Gleb Smirnoff 8624f4347e divert: declare PF_DIVERT domain and stop abusing PF_INET
The divert(4) is not a protocol of IPv4.  It is a socket to
intercept packets from ipfw(4) to userland and re-inject them
back.  It can divert and re-inject IPv4 and IPv6 packets today,
but potentially it is not limited to these two protocols.  The
IPPROTO_DIVERT does not belong to known IP protocols, it
doesn't even fit into u_char.  I guess, the implementation of
divert(4) was done the way it is done basically because it was
easier to do it this way, back when protocols for sockets were
intertwined with IP protocols and domains were statically
compiled in.

Moving divert(4) out of inetsw accomplished two important things:

1) IPDIVERT is getting much closer to be not dependent on INET.
   This will be finalized in following changes.
2) Now divert socket no longer aliases with raw IPv4 socket.
   Domain/proto selection code won't need a hack for SOCK_RAW and
   multiple entries in inetsw implementing different flavors of
   raw socket can merge into one without requirement of raw IPv4
   being the last member of dom_protosw.

Differential revision:	https://reviews.freebsd.org/D36379
2022-08-30 15:09:21 -07:00

174 lines
6.9 KiB
Plaintext

# $FreeBSD$
#
# This is an example that shows how to send ASCII formatted control
# messages to a node using ngctl(8).
#
# What we will do here create a divert(4) tap. This simply dumps
# out all packets diverted by some ipfw(8) divert rule to the console.
#
# Lines that begin with ``$'' (shell prompt) or ``+'' (ngctl prompt)
# indicate user input
#
# First, start up ngctl in interactive mode:
$ ngctl
Available commands:
connect Connects hook <peerhook> of the node at <relpath> to <hook>
debug Get/set debugging verbosity level
help Show command summary or get more help on a specific command
list Show information about all nodes
mkpeer Create and connect a new node to the node at "path"
msg Send a netgraph control message to the node at "path"
name Assign name <name> to the node at <path>
read Read and execute commands from a file
rmhook Disconnect hook "hook" of the node at "path"
show Show information about the node at <path>
shutdown Shutdown the node at <path>
status Get human readable status information from the node at <path>
types Show information about all installed node types
quit Exit program
+
# Now let's create a ng_ksocket(4) node, in the family PF_DIVERT,
# of type SOCK_RAW:
+ mkpeer ksocket foo divert/raw/0
# Note that ``foo'' is the hook name on the socket node, which can be
# anything. The ``inet/raw/divert'' is the hook name on the ksocket
# node, which tells it what kind of socket to create.
# Lets give our ksocket node a global name. How about ``fred'':
+ name foo fred
# Note that we used ngctl's ``name'' command to do this. However,
# the following manually constructed netgraph message would have
# accomplished the exact same thing:
+ msg foo name { name="fred" }
# Here we are using the ASCII <-> binary control message conversion
# routines. ngctl does this for us automatically when we use the
# ``msg'' command.
# Now lets bind the socket associated with the ksocket node to a port
# supplied by the system. We do this by sending the ksocket node a
# ``bind'' control message. Again, ngctl does the conversion of the
# control message from ASCII to binary behind the scenes.
+ msg fred: bind inet/192.168.1.1
# The ksocket accepts arbitrary sockaddr structures, but also has
# special support for the PF_LOCAL and PF_INET protocol families.
# That is why we can specify the struct sockaddr argument to the
# ``bind'' command as ``inet/192.168.1.1'' (since we didn't specify
# a port number, it's assumed to be zero). We could have also
# relied on the generic sockaddr syntax and instead said this:
+ msg fred: bind { family=2 len=16 data=[ 2=192 168 1 1 ] }
