sockaddr_in, sockaddr_in6, sockaddr_storage when use which ?

[test_goraz]

Hi all,

I am playing with some socket programming and one thing confuse me. For ipv4 we have struct sockaddr_in which is able to handle everything related to ipv4 sockets. On other side for ipv6 we have struct sockaddr_in6, which do the same thing for ipv6. It is known that ipv6 created with
socket ( AF_INET, SOCK_STREAM,0) is able to accept connections from ipv4 and ipv6 clients if there is not specificaly written IPV6_V6ONLY in declaration of address type.
For me is confusing the structure
sockaddr_storage ( I read that is used to write address family independent code and that is ok ) but if I have some application for which is necessary to accept connection from ipv4 and ipv6 clients then I can use struct sockaddr_in6, and the application with that structure will work ? Do we need struct sockaddr_storage only on server side when we do not have information will our application be running on ipv4 or ipv6 server.

I am sorry, if this post looks confusing but, it is completealy mess for me to understand this. If someone have knowledge and will to share it thanks in advance

Kind regards,

[rick jones]

Generally speaking, you allocate something sized/aligned like sockaddr_storage and then use casts to either sockaddr_in or sockaddr_in6. The sockaddr_in* structures both have the "family" field in the same place and of the same size so even if your sockaddr_storage happends to be holding a sockaddr_in6, you can still cast it to a sockaddr_in for the purposes of checking the sin_family field. If that is then AF_INET6 you would then switch to using a sockaddr_in6 cast of the sockaddr_storage.

You might want to get the latest edition of Stevens, Fenner and Rudoff's Unix Network Programming. It will have lots of examples for you.

[test_goraz]

Thanks for your answer. I will try to find the book you recommended me. If I have code

struct sockaddr_in saddr;
bzero((char *) &saddr, sizeof(saddr));
saddr.sin_family = AF_INET;
saddr.sin_port = htons((u_short) service);
saddr.sin_addr.s_addr = INADDR_ANY;

I am just wondering how to addapt it using struct sockaddr_storage to be address independent ?

Thanks for answer

[rick jones]

make saddr a pointer to sockaddr_in and then point it at a newly added struct sockaddr_storage foo

struct sockaddr_storage foo;

struct sockaddr_in *saddr;

sadder - (struct sockaddr_in *)&foo;

...

converting the rest of the code with the knowledge that saddr is now a pointer to a struct sockaddr_in.