Aliasing and/or other non-conformant behaviour

[Hodor]

Hello,

Hodor here.

In the code fragment below is there any behaviour that is not defined?

#include <stdio.h>

struct foo {
    int a, b;
};

struct bar {
    struct foo coord;
    const char *description;
};

void init(struct foo *p)
{
    p->a = 42;
    p->b = 84;
}

int main(void)
{
    struct bar obj;
    struct foo *p;
    struct bar *p2;
    
    init((struct foo *) &obj);
    obj.description = "test";
    
    p = (struct foo *) &obj;
    p2 = (struct bar *)p;
    
    printf("%d %d %s\n", p2->coord.a, p2->coord.b, p2->description);
    
    return 0;
}

Edit: I'm not asking whether or not there is any "dangerous" behaviour by the way, I'm only interested in conformance (or not) with the C Standard.

[Salem]

6.7.2.1 Structure and union specifiers
12 Within a structure object, the non-bit-field members and the units in which bit-fields
reside have addresses that increase in the order in which they are declared. A pointer to a
structure object, suitably converted, points to its initial member (or if that member is a
bit-field, then to the unit in which it resides), and vice versa. There may be unnamed
padding within a structure object, but not at its beginning.

It seems a pointer to a struct and a pointer to the first member would always be the same.

[Hodor]

It seems a pointer to a struct and a pointer to the first member would always be the same.

That's how I'm reading it as well. E.g. (struct foo *) &obj == &obj.coord and therefore aliasing rules are not applicable?

Edit: I'm asking about strict-aliasing because someone mentioned that this code snippet might not be conforming due to the aliasing rules. But I am reading/interpreting the standard differently (and think that it's fine) and we could not come to a consensus.

Edit 2: gcc (with -fstrict-aliasing -Wall -Wextra -Wpedantic) says nothing (although that doesn't mean much in itself) but splint reports no applicable errors or warnings either, so the only way forward seems to be the C Standard (C99 or higher) and discussion

Edit 3: My argument is that the types are compatible because of the section you referenced and the equivalence I used as an example. I.e. They are a pointer to the same type (even though in the first instance there needs to be an explicit cast)

[Salem]

A pointer (to the whole struct) and a pointer to the first member can never be the same type.
They are however the same value.

#include <stdio.h>

struct foo {
  int bar;
};

int main ( ) {
  struct foo f;
  struct foo *fp = &f;
  int *ip = &f.bar;
  
  // There is no aliasing here
  *ip = 42;
  fp->bar = 99;

  printf("%p %p\n", fp, ip );
  return 0;
}

fp and ip point to the same memory location, but we haven't coerced any pointers into a different type.
Both assignments refer to the same int, as an int.

Understanding Strict Aliasing - CellPerformance