# This is what you would have to do for protocol families other
# that PF_INET and PF_LOCAL, at least until special handling for
# new ones is added.
# The reason for the ``2=192'' is to skip the two byte IP port number,
# which causes it to be set to zero, the default value for integral
# types when parsing. Now since we didn't ask for a specific port
# number, we need to do a ``getname'' to see what port number we got:
+ msg fred: getname
Rec'd response "getname" (5) from "fred:":
Args: inet/192.168.1.1:1029
# As soon as we sent the message, we got back a response. Here
# ngctl is telling us that it received a control message with the
# NGF_RESP (response) flag set, the response was to a prior ``getname''
# control message, that the originator was the node addressable
# as ``fred:''. The message arguments field is then displayed to
# us in its ASCII form. In this case, what we get back is a struct
# sockaddr, and there we see that our port number is 1029.
# So now let's add the ipfw divert rule for whatever packets we
# want to see. How about anything from 192.168.1.129.
+ ^Z
Suspended
$ ipfw add 100 divert 1029 ip from 192.168.1.129 to any
00100 divert 1029 ip from 192.168.1.129 to any
$ fg
# Now watch what happens when we try to ping from that machine:
+
Rec'd data packet on hook "foo":
0000: 45 00 00 3c 57 00 00 00 20 01 bf ee c0 a8 01 81 E..<W... .......
0010: c0 a8 01 01 08 00 49 5c 03 00 01 00 61 62 63 64 ......I\....abcd
0020: 65 66 67 68 69 6a 6b 6c 6d 6e 6f 70 71 72 73 74 efghijklmnopqrst
0030: 75 76 77 61 62 63 64 65 66 67 68 69 uvwabcdefghi
+
Rec'd data packet on hook "foo":
0000: 45 00 00 3c 58 00 00 00 20 01 be ee c0 a8 01 81 E..<X... .......
0010: c0 a8 01 01 08 00 48 5c 03 00 02 00 61 62 63 64 ......H\....abcd
0020: 65 66 67 68 69 6a 6b 6c 6d 6e 6f 70 71 72 73 74 efghijklmnopqrst
0030: 75 76 77 61 62 63 64 65 66 67 68 69 uvwabcdefghi
+
Rec'd data packet on hook "foo":
0000: 45 00 00 3c 59 00 00 00 20 01 bd ee c0 a8 01 81 E..<Y... .......
0010: c0 a8 01 01 08 00 47 5c 03 00 03 00 61 62 63 64 ......G\....abcd
0020: 65 66 67 68 69 6a 6b 6c 6d 6e 6f 70 71 72 73 74 efghijklmnopqrst
0030: 75 76 77 61 62 63 64 65 66 67 68 69 uvwabcdefghi
+
# So we're seeing the output from the ksocket socket appear on the ``foo''
# hook of ngctl's socket node. Since the packets are getting diverted,
# the 192.168.1.129 machine doesn't see any response from us.
# Of course, any type of socket can be used, even TCP:
+ mkpeer ksocket bar inet/stream/tcp
+ msg bar connect inet/192.168.1.33:13
ngctl: send msg: Operation now in progress
+
Rec'd data packet on hook "foo":
0000: 4d 6f 6e 20 4e 6f 76 20 32 39 20 31 37 3a 34 38 Mon Nov 29 17:48
0010: 3a 33 37 20 31 39 39 39 0d 0a :37 1999..
+
# Or, UNIX domain:
+ mkpeer ksocket bar local/stream/0
+ msg bar bind local/"/tmp/bar.socket"
+
# Here's an example of a more complicated ASCII control message argument.
# If you look in /sys/netgraph/ng_message.h, you will see that a node
# responds to a NGM_LISTHOOKS with a struct hooklist, which contains
# an array of struct linkinfo:
#
# /* Structure used for NGM_LISTHOOKS */
# struct linkinfo {
# char ourhook[NG_HOOKSIZ]; /* hook name */
# char peerhook[NG_HOOKSIZ]; /* peer hook */
# struct nodeinfo nodeinfo;
# };
#
# struct hooklist {
# struct nodeinfo nodeinfo; /* node information */
# struct linkinfo link[0]; /* info about each hook */
# };
#
# By sending a node the ``listhooks'' command using ngctl, we can see
# this structure in ASCII form (lines wrapped for readability):
+ msg bar bind local/"/tmp/bar.socket"
+ msg bar listhooks
Rec'd response "listhooks" (7) from "bar":
Args: { nodeinfo={ type="ksocket" id=9 hooks=1 }
linkinfo=[ { ourhook="local/stream/0" peerhook="bar"
nodeinfo={ name="ngctl1327" type="socket" id=8 hooks=1 } } ] }