c++ - WinSock2 sockaddr structure as a parameter -

what point of having functions such inet_ntop, wsaaddresstostring, recvfrom require either size of actual sockaddr structure or address family provided?

the first 2 bytes of sockaddr structure indicate address family , therefore indicate whether it's sockaddr_in or sockaddr_in6.

so reason additional sizeof(sockaddr_in/sockaddr_in6) , af_inet/af_inet6?

sockaddr legacy type. there newer sockaddr_... types specific address families, sockaddr_in ipv4 , sockaddr_in6 ipv6, , have different byte sizes. apis operate on generic sockaddr* pointers partly historic reasons , partly flexibility. socket apis recognize multiple address types. have provide sizes apis can validate memory buffers large enough pass address data based on specified families (af_inet ipv4, af_inet6 ipv6).

in inet_ntop(), in_addr , in6_addr not contain address family, why have pass separately. responsible making sure specify correct address family input type passing in.

in wsaaddresstostring(), sockaddr* pointer input buffer. have pass allocated size of buffer. instance, since sockaddr_in smaller sockaddr_in6, if allocate sockaddr_in set family af_inet6 instead of af_inet, buffer small function access of ipv6 fields, fail wsaefault error instead of crashing.

in recvfrom(), sockaddr* pointer output buffer. have preallocate , pass allocated size. if allocate sockaddr_in receive ipv6 packet, sender's address not fit in buffer , function fail wsaefault error.