = len - pos;
		if (tmp >= 4) {
			uint32_t val;
			memcpy(&val, msg + pos, sizeof(val));
			hsh = _mm_crc32_u32(hsh, val);
			pos += 4;
		} else if (tmp >= 2) {
			uint16_t val;
			memcpy(&val, msg + pos, sizeof(val));
			hsh = _mm_crc32_u16(hsh, val);
			pos += 2;
		} else {
			uint8_t val;
			memcpy(&val, msg + pos, sizeof(val));
			hsh = _mm_crc32_u8(hsh, val);
			break;
		}
	}
	return ~hsh;
}