#include <unistd.h>
#include <stdint.h>
#include <inttypes.h>
#include <stdio.h>

#include "salsarnd.h"

#define ROTATE(v,c) (((v) << (c)) | ((v) >> (32 - (c))))
#define XOR(v,w) ((v) ^ (w))
#define PLUS(v,w) (((v) + (w)))
#define PLUSONE(v) (PLUS((v),1))

#define QUARTERROUND(a,b,c,d) \
  x[a] = PLUS(x[a],x[b]); x[d] = ROTATE(XOR(x[d],x[a]),16); \
  x[c] = PLUS(x[c],x[d]); x[b] = ROTATE(XOR(x[b],x[c]),12); \
  x[a] = PLUS(x[a],x[b]); x[d] = ROTATE(XOR(x[d],x[a]), 8); \
  x[c] = PLUS(x[c],x[d]); x[b] = ROTATE(XOR(x[b],x[c]), 7);

static void salsa20(void *__out, void *__in)
{
  int i;
  uint32_t x[16];
  uint32_t *input = __in;
  uint32_t *output32 = __out;

  for (i = 0;i < 16;++i) x[i] = input[i];
  for (i = 8;i > 0;i -= 2) {
    QUARTERROUND( 0, 4, 8,12)
    QUARTERROUND( 1, 5, 9,13)
    QUARTERROUND( 2, 6,10,14)
    QUARTERROUND( 3, 7,11,15)
    QUARTERROUND( 0, 5,10,15)
    QUARTERROUND( 1, 6,11,12)
    QUARTERROUND( 2, 7, 8,13)
    QUARTERROUND( 3, 4, 9,14)
  }
  for (i = 0;i < 16;++i) x[i] = PLUS(x[i],input[i]);
  for (i = 0;i < 16;++i) output32[i] = x[i];
}

uint32_t get_random32(void)
{
  static __thread uint32_t salsa20_out[16];
  static __thread uint32_t salsa20_in[16];
  static __thread int salsa20_nrints;
  static __thread int salsa20_rndinit;

  if (!salsa20_rndinit) {
    FILE *frandom;

    salsa20_rndinit = 1;
    frandom = fopen("/dev/urandom", "rb");
    if (frandom == NULL) {
      _exit(1);
    }
    setbuf(frandom, NULL);
    if (fread(salsa20_in, 1, sizeof(salsa20_in), frandom) != sizeof(salsa20_in)) {
      _exit(1);
    }
    fclose(frandom);
  }

  if (salsa20_nrints == 0) {
    if (!++salsa20_in[0]) if (!++salsa20_in[1]) if (!++salsa20_in[2]) if (!++salsa20_in[3])
      if (!++salsa20_in[4]) if (!++salsa20_in[5]) if (!++salsa20_in[6]) if (!++salsa20_in[7])
        if (!++salsa20_in[8]) if (!++salsa20_in[9]) if (!++salsa20_in[10]) if (!++salsa20_in[11])  
          if (!++salsa20_in[12]) if (!++salsa20_in[13]) if (!++salsa20_in[14])   ++salsa20_in[15]; 
    salsa20(salsa20_out, salsa20_in);
    salsa20_nrints = 16;
  }

  return salsa20_out[--salsa20_nrints];
}