1.\" $Id: divert.4,v 1.11 1998/05/25 08:44:28 julian Exp $ 2.\" 3.Dd June 18, 1996 4.Dt DIVERT 4 5.Os FreeBSD 6.Sh NAME 7.Nm divert 8.Nd kernel packet diversion mechanism 9.Sh SYNOPSIS 10.Fd #include <sys/types.h> 11.Fd #include <sys/socket.h> 12.Fd #include <netinet/in.h> 13.Ft int 14.Fn socket PF_INET SOCK_RAW IPPROTO_DIVERT 15.Sh DESCRIPTION 16.Pp 17Divert sockets are similar to raw IP sockets, except that they 18can be bound to a specific 19.Nm 20port via the 21.Xr bind 2 22system call. The IP address in the bind is ignored; only the port 23number is significant. 24A divert socket bound to a divert port will receive all packets diverted 25to that port by some (here unspecified) kernel mechanism(s). 26Packets may also be written to a divert port, in which case they 27re-enter kernel IP packet processing. 28.Pp 29Divert sockets are normally used in conjunction with 30FreeBSD's packet filtering implementation and the 31.Xr ipfw 8 32program. By reading from and writing to a divert socket, matching packets 33can be passed through an arbitrary ``filter'' as they travel through 34the host machine, special routing tricks can be done, etc. 35.Sh READING PACKETS 36Packets are diverted either as they are ``incoming'' or ``outgoing.'' 37Incoming packets are diverted after reception on an IP interface, 38whereas outgoing packets are diverted before next hop forwarding. 39.Pp 40Diverted packets may be read unaltered via 41.Xr read 2 , 42.Xr recv 2 , 43or 44.Xr recvfrom 2 . 45In the latter case, the address returned will have its port set to 46the some tag supplied by the packet diverter, (usually the ipfw rule number) 47and the IP address set to the (first) address of 48the interface on which the packet was received (if the packet 49was incoming) or 50.Dv INADDR_ANY 51(if the packet was outgoing). In the case of an incoming packet the interface 52name will also be placed in the 8 bytes following the address, 53(assuming it fits). 54.Sh WRITING PACKETS 55Writing to a divert socket is similar to writing to a raw IP socket; 56the packet is injected ``as is'' into the normal kernel IP packet 57processing and minimal error checking is done. 58Packets are written as either incoming or outgoing: 59if 60.Xr write 2 61or 62.Xr send 2 63is used to deliver the packet, or if 64.Xr sendto 2 65is used with a destination IP address of 66.Dv INADDR_ANY , 67then the packet is treated as if it were outgoing, i.e., destined 68for a non-local address. Otherwise, the packet is assumed to be 69incoming and full packet routing is done. 70.Pp 71In the latter case, the 72IP address specified must match the address of some local interface, 73or an interface name 74must be found after the IP address. If an interface name is found, 75that interface will be used and the value of the IP address will be 76ignored (other than the fact that it is not 77.Dv INADDR_ANY 78). 79This is to indicate on which interface the packet ``arrived.'' 80.Pp 81Normally, packets read as incoming should be written as incoming; 82similarly for outgoing packets. When reading and then writing back 83packets, passing the same socket address supplied by 84.Xr recvfrom 2 85unmodified to 86.Xr sendto 2 87simplifies things (see below). 88.Pp 89The port part of the socket address passed to the 90.Xr sendto 2 91contains a tag that should be meaningful to the diversion module. In the 92case of 93.Xr Ipfw 8 94the tag is interpretted as the rule number 95.Em after which 96rule processing should restart. 97.Sh LOOP AVOIDANCE 98Packets written into a divert socket (using 99.Xr sendto 2 100) Re-enter the packet filter at the rule number 101following the tag given in the port part of the socket address, which 102is usually already set at the rule number that caused the diversion 103(not the next rule if there are several at the same number). If the 'tag' 104is altered to indicate an alternative re-entry point, care should be taken 105to avoid loops, where the same packet is diverted more than once at the 106same rule. 107.Sh DETAILS 108To enable divert sockets, your kernel must be compiled with the option 109.Dv IPDIVERT . 110.Pp 111If a packet is diverted but no socket is bound to the 112port, or if 113.Dv IPDIVERT 114is not enabled in the kernel, the packet is dropped. 115.Pp 116Incoming packet fragments which get diverted are fully reassembled 117before delivery; the diversion of any one fragment causes the entire 118packet to get diverted. 119If different fragments divert to different ports, 120then which port ultimately gets chosen is unpredictable. 121.Pp 122Packets are received and sent unchanged, except that 123packets written as outgoing have their IP header checksums overwritten 124with the correct value. 125Packets written as incoming and having incorrect checksums will be dropped. 126Otherwise, all header fields are unchanged (and therefore in network order). 127.Pp 128Binding to port numbers less than 1024 requires super-user access, as does 129creating a socket of type SOCK_RAW. 130.Sh ERRORS 131Writing to a divert socket can return these errors, along with 132the usual errors possible when writing raw packets: 133.Bl -tag -width Er 134.It Bq Er EINVAL 135The packet had an invalid header, or the IP options in the packet 136and the socket options set were incompatible. 137.It Bq Er EADDRNOTAVAIL 138The destination address contained an IP address not equal to 139.Dv INADDR_ANY 140that was not associated with any interface. 141.El 142.Sh SEE ALSO 143.Xr bind 2 , 144.Xr recvfrom 2 , 145.Xr sendto 2 , 146.Xr socket 2 , 147.Xr ipfw 8 148.Sh BUGS 149This is an attempt to provide a clean way for user mode processes 150to implement various IP tricks like address translation, but it 151could be cleaner, and it's too dependent on 152.Xr ipfw 8 . 153.Pp 154It's questionable whether incoming fragments should be reassembled 155before being diverted. For example, if only some fragments of a 156packet destined for another machine don't get routed through the 157local machine, the packet is lost. This should probably be 158a settable socket option in any case. 159.Sh AUTHORS 160.An Archie Cobbs Aq archie@whistle.com , 161Whistle Communications Corp. 162