What is there a need to zero out sockaddr_in structs?

Hi,

I happened to read up on some nix sockets and I'm kinda hooked.

The sample scripts often mentioning about zero-ing up the structs of sockaddr_in.

Code:

---

memset(&(my_addr.sin_zero), ’\0’, 8); // zero the rest of the struct

---

Are there any reasons that remaining portions of structs be "Zeroed"?

Thanks and Regards,
Steven.

The same reason for zeroing any struct, when you pass it to some black-box function, that function may work on some of the items if they are not null, this may break things.

It's good programming practice to zero things even if you think the function dosen't use the extra structure members.

hi

sockaddr_in is written for a convenience use of internet sockets over sockaddr. In order to make the sizes of both structs equal you need 8 more bytes which are zeroed by memset().

Quote:

---

Originally Posted by rockytriton

The same reason for zeroing any struct, when you pass it to some black-box function, that function may work on some of the items if they are not null, this may break things.

---

How can it happen that they are not null when you have just declared a new struct? As in,

Code:

---

struct foo bar = malloc(sizeof(struct foo));

---

malloc doesn't say anything about the content of the memory returned - it's uninitialised.
Debug versions of malloc in particular often fill the memory with a pattern to help detect common malloc related programming errors.

Use calloc if you really want allocated memory full of zero bytes, but beware of the type of memory you're actually allocating.
http://www.eskimo.com/~scs/C-faq/q7.31.html

Aha - thank you.