Background
OS: Ubuntu 16.04 x64 running on VirtualBox
I am a developer with minimal Ubuntu/Linux knowledge and have been assigned to a project where the goal is to utilize tcpcrypt when communicating with certain endpoints.
tcpcrypt comes with a shell script which sets required entries in iptables to route packets to tcpcrypt for encrpt/decrypt. After execution of this script iptables looks like:
filter
Chain INPUT (policy ACCEPT 4 packets, 552 bytes)
pkts bytes target prot opt in out source destination
0 0 ACCEPT tcp -- lo any anywhere anywhere tcp dpt:65530 tos match0x22/0xff
0 0 NFQUEUE tcp -- any any !localhost anywhere tcp dpt:65530 flags:FIN,SYN,RST,PSH,ACK,URG/SYN NFQUEUE num 666
0 0 NFQUEUE tcp -- any any anywhere anywhere multiport sports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s tcp flags:FIN,SYN,RST,PSH,ACK,URG/SYN,ACK NFQUEUE num 666
Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
pkts bytes target prot opt in out source destination
Chain OUTPUT (policy ACCEPT 4 packets, 536 bytes)
pkts bytes target prot opt in out source destination
0 0 NFQUEUE tcp -- any any anywhere anywhere multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s tos match0x04/0xff owner UID match tcpcryptd NFQUEUE num 666
0 0 NFQUEUE tcp -- any any anywhere anywhere tcp spt:65530 flags:FIN,SYN,RST,PSH,ACK,URG/SYN,ACK NFQUEUE num 666
nat
Chain PREROUTING (policy ACCEPT)
target prot opt source destination
REDIRECT tcp -- multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s redir ports 65530
Chain INPUT (policy ACCEPT)
target prot opt source destination
Chain OUTPUT (policy ACCEPT)
target prot opt source destination
ACCEPT tcp -- anywhere anywhere multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s owner UID match tcpcryptd
REDIRECT tcp -- anywhere anywhere multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s redir ports 65530
mangle
Chain PREROUTING (policy ACCEPT)
target prot opt source destination
Chain INPUT (policy ACCEPT)
target prot opt source destination
Chain FORWARD (policy ACCEPT)
target prot opt source destination
Chain OUTPUT (policy ACCEPT)
target prot opt source destination
Chain POSTROUTING (policy ACCEPT)
target prot opt source destination
TOS all -- anywhere anywhere tos match0x04/0xff TOS and 0x00
With these entries, every packet is put on a queue where tcpcrypt picks for enc/dec.
UPDATE
This is the script for iptables:
#!/bin/sh
#DAEMON_USER DIVERT_PORT ONLY_PORTS OMIT_PORTS
# determine which operation is requested (Append or Delete)
if [ "$1" = "start" -o -z "$1" ]; then
# during startup, bail early if any of these commands fails
set -e
OP="-A"
elif [ "$1" = "stop" -o "$1" = "-f" ] ; then
OP="-D"
else
echo "Expected \"start\" or \"stop\" as first argument" >&2
exit 1
fi
# determine which ports should be tcpcrypt-enabled
if [ -z "$ONLY_PORTS" -a -z "$OMIT_PORTS" ] ; then
echo "Expected either OMIT_PORTS or ONLY_PORTS environment variables to be set" >&2
exit 1
fi
if [ -n "$ONLY_PORTS" -a -n "$OMIT_PORTS" ] ; then
echo "Expected only one of OMIT_PORTS or ONLY_PORTS environment variables to be set" >&2
exit 1
fi
if [ -n "$OMIT_PORTS" ] ; then
PORT_TEST=!
