Question

From http://www.doc.ic.ac.uk/teaching/projects/Distinguished03/AndrewSuffield.pdf:

#include <string.h>

struct x
{
  char s[10];
  int a[4];
};

void bang(struct x *d)
{
  strcat(d->s, "!");
}

int main(void)
{
  struct x d;
  strcpy(d.s, "012345678");
  d.a[0] = 3;
  d.a[1] = 2;
  d.a[2] = 1;
  d.a[3] = 0;
  bang(&d);
  return a[0];
}

In this example, struct x contains a 10-byte string immediately followed by a 4-integer array. d is initialized with a 9-character string (occupying 10 bytes because of the trailing NULL) and four integers. bang() appends a ! to the string, making it “012345678!” plus a trailing NULL.

The NULL byte at the end of the string will overwrite the first byte of d.a[0]. On a
big-endian host, this will have no effect because that byte was already zero. On a little-endian host, this will change the value of d.a[0] to zero

Two questions:

1. Will there not be structure holes present between s & a and so the above argument does not hold. gcc gives return vaue as 3.
2. return a[0] does not work on my system (gcc).

Answer 1

1. There could very well be padding between the fields. On any modern system there will be. Try passing strcat a longer string.
2. return a[0] is clearly a typo. It should read return d.a[0].