Raw sockets allow new IPv4 protocols to be implemented in user space. A raw socket receives or sends the raw datagram not including link level headers.
The IPv4 layer generates an IP header when sending a packet unless the IP_HDRINCL socket option is enabled on the socket. When it is enabled, the packet must contain an IP header. For receiving the IP header is always included in the packet.
Only processes with an effective user id of 0 or the CAP_NET_RAW capability are allowed to open raw sockets.
All packets or errors matching the protocol number specified for the raw socket are passed to this socket. For a list of the allowed protocols see RFC1700 assigned numbers and getprotobyname(3).
A protocol of IPPROTO_RAW implies enabled IP_HDRINCL and receives all IP protocols. Sending is not allowed.
If IP_HDRINCL is specified and the IP header has a non-zero destination address then the destination address of the socket is used to route the packet. When MSG_DONTROUTE is specified the destination address should refer to a local interface, otherwise a routing table lookup is done anyways but gatewayed routes are ignored.
In Linux 2.2 all IP header fields and options can be set using IP socket options. This means raw sockets are usually only needed for new protocols or protocols with no user interface (like ICMP).
Enable a special filter for raw sockets bound to the IPPROTO_ICMP protocol. The value has a bit set for each ICMP message type which should be filtered out. The default is to filter no ICMP messages.
Raw sockets fragment a packet when its total length exceeds the interface MTU (but see BUGS). A more network friendly and faster alternative is to implement path MTU discovery as described in the IP_PMTU_DISCOVER section of ip(7).
A raw socket can be bound to a specific local address using the bind(2) call. If it isn't bound all packets with the specified IP protocol are received. In addition a RAW socket can be bound to a specific network device using SO_BINDTODEVICE; see socket(7).
An IPPROTO_RAW socket is send only. If you really want to receive all IP packets use a packet(7) socket with the ETH_P_IP protocol. Note that packet sockets don't reassemble IP fragments, unlike raw sockets.
If you want to receive all ICMP packets for a datagram socket it is often better to use IP_RECVERR on that particular socket; see ip(7).
Raw sockets may tap all IP protocols in Linux, even protocols like ICMP or TCP which have a protocol module in the kernel. In this case the packets are passed to both the kernel module and the raw socket(s). This should not be relied upon in portable programs, many other BSD socket implementation have limitations here.
Linux never changes headers passed from the user (except for filling in some zeroed fields as described for IP_HDRINCL). This differs from many other implementations of raw sockets.
RAW sockets are generally rather unportable and should be avoided in programs intended to be portable.
Packet too big. Either Path MTU Discovery is enabled (the IP_PMTU_DISCOVER socket flag) or the packet size exceeds the maximum allowed IPv4 packet size of 64KB.
User tried to send to a broadcast address without having the broadcast flag set on the socket.
An ICMP error has arrived reporting a parameter problem.
An invalid memory address was supplied.
Invalid flag has been passed to a socket call (like MSG_OOB).
IP_RECVERR and ICMP_FILTER are new in Linux 2.2. They are Linux extensions and should not be used in portable programs.
Transparent proxy extensions are not described.
When the IP_HDRINCL option is set datagrams will not be fragmented and are limited to the interface MTU. This is a limitation in Linux 2.2.
RFC1191 for path MTU discovery.