#define ITERATIONS 3
#define N (3 * ITERATIONS) /* number of bits in the input numbers */
#define HI_OVERFLOW 3
#define BITS_IN_INT 32
#define HI_HALF (BITS_IN_INT - HI_OVERFLOW - N)
#define LO_HALF (HI_HALF - N)
 
/* #define ADD_ACC(P,B) (P += ((B) << HI_HALF)) */
#define ADD_ACC(P,B) add_acc(&P, B)
#define GET_RESULT(P) (((P << HI_OVERFLOW) >> HI_OVERFLOW) >> LO_HALF)
#define GET_LOW_FOUR(P) ((P >> (LO_HALF -1)) & 0x0f)

void add_acc(int* p, int b){ 
	int a = b << HI_HALF;
	*p += a;
}

typedef enum
{ zero, plusB, plus2B, plus3B, plus4B, subB, sub2B, sub3B, sub4B } command;

static const command command_table[16] =
{ zero, plusB, plusB, plus2B, plus2B, plus3B, plus3B, plus4B,
  sub4B, sub3B, sub3B, sub2B, sub2B, subB, subB, zero };

int mul(int a, int b)
{
    int p = 0;
    int i;
    int threeB;
    a &= ((1 << N) - 1);
    /*    b &= ((1 << N) - 1); */
    p = (a << LO_HALF);
    threeB = b + (b << 1);
    for(i = 0; i < ITERATIONS; ++i){
	switch( command_table[ GET_LOW_FOUR(p) ] ){
	case plusB:
	    ADD_ACC(p,b);
	    break;
	case plus2B:
	    ADD_ACC(p,b << 1);
	    break;
	case plus3B:
	    ADD_ACC(p,threeB);
	    break;
	case plus4B:
	    ADD_ACC(p,b << 2);
	    break;
	case subB:
	    ADD_ACC(p,-b);
	    break;
	case sub2B:
	    ADD_ACC(p,-(b << 1));
	    break;
	case sub3B:
	    ADD_ACC(p,-threeB);
	    break;
	case sub4B:
	    ADD_ACC(p,-(b << 2));
	    break;
	case zero:
	default:
	}
	p >>= 3;
	
    }
    return GET_RESULT(p);
	    

}