demo of using adc instruction to add a bignum

adc-demo.c

#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>

typedef int bigit;
typedef long int iptr;


void asm_add(xp, xl, yp, yl, zp) bigit *xp, *yp, *zp; iptr xl, yl;
{
#define UNROLL 2
  volatile bigit sum;

  volatile iptr c1 = yl / UNROLL * UNROLL;
  volatile iptr c2 = yl % UNROLL;
  volatile iptr c3 = (xl - yl) / UNROLL * UNROLL;
  volatile iptr c4 = (xl - yl) % UNROLL;

  volatile bigit *x1 = xp - c1 + 1;
  volatile bigit *y1 = yp - c1 + 1;
  volatile bigit *z1 = zp - c1 + 1;

  volatile bigit *x2 = xp - yl + 1;
  volatile bigit *y2 = yp - yl + 1;
  volatile bigit *z2 = zp - yl + 1;

  volatile bigit *x3 = xp - xl + 1 + c4;
  volatile bigit *z3 = zp - xl + 1 + c4;

  volatile bigit *x4 = xp - xl + 1;
  volatile bigit *z4 = zp - xl + 1;

  volatile bigit *zz = zp - xl;

  // process segment shared by xp and yp
  __asm__ (
      "clc                           \n\t"
      "1:                            \n\t"
      "jecxz 2f                      \n\t"

      "movl -4(%[x],%[c],4), %[sum]  \n\t"
      "adcl -4(%[y],%[c],4), %[sum]  \n\t"
      "movl %[sum], -4(%[z],%[c],4)  \n\t"

      "movl -8(%[x],%[c],4), %[sum]  \n\t"
      "adcl -8(%[y],%[c],4), %[sum]  \n\t"
      "movl %[sum], -8(%[z],%[c],4)  \n\t"

      "lea -2(%[c]), %[c]            \n\t"
      "jmp 1b                        \n\t"
      "2:                            \n\t"
      : [c]"+c"(c1), [sum]"+r"(sum), [x]"+r"(x1), [y]"+r"(y1), [z]"+r"(z1));

  __asm__ (
      "1:                            \n\t"
      "jecxz 2f                      \n\t"
      "movl -4(%[x],%[c],4), %[sum]  \n\t"
      "adcl -4(%[y],%[c],4), %[sum]  \n\t"
      "movl %[sum], -4(%[z],%[c],4)  \n\t"
      "lea -1(%[c]), %[c]            \n\t"
      "jmp 1b                        \n\t"
      "2:                            \n\t"
      : [c]"+c"(c2), [sum]"+r"(sum), [x]"+r"(x2), [y]"+r"(y2), [z]"+r"(z2));


  // copy segment in xp but not in yp
  __asm__ (
      "1:                            \n\t"
      "jecxz 2f                      \n\t"

      "movl -4(%[x],%[c],4), %[sum]  \n\t"
      "adcl $0, %[sum]               \n\t"
      "movl %[sum], -4(%[z],%[c],4)  \n\t"

      "movl -8(%[x],%[c],4), %[sum]  \n\t"
      "adcl $0, %[sum]               \n\t"
      "movl %[sum], -8(%[z],%[c],4)  \n\t"

      "lea -2(%[c]), %[c]            \n\t"
      "jmp 1b                        \n\t"
      "2:                            \n\t"
      : [c]"+c"(c3), [sum]"+r"(sum), [x]"+r"(x3), [z]"+r"(z3));

  __asm__ (
      "1:                            \n\t"
      "jecxz 2f                      \n\t"
      "movl -4(%[x],%[c],4), %[sum]  \n\t"
      "adcl $0, %[sum]               \n\t"
      "movl %[sum], -4(%[z],%[c],4)  \n\t"
      "lea -1(%[c]), %[c]            \n\t"
      "jmp 1b                        \n\t"
      "2:                            \n\t"
      : [c]"+c"(c4), [sum]"+r"(sum), [x]"+r"(x4), [z]"+r"(z4));

  __asm__ (
      "movl $0, %[sum]         \n\t"
      "adcl $0, %[sum]         \n\t"
      "movl %[sum], (%[z])    \n\t"
      : [sum]"+r"(sum), [z]"+r"(zz));

}

void test1() {
    iptr xl = 4;
    iptr yl = 4;

    int *x = malloc(sizeof(int) * xl);
    int *y = malloc(sizeof(int) * yl);
    int *z = malloc(sizeof(int) * (xl + 1));

    int *xp = x + xl - 1;
    int *yp = y + yl - 1;
    int *zp = z + xl;

    int H = 2147483648;  // 2^31

    x[0] = H-1;
    x[1] = H-1;
    x[2] = H-1;
    x[3] = H;

    y[0] = H;
    y[1] = H;
    y[2] = H;
    y[3] = H;

    asm_add(xp, xl, yp, yl, zp);

    for (int j = 0; j < xl + 1; j++) {
        printf("z[%d] = %u\n", j, z[j]);
    }
}


void test2() {
    iptr xl = 4;
    iptr yl = 2;

    int *x = malloc(sizeof(int) * xl);
    int *y = malloc(sizeof(int) * yl);
    int *z = malloc(sizeof(int) * (xl + 1));

    int *xp = x + xl - 1;
    int *yp = y + yl - 1;
    int *zp = z + xl;

    int H = 2147483648;  // 2^31

    x[0] = H+H-1;
    x[1] = H+H-1;
    x[2] = H-1;
    x[3] = H;

    y[0] = H;
    y[1] = H;

    asm_add(xp, xl, yp, yl, zp);

    for (int j = 0; j < xl + 1; j++) {
        printf("z[%d] = %u\n", j, z[j]);
    }
}


void test3() {
    iptr xl = 4;
    iptr yl = 4;

    int *x = malloc(sizeof(int) * xl);
    int *y = malloc(sizeof(int) * yl);
    int *z = malloc(sizeof(int) * (xl + 1));

    int *xp = x + xl - 1;
    int *yp = y + yl - 1;
    int *zp = z + xl;

    int H = 2147483648;  // 2^31

    x[0] = 1;
    x[1] = 2;
    x[2] = 3;
    x[3] = 4;

    y[0] = 1;
    y[1] = 2;
    y[2] = 3;
    y[3] = 4;

    asm_add(xp, xl, yp, yl, zp);

    for (int j = 0; j < xl + 1; j++) {
        printf("z[%d] = %u\n", j, z[j]);
    }
}

void test4() {
    iptr xl = 5;
    iptr yl = 5;

    int *x = malloc(sizeof(int) * xl);
    int *y = malloc(sizeof(int) * yl);
    int *z = malloc(sizeof(int) * (xl + 1));

    int *xp = x + xl - 1;
    int *yp = y + yl - 1;
    int *zp = z + xl;

    int H = 2147483648;  // 2^31

    x[0] = 1;
    x[1] = 2;
    x[2] = 3;
    x[3] = 4;
    x[4] = 5;

    y[0] = 1;
    y[1] = 2;
    y[2] = 3;
    y[3] = 4;
    y[4] = 5;

    asm_add(xp, xl, yp, yl, zp);

    for (int j = 0; j < xl + 1; j++) {
        printf("z[%d] = %u\n", j, z[j]);
    }
}

int main () {
  test1();
  printf("----------------------\n");
  test2();
  printf("----------------------\n");
  test3();
  printf("----------------------\n");
  test4();
}

results.txt

z[0] = 1
z[1] = 0
z[2] = 0
z[3] = 0
z[4] = 0