PORTS="$OMIT_PORTS"
fi
if [ -n "$ONLY_PORTS" ] ; then
PORT_TEST=
PORTS="$ONLY_PORTS"
fi
# more necessary configuration
if [ -z "$DAEMON_USER" ] ; then
echo "Expected DAEMON_USER environment variable to be set" >&2
exit 1
fi
if [ -z "$DIVERT_PORT" ] ; then
echo "Expected DIVERT_PORT environment variable to be set" >&2
exit 1
fi
# some shorthand to make rules more concise
from_enabled_port="-m multiport $PORT_TEST --source-ports $PORTS"
to_enabled_port="-m multiport $PORT_TEST --destination-ports $PORTS"
NFQUEUE="NFQUEUE --queue-num $DIVERT_PORT"
CRYPT_PORT="65530"
REDIRECT="REDIRECT --to-port $CRYPT_PORT"
INJECT_TOS="0x22"
HANDSHAKE_TOS="0x04"
filter="$ECHO iptables -t filter $OP"
# Injection from daemon: Accept
$filter INPUT -i lo -p tcp --dport $CRYPT_PORT \
-m tos --tos $INJECT_TOS \
-j ACCEPT
# SYN redirected to daemon:
# Queue for daemon to initiate proxy connection with original destination
$filter INPUT -p tcp --dport $CRYPT_PORT --tcp-flags ALL SYN \
-j $NFQUEUE
# SYN+ACK on proxy connection:
# Queue for daemon to complete original handshake
$filter INPUT -p tcp $from_enabled_port --tcp-flags ALL SYN,ACK \
-j $NFQUEUE
# Handshake packet of proxy connection from daemon:
# Queue for daemon to set tcp options via DIVERT_MODIFY
$filter OUTPUT -p tcp $to_enabled_port \
-m tos --tos $HANDSHAKE_TOS \
-m owner --uid-owner $DAEMON_USER \
-j $NFQUEUE
# SYN+ACK on redirected connection:
# Queue for daemon to delay handshake until proxy connection succeeds
$filter OUTPUT -p tcp --sport $CRYPT_PORT --tcp-flags ALL SYN,ACK \
-j $NFQUEUE
nat="$ECHO iptables -t nat $OP"
# Inbound connection for enabled ports:
# Redirect to daemon (at localhost:$CRYPT_PORT) for encryption
#
# (The nat module will now translate addresses in both directions,
# for the lifetime of this connection.)
$nat PREROUTING -p tcp $to_enabled_port \
-j $REDIRECT
# Proxy connection from daemon to enabled port: Accept
$nat OUTPUT -p tcp $to_enabled_port \
-m owner --uid-owner $DAEMON_USER \
-j ACCEPT
# Outbound connections to enabled ports on remote hosts:
# Redirect to daemon (at localhost port $CRYPT_PORT) for encryption
#
# (The nat module will now translate addresses in both directions,
# for the lifetime of this connection.)
$nat OUTPUT \! -o lo -p tcp $to_enabled_port \
-j $REDIRECT
mangle="$ECHO iptables -t mangle $OP"
# Packets leaving the machine with bookkeeping mark: Remove mark
$mangle POSTROUTING -m tos --tos $HANDSHAKE_TOS \
-j TOS --set-tos 0x00
Question
How should I modify iptables
with current(see above) entries to achieve the following restrictions:
- Only packets with a certain destination should be queued for tcpcrypt.
- All other packets should not be queued for tcpcrypt and must travel freely.
What I have tried
A) I tried adding desired IP address to tcp destination in OUTPUT
chain which looks like:
Chain OUTPUT (policy ACCEPT 4 packets, 536 bytes)
pkts bytes target prot opt in out source destination
0 0 NFQUEUE tcp -- any any anywhere XXX.XXX.XXX.XXX multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s tos match0x04/0xff owner UID match tcpcryptd NFQUEUE num 666
0 0 NFQUEUE tcp -- any any anywhere XXX.XXX.XXX.XXX tcp spt:65530 flags:FIN,SYN,RST,PSH,ACK,URG/SYN,ACK NFQUEUE num 666
B) I tried adding source and destination parameters to NAT rules:
target prot opt source destination
REDIRECT tcp -- XXX.XXX.XXX.XXX XXX.XXX.XXX.XXX multiport dports !ssh,261,https,nntps,614,ldaps,684,695,ftps-data,ftps,telnets:pop3s redir ports 65530
But still all packets regardless of destination address, are sent to tcpcrypt.
iptables --list -n -v
?iptables -t TABLE -L -v
for each of filter, nat and mangle tables. Itried your suggested-n
, with this option all of source/destination values are displayed as0.0.0.0/0