From: mark_cruzNOTFORSPAM@hotmail.com   
      
   Hello,   
      
   I came across the kernel driver code implementing some sort of IGMP snooping   
   backend and as part of its functionalilty it creates a new socket address   
   family, AF_IGMPSNOOP, but actually implements just a few operations for this   
   type of socket:   
      
   static struct proto_ops igmp_snoop_ops = {   
      family:       AF_IGMP_SNOOP,   
      release:      _igmp_snoop_sock_release,   
      bind:         sock_no_bind,   
      connect:      sock_no_connect,   
      socketpair:   sock_no_socketpair,   
      accept:       sock_no_accept,   
      getname:      sock_no_getname,   
      poll:         datagram_poll,   
      ioctl:        sock_no_ioctl,   
      listen:       sock_no_listen,   
      shutdown:     sock_no_shutdown,   
      setsockopt:   _igmp_snoop_setsockopt,   
      getsockopt:   sock_no_getsockopt,   
      sendmsg:      _igmp_snoop_sock_sendmsg,   
      recvmsg:      _igmp_snoop_sock_recvmsg,   
      mmap:         sock_no_mmap,   
      sendpage:     sock_no_sendpage,   
   }   
      
   However from a user space perspective a socket is created as:   
      
   fd = socket (AF_IGMPSNOOP, SOCK_RAW, IPPROTO_IGMP)   
      
   I'm wondering what is the rationale for adding a new family, while in fact a   
   raw socket is being used to access the stack for IGMP packets? Would not it   
   be the same to just create AF_INET raw socket and do the things?   
      
   Looking forward to hearing form you !   
   Thanks.   
      
   Mark   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)