Re: strict aliasing and pointer to struct

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



nrmj <maa1666@xxxxxxxx> writes:

> I have been googling a lot on the web about alias
> analysis and strict aliasing, but haven't found much
> about what is allowed or not when working with
> structures.

I don't think these questions have anything to do with gcc.  I
recommend that you ask on some place like comp.std.c.

> 1) If I understand the C standard correctly, the
> compiler is free to assume that two pointers pointing
> to different types don't alias.
> 
>    If two struct have different tag name, like:
>     struct s1 {int i;};
>     struct s2 {int i;};
>     struct s1 *p1;
>     struct s2 *p2;
> 
>    p1 and p2 are considered by the compiler to point
> to different locations and don't alias, because p1
> points to type "struct s1" and p2 points to type
> "struct s2" which are different types, even if they
> have the same members, right ?

You have to watch out for the exception in C99 6.5.2.3 if there is a
visible union declaration which contains both structs.  But otherwise,
yes, they are different types.

> 2) Is it safe to cast pointer to struct B to pointer
> to struct A, in order to fake inheritance, like in
> this code sniplet ?

No, that is not safe.

But it is safe to do

  struct Color_Point { struct Point p; Color color; };
  p = &my_color_point.p;

> 3) When programming with socket, the following cast to
> (struct sockaddr *) is very common.
>    But is it really safe ?
> 
>    struct sockaddr_in my_addr;
>    ...
>    bind(sockfd, (struct sockaddr *)&my_addr,
> sizeof(struct sockaddr));

That is safe because the code does not dereference the resulting
pointer.

Ian

[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux