shorewall_code/Shorewall/firewall

#!/bin/sh
#
#     The Shoreline Firewall (Shorewall) Packet Filtering Firewall - V1.4 3/14/2003
#
#     This program is under GPL [http://www.gnu.org/copyleft/gpl.htm]
#
#     (c) 1999,2000,2001,2002,2003 - Tom Eastep (teastep@shorewall.net)
#
#	On most distributions, this file should be called:
#	/etc/rc.d/init.d/shorewall or /etc/init.d/shorewall
#
#	Complete documentation is available at http://shorewall.net
#
#	This program is free software; you can redistribute it and/or modify
#	it under the terms of Version 2 of the GNU General Public License
#	as published by the Free Software Foundation.
#
#	This program is distributed in the hope that it will be useful,
#	but WITHOUT ANY WARRANTY; without even the implied warranty of
#	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
#	GNU General Public License for more details.
#
#	You should have received a copy of the GNU General Public License
#	along with this program; if not, write to the Free Software
#	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA
#
#	If an error occurs while starting or restarting the firewall, the
#	firewall is automatically stopped.
#
#	Commands are:
#
#	   shorewall start			  Starts the firewall
#	   shorewall restart			  Restarts the firewall
#	   shorewall stop			  Stops the firewall
#	   shorewall status			  Displays firewall status
#	   shorewall reset			  Resets iptabless packet and
#						  byte counts
#	   shorewall clear			  Remove all Shorewall chains
#						  and rules/policies.
#	   shorewall refresh	.		  Rebuild the common chain
#          shorewall check                        Verify the more heavily-used
#                                                 configuration files.
#
# Search a list looking for a match -- returns zero if a match found
# 1 otherwise
#
list_search() # $1 = element to search for , $2-$n = list
{
    local e=$1

    while [ $# -gt 1 ]; do
	shift
	[ "x$e" = "x$1" ] && return 0
    done

    return 1
}

#
# Function to count list elements
#
list_count() {
    arg_count() {
	echo $#
    }

    arg_count `separate_list $1`
}

#
# Mutual exclusion -- These functions are jackets for the mutual exclusion
#		      routines in $FUNCTIONS. They invoke
#		      the corresponding function in that file if the user did
#		      not specify "nolock" on the runline.
#
my_mutex_on() {
    [ -n "$nolock" ] || { mutex_on; have_mutex=Yes; }
}

my_mutex_off() {
    [ -n "$have_mutex" ] && { mutex_off; have_mutex=; }
}

#
# Message to stderr
#
error_message() # $* = Error Message
{
   echo "   $@" >&2
}

#
# Fatal error -- stops the firewall after issuing the error message
#
fatal_error() # $* = Error Message
{
    echo "   Error: $@" >&2
    if [ $command = check ]; then
	[ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
    else
	stop_firewall
    fi
    exit 2
}

#
# Fatal error during startup -- generate an error message and abend with
#				altering the state of the firewall
#
startup_error() # $* = Error Message
{
    echo "   Error: $@" >&2
    my_mutex_off
    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
    kill $$
    exit 2
}

#
# Send a message to STDOUT and the System Log
#
report () { # $* = message
    echo "$@"
    logger "$@"
}

#
# Perform variable substitution on the passed argument and echo the result
#
expand() # $1 = contents of variable which may be the name of another variable
{
    eval echo \"$1\"
}

#
# Perform variable substitition on the values of the passed list of variables
#
expandv() # $* = list of variable names
{
    local varval

    while [ $# -gt 0 ]; do
	eval varval=\$${1}
	eval $1=\"$varval\"
	shift
    done
}

#
# Replace all leading "!" with "! " in the passed argument list
#

fix_bang() {
    local i;

    for i in $@; do
	case $i in
	    !*)
		echo "! ${i#!}"
		;;
	    *)
		echo $i
		;;
	esac
    done
}

#
# Run iptables and if an error occurs, stop the firewall and quit
#
run_iptables() {

    if ! iptables $@ ; then
	[ -z "$stopping" ] && { stop_firewall; exit 2; }
    fi
}

#
# Version of 'run_iptables' that inserts white space after "!" in the arg list
#
run_iptables2() {

    if [ "x${*%!*}" = "x$*" ]; then
	#
	# No "!" in the command -- just execute it
	#
	run_iptables $@
	return
    fi
    #
    # Need to insert white space before each "!"
    #
    run_iptables `fix_bang $@`
}

#
# Run ip and if an error occurs, stop the firewall and quit
#
run_ip() {
    if ! ip $@ ; then
	[ -z "$stopping" ] && { stop_firewall; exit 2; }
    fi
}

#
# Run arp and if an error occurs, stop the firewall and quit
#
run_arp() {
    if ! arp $@ ; then
	[ -z "$stopping" ] && { stop_firewall; exit 2; }
    fi
}

#
# Run tc and if an error occurs, stop the firewall and quit
#
run_tc() {
    if ! tc $@ ; then
	[ -z "$stopping" ] && { stop_firewall; exit 2; }
    fi
}

#
# Create a filter chain
#
# If the chain isn't one of the common chains then add a rule to the chain
# allowing packets that are part of an established connection. Create a
# variable ${1}_exists and set its value to Yes to indicate that the chain now
# exists.
#
createchain() # $1 = chain name, $2 = If "yes", create default rules
{
    run_iptables -N $1

    if [ $2 = yes ]; then
	run_iptables -A $1 -m state --state ESTABLISHED,RELATED -j ACCEPT
	[ -z "$NEWNOTSYN" ] && \
	    run_iptables -A $1 -m state --state NEW -p tcp ! --syn -j newnotsyn
    fi

    eval ${1}_exists=Yes
}

#
# Determine if a chain exists
#
# When we create a chain "chain", we create a variable named chain_exists and
# set its value to Yes. This function tests for the "_exists" variable
# corresponding to the passed chain having the value of "Yes".
#
havechain() # $1 = name of chain
{
    eval test \"\$${1}_exists\" = Yes
}

#
# Query NetFilter about the existence of a filter chain
#
chain_exists() # $1 = chain name
{
    qt iptables -L $1 -n
}

#
# Query NetFilter about the existence of a mangle chain
#
mangle_chain_exists() # $1 = chain name
{
    qt iptables -t mangle -L $1 -n
}

#
# Ensure that a chain exists (create it if it doesn't)
#
ensurechain() # $1 = chain name
{
    havechain $1 || createchain $1 yes
}

#
# Add a rule to a chain creating the chain if necessary
#
addrule() # $1 = chain name, remainder of arguments specify the rule
{
    ensurechain $1
    run_iptables -A $@
}

#
# Create a nat chain
#
# Create a variable ${1}_nat_exists and set its value to Yes to indicate that
# the chain now exists.
#
createnatchain() # $1 = chain name
{
    run_iptables -t nat -N $1

    eval ${1}_nat_exists=Yes
}

#
# Determine if a nat chain exists
#
# When we create a chain "chain", we create a variable named chain_nat_exists
# and set its value to Yes. This function tests for the "_exists" variable
# corresponding to the passed chain having the value of "Yes".
#
havenatchain() # $1 = name of chain
{
    eval test \"\$${1}_nat_exists\" = Yes
}

#
# Ensure that a nat chain exists (create it if it doesn't)
#
ensurenatchain() # $1 = chain name
{
    havenatchain $1 || createnatchain $1
}

#
# Add a rule to a nat chain creating the chain if necessary
#
addnatrule() # $1 = chain name, remainder of arguments specify the rule
{
    ensurenatchain $1
    run_iptables2 -t nat -A $@
}

#
# Delete a chain if it exists
#
deletechain() # $1 = name of chain
{
    qt iptables -L $1 -n && qt iptables -F $1 && qt iptables -X $1
}

#
# Determine if a chain is a policy chain
#
is_policy_chain() # $1 = name of chain
{
    eval test \"\$${1}_is_policy\" = Yes
}

#
# Set a standard chain's policy
#
setpolicy() # $1 = name of chain, $2 = policy
{
    run_iptables -P $1 $2
}

#
# Set a standard chain to enable established and related connections
#
setcontinue() # $1 = name of chain
{
    run_iptables -A $1 -m state --state ESTABLISHED,RELATED -j ACCEPT
}

#
# Flush one of the NAT table chains
#
flushnat() # $1 = name of chain
{
    run_iptables -t nat -F $1
}

#
# Flush one of the Mangle table chains
#
flushmangle() # $1 = name of chain
{
    run_iptables -t mangle -F $1
}

#
# Chain name base for an interface
#
chain_base() #$1 = interface
{
    local c=${1%%+*}

    case $c in
        *.*)
	    echo ${c%.*}_${c#*.}
	    ;;
	*)
	    echo ${c:=common}
	    ;;
    esac
}

#
# Find interfaces to a given zone
#
# Search the variables representing the contents of the interfaces file and
# for each record matching the passed ZONE, echo the expanded contents of
# the "INTERFACE" column
#
find_interfaces() # $1 = interface zone
{
    local zne=$1
    local z
    local interface

    for interface in $all_interfaces; do
	eval z=\$`chain_base ${interface}`_zone
	[ "x${z}" = x${zne} ] && echo $interface
    done
}

#
# Forward Chain for an interface
#
forward_chain() # $1 = interface
{
   echo `chain_base $1`_fwd
}

#
# Input Chain for an interface
#
input_chain() # $1 = interface
{
   echo `chain_base $1`_in
}

#
# Output Chain for an interface
#
output_chain() # $1 = interface
{
   echo `chain_base $1`_out
}

#
# Masquerade Chain for an interface
#
masq_chain() # $1 = interface
{
   echo `chain_base $1`_masq
}

#
# MAC Verification Chain for an interface
#
mac_chain() # $1 = interface
{
   echo `chain_base $1`_mac
}

#
# DNAT Chain from a zone
#
dnat_chain() # $1 = zone
{
   echo ${1}_dnat
}

#
# SNAT Chain to a zone
#
snat_chain() # $1 = zone
{
   echo `chain_base $1`_snat
}

#
# ECN Chain to an interface
#
ecn_chain() # $1 = interface
{
   echo ${1}_ecn
}

#
# First chains for an interface
#
first_chains() #$1 = interface
{
   local c=`chain_base $1`

   echo ${c}_fwd ${c}_in
}

#
# Find hosts in a given zone
#
# Read hosts file and for each record matching the passed ZONE,
# echo the expanded contents of the "HOST(S)" column
#
find_hosts() # $1 = host zone
{
    local hosts

    while read z hosts options; do
	[ "x`expand $z`" = "x$1" ] && expandv hosts && echo `separate_list $hosts`
    done < $TMP_DIR/hosts
}

#
# Determine the interfaces on the firewall
#
# For each zone, create a variable called ${zone}_interfaces. This
# variable contains a space-separated list of interfaces to the zone
#
determine_interfaces() {
    for zone in $zones; do
	interfaces=`find_interfaces $zone`
	interfaces=`echo $interfaces` # Remove extra trash
	eval ${zone}_interfaces=\"\$interfaces\"
    done
}

#
# Determine the defined hosts in each zone and generate report
#
determine_hosts() {

    for zone in $zones; do
	hosts=`find_hosts $zone`
	hosts=`echo $hosts` # Remove extra trash

	eval interfaces=\$${zone}_interfaces

	for interface in $interfaces; do
	    if [ -z "$hosts" ]; then
		hosts=$interface:0.0.0.0/0
	    else
		hosts="$hosts $interface:0.0.0.0/0"
	    fi
	done

	interfaces=

	for host in $hosts; do
	    interface=${host%:*}
	    if ! list_search $interface $interfaces; then
		if [ -z "$interfaces" ]; then
		    interfaces=$interface
		else
		    interfaces="$interfaces $interface"
		fi
	    fi

	    [ "${host#*:}" = "0.0.0.0/0" ] || \
		eval ${zone}_is_complex=Yes
	done

	eval ${zone}_interfaces="\$interfaces"
	eval ${zone}_hosts="\$hosts"

	if [ -n "$hosts" ]; then
	    eval display=\$${zone}_display
	    display_list "$display Zone:" $hosts
	else
	    error_message "Warning: Zone $zone is empty"
	fi
    done
}

#
# Ensure that the passed zone is defined in the zones file or is the firewall
#
validate_zone() # $1 = zone
{
    list_search $1 $zones $FW
}

#
# Validate the zone names and options in the interfaces file
#
validate_interfaces_file() {
    while read z interface subnet options; do
	expandv z interface subnet options
	r="$z $interface $subnet $options"

	[ "x$z" = "x-" ] && z=

	if [ -n "$z" ]; then
	    validate_zone $z || startup_error "Invalid zone ($z) in record \"$r\""
	fi

	if [ -n "`ip link show $interface 2> /dev/null | grep LOOPBACK`" ]; then
	    startup_error "The loopback interface ($interface) may not be defined in /etc/shorewall/interfaces"
	fi

	list_search $interface $all_interfaces && \
	    startup_error "Duplicate Interface $interface"

	case $interface in
	    *:*)
		startup_error "Invalid Interface Name: $interface"
		;;
	esac

	all_interfaces="$all_interfaces $interface"
	options=`separate_list $options`
	interface=`chain_base $interface`

	eval ${interface}_broadcast="$subnet"
	eval ${interface}_zone="$z"
	eval ${interface}_options=\"$options\"

	for option in $options; do
	    case $option in
		dhcp|norfc1918|tcpflags|newnotsyn)
		    ;;
		routefilter|dropunclean|logunclean|blacklist|proxyarp|maclist|-)
		    ;;
		routeback)
		    [ -n "$z" ] || startup_error "The routeback option may not be specified on a multi-zone interface"
		    eval ${z}_routeback=\"$interface:0.0.0.0/0 \$${z}_routeback\"
		    ;;
		*)
		    error_message "Warning: Invalid option ($option) in record \"$r\""
		    ;;
	    esac
	done

    [ -z "$all_interfaces" ] && startup_error "No Interfaces Defined"

    done < $TMP_DIR/interfaces
}

#
# Validate the zone names and options in the hosts file
#
validate_hosts_file() {
    while read z hosts options; do
	expandv z hosts options
	r="$z $hosts $options"
	validate_zone $z || startup_error "Invalid zone ($z) in record \"$r\""

	for host in `separate_list $hosts`; do
	    interface=${host%:*}

	    list_search  $interface $all_interfaces || \
		startup_error "Unknown interface ($interface) in record \"$r\""

	    for option in `separate_list $options`; do
		case $option in
		    maclist|-)
			;;
		    routeback)
			eval ${z}_routeback=\"$host \$${z}_routeback\"
			;;
		    *)
			error_message "Warning: Invalid option ($option) in record \"$r\""
			;;
		esac
	    done
	done
    done < $TMP_DIR/hosts
}

#
# Format a match by the passed MAC address
# The passed address begins with "~" and uses "-" as a separator between bytes
# Example: ~01-02-03-04-05-06
#
mac_match() # $1 = MAC address formated as described above
{
    echo "--match mac --mac-source `echo $1 | sed 's/~//;s/-/:/g'`"
}

#
# validate the policy file
#
validate_policy()
{
    local clientwild
    local serverwild
    local zone
    local zone1
    local pc
    local chain
    local policy
    local loglevel
    local synparams

    print_policy() # $1 = source zone, $2 = destination zone
    {
	[ $command != check ] || \
	    [ $1 = $2 ]  || \
	    [ $1 = all ] || \
	    [ $2 = all ] || \
	    echo "   Policy for $1 to $2 is $policy"
    }

    all_policy_chains=

    strip_file policy

    while read client server policy loglevel synparams; do
	expandv client server policy loglevel synparams

	clientwild=
	serverwild=

	case "$client" in
	all|ALL)
	    clientwild=Yes
	    ;;
	*)
	    if ! validate_zone $client; then
		startup_error "Undefined zone $client"
	    fi
	esac

	case "$server" in
	all|ALL)
	    serverwild=Yes
	    ;;
	*)
	    if ! validate_zone $server; then
		startup_error "Undefined zone $server"
	    fi
	esac

	case $policy in
	ACCEPT|REJECT|DROP|CONTINUE|NONE)
	    ;;
	*)
	    startup_error "Invalid policy $policy"
	    ;;
	esac

	chain=${client}2${server}

	[ "x$chain" = "x${FW}2${FW}" ] && \
	    startup_error "fw->fw policy not allowed: $policy"

	if is_policy_chain $chain ; then
	    startup_error "Duplicate policy $policy"
	fi

	[ "x$loglevel" = "x-" ] && loglevel=

	chain=${client}2${server}

	[ $policy = NONE ] || all_policy_chains="$all_policy_chains $chain"

	eval ${chain}_is_policy=Yes
	eval ${chain}_policy=$policy
	eval ${chain}_loglevel=$loglevel
	eval ${chain}_synparams=$synparams

	if [ -n "${clientwild}" ]; then
	    if [ -n "${serverwild}" ]; then
		for zone in $zones $FW all; do
		    for zone1 in $zones $FW all; do
			eval pc=\$${zone}2${zone1}_policychain

			if [ -z "$pc" ]; then
			    eval ${zone}2${zone1}_policychain=$chain
			    eval ${zone}2${zone1}_policy=$policy
			    print_policy $zone $zone1
			fi
		    done
		done
	    else
		for zone in $zones $FW all; do
		    eval pc=\$${zone}2${server}_policychain

		    if [ -z "$pc" ]; then
			eval ${zone}2${server}_policychain=$chain
			eval ${zone}2${server}_policy=$policy
			print_policy $zone $server
		    fi
		done
	    fi
	elif [ -n "$serverwild" ]; then
	    for zone in $zones $FW all; do
		eval pc=\$${client}2${zone}_policychain

		    if [ -z "$pc" ]; then
			eval ${client}2${zone}_policychain=$chain
			eval ${client}2${zone}_policy=$policy
			print_policy $client $zone
		    fi
	    done
	else
	    eval ${chain}_policychain=${chain}
	    print_policy $client $server
	fi

    done < $TMP_DIR/policy
}

#
# Find broadcast addresses
#
find_broadcasts() {
    for interface in $all_interfaces; do
	eval bcast=\$`chain_base $interface`_broadcast
	if [ "x$bcast" = "xdetect" ]; then
	    addr="`ip addr show $interface 2> /dev/null`"
	    if [ -n "`echo "$addr" | grep 'inet.*brd '`" ]; then
		addr="`echo "$addr" | \
		grep "inet " | sed 's/^.* inet.*brd //;s/scope.*//'`"
		echo $addr | cut -d' ' -f 1
	    fi
	elif [ "x${bcast}" != "x-" ]; then
	    echo `separate_list $bcast`
	fi
    done
}

#
# Find interface address--returns the first IP address assigned to the passed
# device
#
find_interface_address() # $1 = interface
{
    #
    # get the line of output containing the first IP address
    #
    addr=`ip addr show $1 2> /dev/null | grep inet | head -n1`
    #
    # If there wasn't one, bail out now
    #
    [ -n "$addr" ] || fatal_error "Can't determine the IP address of $1"
    #
    # Strip off the trailing VLSM mask (or the peer IP in case of a P-t-P link)
    # along with everything else on the line
    #
    echo $addr | sed 's/inet //;s/\/.*//;s/ peer.*//'
}

#
# Find interface addresses--returns the set of addresses assigned to the passed
# device
#
find_interface_addresses() # $1 = interface
{
    ip addr show $1 | grep inet | sed 's/inet //;s/\/.*//;s/ peer.*//'
}

#
# Find interfaces that have the passed option specified
#
find_interfaces_by_option() # $1 = option
{
    for interface in $all_interfaces; do
	eval options=\$`chain_base ${interface}`_options
	list_search $1 $options && echo $interface
    done
}

#
# Find hosts with the passed option
#
find_hosts_by_option() # $1 = option
{
    while read ignore hosts options; do
	expandv options
	list_search $1 `separate_list $options` && \
	    echo `expand $hosts`
    done < $TMP_DIR/hosts

    for interface in $all_interfaces; do
	eval options=\$`chain_base ${interface}`_options
	list_search $1 $options && \
	    echo ${interface}:0.0.0.0/0
    done
}

#
# Determine if there are interfaces of the given zone and option
#
# Returns zero if any such interfaces are found and returns one otherwise.
#
have_interfaces_in_zone_with_option() # $1 = zone, $2 = option
{
    local zne=$1
    local z
    local interface

    for interface in $all_interfaces; do
	eval z=\$`chain_base ${interface}`_zone

	[ "x$z" = "x$zne" ] && \
	    list_search $1 $options && \
	    return 0
    done

    return 1
}

#
# Flush and delete all user-defined chains in the filter table
#
deleteallchains() {
    run_iptables -F
    run_iptables -X
}

#
# Source a user exit file if it exists
#
run_user_exit() # $1 = file name
{
    local user_exit=`find_file $1`

    if [ -f $user_exit ]; then
	echo "Processing $user_exit ..."
	. $user_exit
    fi
}

#
# Add a logging rule.
#
log_rule() # $1 = log level, $2 = chain, $3 = disposition , $... = predicates for the rule
{
    local level=$1
    local chain=$2
    local disposition=$3
    local rulenum=

    shift;shift;shift

    if [ -n "$LOGRULENUMBERS" ]; then
	eval rulenum=\$${chain}_logrules

	[ -z "$rulenum" ] && rulenum=1

	case $level in
	    ULOG)
		eval iptables -A $chain $@ -j ULOG $LOGPARMS --ulog-prefix '"`printf "$LOGFORMAT" $chain $rulenum $disposition`"'
		;;
	    *)
		eval iptables -A $chain $@ -j LOG $LOGPARMS --log-level $level --log-prefix '"`printf "$LOGFORMAT" $chain $rulenum $disposition`"'
		;;
	esac
	
	if [ $? -ne 0 ] ; then
	    [ -z "$stopping" ] && { stop_firewall; exit 2; }
	fi

	rulenum=$(($rulenum + 1))

	eval ${chain}_logrules=$rulenum
    else
	case $level in
	    ULOG)
		eval iptables -A $chain $@ -j ULOG $LOGPARMS --ulog-prefix '"`printf "$LOGFORMAT" $chain $disposition`"'
		;;
	    *)
		eval iptables -A $chain $@ -j LOG $LOGPARMS --log-level $level --log-prefix '"`printf "$LOGFORMAT" $chain $disposition`"'
		;;
	esac
	
	if [ $? -ne 0 ] ; then
	    [ -z "$stopping" ] && { stop_firewall; exit 2; }
	fi
    fi
}

#
# Stop the Firewall
#
stop_firewall() {
    #
    # Turn off trace unless we were tracing "stop" or "clear"
    #
    case $command in
	stop|clear)
	    ;;
	check)
	    kill $$
	    exit 2
	    ;;
	*)
	    set +x
	    ;;
    esac

    stopping="Yes"

    terminator=

    deletechain shorewall

    run_user_exit stop

    [ -n "$MANGLE_ENABLED" ] && \
	run_iptables -t mangle -F && \
	run_iptables -t mangle -X

    [ -n "$NAT_ENABLED" ] && delete_nat
    delete_proxy_arp
    [ -n "$CLEAR_TC" ] && delete_tc

    setpolicy INPUT DROP
    setpolicy OUTPUT DROP
    setpolicy FORWARD DROP

    deleteallchains

    hosts=

    strip_file routestopped

    while read interface host; do
	expandv interface host
	[ "x$host" = "x-" -o -z "$host" ] && host=0.0.0.0/0
	for h in `separate_list $host`; do
	    hosts="$hosts $interface:$h"
	done
    done < $TMP_DIR/routestopped

    for host in $hosts; do
	interface=${host%:*}
	subnet=${host#*:}
	iptables -A INPUT  -i $interface -s $subnet -j ACCEPT
	iptables -A OUTPUT -o $interface -d $subnet -j ACCEPT

	for host1 in $hosts; do
	    [ "$host" != "$host1" ] && \
		iptables -A FORWARD -i $interface -s $subnet \
		    -o ${host1%:*} -d ${host1#*:} -j ACCEPT
	done
    done

    iptables -A INPUT  -i lo -j ACCEPT
    iptables -A OUTPUT -o lo -j ACCEPT

    for interface in `find_interfaces_by_option dhcp`; do
	iptables -A INPUT  -p udp -i $interface --dport 67:68 -j ACCEPT
	iptables -A OUTPUT -p udp -o $interface --dport 67:68 -j ACCEPT
    done

    case "$IP_FORWARDING" in
    [Oo][Nn])
	echo 1 > /proc/sys/net/ipv4/ip_forward
	;;
    [Oo][Ff][Ff])
	echo 0 > /proc/sys/net/ipv4/ip_forward
	;;
    esac

    run_user_exit stopped

    logger "Shorewall Stopped"

    rm -rf $TMP_DIR

    case $command in
    stop|clear)
	;;
    *)
	#
	# The firewall is being stopped when we were trying to do something
	# else. Remove the lock file and Kill the shell in case we're in a
	# subshell
	#
	my_mutex_off
	kill $$
	;;
    esac
}

#
# Remove all rules and remove all user-defined chains
#
clear_firewall() {
    stop_firewall

    run_iptables -F

    echo 1 > /proc/sys/net/ipv4/ip_forward

    setpolicy INPUT ACCEPT
    setpolicy FORWARD ACCEPT
    setpolicy OUTPUT ACCEPT

    run_user_exit clear

    logger "Shorewall Cleared"
}

#
# Set up ipsec tunnels
#
setup_tunnels() # $1 = name of tunnels file
{
    local inchain
    local outchain

    setup_one_ipsec() # $1 = gateway $2 = Tunnel Kind $3 = gateway zones
    {
	options="-m state --state NEW -j ACCEPT"
	addrule $inchain	  -p 50	 -s $1 -j ACCEPT
	addrule $outchain	  -p 50	 -d $1 -j ACCEPT
	run_iptables -A $inchain  -p 51	 -s $1 -j ACCEPT
	run_iptables -A $outchain -p 51	 -d $1 -j ACCEPT

	run_iptables -A $outchain -p udp -d $1 --dport 500 --sport 500 $options

	if [ $2 = ipsec ]; then
	    run_iptables -A $inchain  -p udp -s $1 --sport 500 --dport 500 $options
	else
	    run_iptables -A $inchain  -p udp -s $1 --dport 500 $options
	    run_iptables -A $inchain  -p udp -s $1 --dport 4500 $options
	fi

	for z in `separate_list $3`; do
	    if validate_zone $z; then
		addrule ${FW}2${z} -p udp --sport 500 --dport 500 $options
		if [ $2 = ipsec ]; then
		    addrule ${z}2${FW} -p udp --sport 500 --dport 500 $options
		else
		    addrule ${z}2${FW} -p udp --dport 500 $options
		    addrule ${z}2${FW} -p udp --dport 4500 $options
		fi
	    else
		error_message "Warning: Invalid gateway zone ($z)" \
		" -- Tunnel \"$tunnel\" may encounter keying problems"
	    fi
	done

	echo "   IPSEC tunnel to $gateway defined."
    }

    setup_one_other() # $1 = TYPE, $2 = gateway, $3 = protocol
    {
	addrule $inchain  -p $3 -s $2 -j ACCEPT
	addrule $outchain -p $3 -d $2 -j ACCEPT

	echo "   $1 tunnel to $2 defined."
    }

    setup_pptp_client() # $1 = gateway
    {
	addrule $outchain -p 47               -d $1 -j ACCEPT
	addrule $inchain  -p 47               -j ACCEPT
	addrule $outchain -p tcp --dport 1723 -d $1 -j ACCEPT

	echo "   PPTP tunnel to $1 defined."
    }

    setup_pptp_server()
    {
	addrule $inchain  -p 47               -j ACCEPT
	addrule $outchain -p 47               -j ACCEPT
	addrule $inchain  -p tcp --dport 1723 -j ACCEPT

	echo "   PPTP server defined."
    }

    setup_one_openvpn() # $1 = gateway, $2 = kind[:port]
    {
	case $2 in
	    *:*)
		p=${2#*:}
		;;
	    *)
		p=5000
		;;
	esac

	addrule $inchain  -p udp -s $1 --sport $p --dport $p -j ACCEPT
	addrule $outchain -p udp -d $1 --sport $p --dport $p -j ACCEPT

	echo "   OPENVPN tunnel to $1:$p defined."
    }

    strip_file tunnels $1

    while read kind z gateway z1; do
	expandv kind z gateway z1
	tunnel="`echo $kind $z $gateway $z1`"
	if validate_zone $z; then
	    inchain=${z}2${FW}
	    outchain=${FW}2${z}
	    case $kind in
		ipsec|IPSEC)
		    setup_one_ipsec $gateway ipsec $z1
		    ;;
		ipsecnat|IPSECNAT)
		    setup_one_ipsec $gateway ipsecnat $z1
		    ;;
		ipip|IPIP)
		    setup_one_other IPIP $gateway 4
		    ;;
		gre|GRE)
		    setup_one_other GRE $gateway 47
		    ;;
		6to4|6TO4)
		    setup_one_other 6to4 $gateway 41
		    ;;
		pptpclient|PPTPCLIENT)
		    setup_pptp_client $gateway
		    ;;
		pptpserver|PPTPSERVER)
		    setup_pptp_server
		    ;;
		openvpn|OPENVPN|openvpn:*|OPENVPN:*)
		    setup_one_openvpn $gateway $kind
		    ;;
		*)
		    error_message "Tunnels of type $kind are not supported:" \
			"Tunnel \"$tunnel\" Ignored"
		    ;;
	    esac
	else
	    error_message "Invalid gateway zone ($z)" \
		" -- Tunnel \"$tunnel\" Ignored"
	fi
    done < $TMP_DIR/tunnels
}

#
# Setup Proxy ARP
#
setup_proxy_arp() {

    print_error() {
	error_message "Invalid value for HAVEROUTE - ($haveroute)"
	error_message "Entry \"$address $interface $external $haveroute\" ignored"
    }

    setup_one_proxy_arp() {
	case $haveroute in
	[Nn][Oo])
	    haveroute=
	    ;;
	[Yy][Ee][Ss])
	    ;;
	*)
	    if [ -n "$haveroute" ]; then
		print_error
		return
	    fi
	    ;;
	esac

	[ -z "$haveroute" ] && run_ip route add $address dev $interface

	run_arp -Ds $address $external pub

	echo 1 > /proc/sys/net/ipv4/conf/$interface/proxy_arp
	echo 0 > /proc/sys/net/ipv4/conf/$external/proxy_arp

	echo $address $interface $external $haveroute >> ${STATEDIR}/proxyarp

	echo "   Host $address connected to $interface added to ARP on $external"
    }

    > ${STATEDIR}/proxyarp

    while read address interface external haveroute; do
	expandv address interface external haveroute
	setup_one_proxy_arp
    done < $TMP_DIR/proxyarp

    interfaces=`find_interfaces_by_option proxyarp`

    for interface in $interfaces; do
	if echo 1 > /proc/sys/net/ipv4/conf/$interface/proxy_arp 2> /dev/null; then
	    echo "   Enabled proxy ARP on $interface"
        else
	    error_message "Warning: Unable to enable proxy ARP on $interface"
        fi
    done
}

#
# Set up MAC Verification
#
setup_mac_lists() {
    local interface
    local mac
    local addresses
    local address
    local chain
    local logpart
    local macpart
    local blob
    local hosts
    #
    # Generate the list of interfaces having MAC verification
    #
    maclist_interfaces=

    for hosts in $maclist_hosts; do
	interface=${hosts%:*}
	if ! list_search $interface $maclist_interfaces; then\
	    if [ -z "$maclist_interfaces" ]; then
		maclist_interfaces=$interface
	    else
		maclist_interfaces="$maclist_interfaces $interface"
	    fi
	fi
    done

    echo "Setting up MAC Verification on $maclist_interfaces..."
    #
    # Be sure that they are all ethernet interfaces
    #
    for interface in $maclist_interfaces; do
	case $interface in
	    eth*|wlan*)
		;;
	    *)
		fatal_error "MAC verification is only supported on ethernet and 802.11b devices: $interface"
		;;
	esac

	createchain `mac_chain $interface` no
    done
    #
    # Process the maclist file producing the verification rules
    #

    while read interface mac addresses; do
	expandv interface mac addresses

	chain=`mac_chain $interface`

	if ! havechain $chain ; then
	    fatal_error "No hosts on $interface have the maclist option specified"
	fi

	macpart=`mac_match $mac`

	if [ -z "$addresses" ]; then
	    run_iptables -A $chain $macpart -j RETURN
	else
	    for address in `separate_list $addresses` ; do
		run_iptables2 -A $chain $macpart -s $address -j RETURN
	    done
	fi
    done < $TMP_DIR/maclist
    #
    # Must take care of our own broadcasts and multicasts then terminate the verification
    # chains
    #
    for interface in $maclist_interfaces; do
	chain=`mac_chain $interface`
	blob=`ip addr show $interface 2> /dev/null | grep inet | sed 's/inet //; s/brd //; s/scope.*//;'`

	[ -z "$blob" ] && \
	    fatal_error "Interface $interface must be up before Shorewall can start"

	set -- $blob

	while [ $# -gt 0 ]; do
	    address=${1%/*}

	    case $1 in
		*/32)
		    ;;
		*)
		    run_iptables -A $chain -s $address -d $2 -j RETURN
		    shift
		    ;;
	    esac

	    run_iptables -A $chain -s $address -d 255.255.255.255 -j RETURN
	    run_iptables -A $chain -s $address -d 224.0.0.0/4     -j RETURN
	    shift
	done

	if [ -n "$MACLIST_LOG_LEVEL" ]; then
	    log_rule $MACLIST_LOG_LEVEL $chain $MACLIST_DISPOSITION 
	fi

	run_iptables -A $chain -j $maclist_target
    done
    #
    # Generate jumps from the input and forward chains
    #
    for hosts in $maclist_hosts; do
	interface=${hosts%:*}
	hosts=${hosts#*:}
	for chain in `first_chains $interface` ; do
	    run_iptables -A $chain -s $hosts -m state --state NEW \
		-j `mac_chain $interface`
	done
    done
}

#
# Set up SYN flood protection
#
setup_syn_flood_chain ()
	# $1 = policy chain
	# $2 = synparams
{
    local chain=$1
    local limit=${2%:*}
    local limit_burst=${2#*:}

    run_iptables -N @$chain
    run_iptables -A @$chain \
	-m limit --limit $limit --limit-burst $limit_burst \
	-j RETURN
    run_iptables -A @$chain -j DROP
}

#
# Enable SYN flood protection on a chain
#
# Insert a jump rule to the protection chain from the first chain. Inserted
# as the second rule and restrict the jump to SYN packets
#
enable_syn_flood_protection() # $1 = chain, $2 = protection chain
{
    run_iptables -I $1 2 -p tcp --syn -j @$2
    echo "      Enabled SYN flood protection"
}

#
# Delete existing Proxy ARP
#
delete_proxy_arp() {
    if [ -f ${STATEDIR}/proxyarp ]; then
	while read address interface external haveroute; do
	    qt arp -i $external -d $address pub
	    [ -z "$haveroute" ] && qt ip route del $address dev $interface
	done < ${STATEDIR}/proxyarp

	rm -f ${STATEDIR}/proxyarp
    fi

    [ -d ${STATEDIR} ] && touch ${STATEDIR}/proxyarp

    for f in `ls /proc/sys/net/ipv4/conf/*/proxy_arp`; do
	echo 0 > $f
    done
}

#
# Setup Static Network Address Translation (NAT)
#
setup_nat() {
    local allints
    #
    # At this point, we're just interested in the network translation
    #
    > ${STATEDIR}/nat

    echo "Setting up NAT..."

    while read external interface internal allints localnat; do
	expandv external interface internal allints localnat

	iface=${interface%:*}

	if [ -n "$ADD_IP_ALIASES" ]; then
	    qt ip addr del $external dev $iface
	fi

	if [ -z "$allints" -o "$allints" = "Yes" -o "$allints" = "yes" ]
	then
	    addnatrule nat_in  -d $external -j DNAT --to-destination $internal
	    addnatrule nat_out -s $internal -j SNAT --to-source $external

	    if [ "$localnat" = "Yes" -o "$localnat" = "yes" ]; then
		run_iptables2 -t nat -A OUTPUT -d $external \
		    -j DNAT --to-destination $internal
	    fi
	else
	    addnatrule `input_chain $iface` \
		-d $external -j DNAT --to-destination $internal
	    addnatrule `output_chain $iface` \
		-s $internal -j SNAT --to-source $external
	fi

	if [ -n "$ADD_IP_ALIASES" ]; then
	    list_search $external $aliases_to_add || \
		aliases_to_add="$aliases_to_add $external $interface"
	fi

	echo "   Host $internal NAT $external on $interface"
    done < $TMP_DIR/nat
}

#
# Delete existing Static NAT
#
delete_nat() {
    run_iptables -t nat -F
    run_iptables -t nat -X

    if [ -f ${STATEDIR}/nat ]; then
	while read external interface; do
	    qt ip addr del $external dev $interface
	done < ${STATEDIR}/nat

	rm -f {$STATEDIR}/nat
    fi

    [ -d ${STATEDIR} ] && touch ${STATEDIR}/nat
}

#
# Setup ECN disabling rules
#
setup_ecn() # $1 = file name
{
    local interfaces=""
    local hosts
    local h

    strip_file ecn $1

    echo "Processing $1..."

    while read interface host; do
	expandv interface host
	list_search  $interface $all_interfaces || \
	    startup_error "Unknown interface $interface"
	list_search $interface $interfaces || \
	    interfaces="$interfaces $interface"
	[ "x$host" = "x-" ] && host=
	for h in `separate_list ${host:-0.0.0.0/0}`; do
	    hosts="$hosts $interface:$h"
	done
    done < $TMP_DIR/ecn

    if [ -n "$interfaces" ]; then
	echo "Setting up ECN control on${interfaces}..."

	for interface in $interfaces; do
	    chain=`ecn_chain $interface`
	    if mangle_chain_exists $chain; then
		flushmangle $chain
	    else
		run_iptables -t mangle -N $chain
		run_iptables -t mangle -A POSTROUTING -p tcp -o $interface -j $chain
		run_iptables -t mangle -A OUTPUT      -p tcp -o $interface -j $chain
	    fi
	done

	for host in $hosts; do
	    interface=${host%:*}
	    h=${host#*:}
	    run_iptables -t mangle -A `ecn_chain $interface` -p tcp -d $h -j ECN --ecn-tcp-remove
	    echo "   ECN Disabled to $h through $interface"
	done
    fi
}

#
# Process a TC Rule - $marking_chain is assumed to contain the name of the
#                     default marking chain
#
process_tc_rule()
{
    add_a_tc_rule() {
	r=
	chain=$marking_chain

	if [ "x$source" != "x-"	 ]; then
	    case $source in
		*.*.*)
		    r="-s $source "
		    ;;
		~*)
		    r=`mac_match $source`
		    ;;
		$FW)
		    chain=tcout
		    ;;
		*)
		    if ! list_search $source $all_interfaces; then
			fatal_error "Unknown interface $source in rule \"$rule\""
		    fi

		    r="-i $source "
		    ;;
	    esac
	fi

	if [ "$mark" != "${mark%:*}" ]; then

	    [ "$chain" = tcout ] && \
		fatal_error "Chain designator not allowed when source is \$FW; rule \"$rule\""

	    case "${mark#*:}" in
		p|P)
		    chain=tcpre
		    ;;
		f|F)
		    chain=tcfor
		    ;;
		*)
		    fatal_error "Invalid chain designator: (${mark#*:}) in rule \"$rule\""
		    ;;
	    esac

	    mark="${mark%:*}"
	fi

	[ "x$dest"   = "x-"  ] || r="${r}-d $dest "
	[ "$proto"   = "all" ] || r="${r}-p $proto "
	[ "x$port"   = "x-"  ] || r="${r}--dport $port "
	[ "x$sport"  = "x-"  ] || r="${r}--sport $sport "

	run_iptables2 -t mangle -A $chain $r -j MARK --set-mark $mark

    }

    for source in `separate_list ${sources:=-}`; do
	for dest in `separate_list ${dests:=-}`; do
	    for port in `separate_list ${ports:=-}`; do
		for sport in `separate_list ${sports:=-}`; do
		    add_a_tc_rule
		done
	    done
	done
    done

    echo "   TC Rule \"$rule\" added"
}

#
# Setup queuing and classes
#
setup_tc1() {
    #
    # Create the TC mangle chains
    #

    run_iptables -t mangle -N tcpre
    run_iptables -t mangle -N tcfor
    run_iptables -t mangle -N tcout
    #
    # Process the TC Rules File
    #
    strip_file tcrules

    while read mark sources dests proto ports sports; do
	expandv mark sources dests proto ports sports
	rule=`echo "$mark $sources $dests $proto $ports $sports"`
	process_tc_rule
    done < $TMP_DIR/tcrules
    #
    # Link to the TC mangle chains from the main chains
    #

    run_iptables -t mangle -A FORWARD    -j tcfor
    run_iptables -t mangle -A PREROUTING -j tcpre
    run_iptables -t mangle -A OUTPUT     -j tcout

    run_user_exit tcstart

}

setup_tc() {

    echo "Setting up Traffic Control Rules..."

    setup_tc1
}

#
# Clear Traffic Shaping
#
delete_tc()
{

    clear_one_tc() {
	tc qdisc del dev $1 root    2> /dev/null
	tc qdisc del dev $1 ingress 2> /dev/null
    }

    run_user_exit tcclear

    run_ip link list | \
    while read inx interface details; do
	case $inx in
	    [0-9]*)
		clear_one_tc ${interface%:}
		;;
	    *)
		;;
	esac
    done
}

#
# Check the configuration
#
check_config() {

    disclaimer() {
	echo
	echo "Notice:  The 'check' command is unsupported and problem"
	echo "         reports complaining about errors that it didn't catch"
	echo "         will not be accepted"
	echo
    }

    disclaimer

    report_capabilities

    echo "Verifying Configuration..."

    verify_os_version

    load_kernel_modules

    echo "Determining Zones..."

    determine_zones

    [ -z "$zones" ] && startup_error "ERROR: No Zones Defined"

    display_list "Zones:" $zones

    echo "Validating interfaces file..."

    validate_interfaces_file

    echo "Validating hosts file..."

    validate_hosts_file

    echo "Determining Hosts in Zones..."

    determine_interfaces
    determine_hosts

    echo "Validating policy file..."

    validate_policy

    echo "Validating rules file..."

    rules=`find_file rules`
    strip_file rules $rules
    process_rules

    rm -rf $TMP_DIR

    echo "Configuration Validated"

    disclaimer

}

#
# Refresh queuing and classes
#
refresh_tc() {

    echo "Refreshing Traffic Control Rules..."

    [ -n "$CLEAR_TC" ] && delete_tc

    [ -n "$MARK_IN_FORWARD_CHAIN" ] && chain=tcfor || chain=tcpre

    if mangle_chain_exists $chain; then
        #
        # Flush the TC mangle chains
        #
	run_iptables -t mangle -F $chain

        run_iptables -t mangle -F tcout
        #
        # Process the TC Rules File
        #
        strip_file tcrules

        while read mark sources dests proto ports sports; do
	    expandv mark sources dests proto ports sports
	    rule=`echo "$mark $sources $dests $proto $ports $sports"`
	    process_tc_rule
        done < $TMP_DIR/tcrules

	run_user_exit tcstart
    else
	setup_tc1
    fi

}

#
# Add a NAT rule - Helper function for the rules file processor
#
# The caller has established the following variables:
#    command        = The current command -- if 'check', we just go through
#                     the motions.
#    cli	    = Source IP, interface or MAC Specification
#    serv	    = Destination IP Specification
#    servport	    = Port the server is listening on
#    dest_interface = Destination Interface Specification
#    proto	    = Protocol Specification
#    addr	    = Original Destination Address
#    dports	    = Destination Port Specification. 'dports' may be changed
#		      by this function
#    cport	    = Source Port Specification
#    multiport	    = String to invoke multiport match if appropriate
#
add_nat_rule() {
    local chain
    local excludedests=

    # Be sure we can NAT

    if [ -z "$NAT_ENABLED" ]; then
	fatal_error "Rule \"$rule\" requires NAT which is disabled"
    fi

    # Parse SNAT address if any

    if [ "$addr" != "${addr%:*}" ]; then
	snat="${addr#*:}"
	addr="${addr%:*}"
    else
	snat=""
    fi

    # Set original destination address

    case $addr in
	all)
	    addr=
	    ;;
	detect)
	    addr=
	    if [ -n "$DETECT_DNAT_IPADDRS" -a "$source" != "$FW" ]; then
		eval interfaces=\$${source}_interfaces
		for interface in $interfaces; do
		    addr=${addr:+$addr,}`find_interface_address $interface`
		done
	    fi
	    ;;
	!*)
	    if [ `list_count $addr` -gt 1 ]; then
		excludedests="`separate_list ${addr#\!}`"
		addr=
	    fi
	    ;;
    esac

    addr=${addr:-0.0.0.0/0}

    # Select target

    if [ -n "$serv" ]; then
	servport="${servport:+:$servport}"
	serv1=
	for srv in `separate_list $serv`; do
	    serv1="$serv1 --to-destination ${srv}${servport}"
	done
	target1="DNAT $serv1"
    else
	target1="REDIRECT --to-port $servport"
    fi

    if [ $source = $FW ]; then
	[ -n "$excludezones" ] && fatal_error "Invalid Source in rule \"$rule\""
    fi

    # Generate nat table rules

    if [ $command != check ]; then
	if [ "$source" = "$FW" ]; then
	    if [ -n "$excludedests" ]; then
		chain=nonat${nonat_seq}
		nonat_seq=$(($nonat_seq + 1))
		createnatchain $chain
		run_iptables -t nat -A OUTPUT $cli $proto $multiport $sports $dports -j $chain

		for adr in $excludedests; do
		    addnatrule $chain -d $adr -j RETURN
		done

		if [ -n "$loglevel" ]; then
		    log_rule $loglevel $chain $logtarget -t nat
		fi

		addnatrule $chain -j $target1
	    else
		for adr in `separate_list $addr`; do
		    run_iptables2 -t nat -A OUTPUT $proto $sports -d $adr \
			$multiport $dports -j $target1
		done
	    fi
	else
	    chain=`dnat_chain $source`

	    if [ -n "${excludezones}${excludedests}" ]; then
		chain=nonat${nonat_seq}
		nonat_seq=$(($nonat_seq + 1))
		createnatchain $chain
		addnatrule `dnat_chain $source` $cli $proto $multiport $sports $dports -j $chain
		for z in $excludezones; do
		    eval hosts=\$${z}_hosts
		    for host in $hosts; do
			for adr in `separate_list $addr`; do
			    addnatrule $chain -s ${host#*:} -d $adr -j RETURN
			done
		    done
		done

		for adr in $excludedests; do
		    addnatrule $chain -d $adr -j RETURN
		done

		for adr in `separate_list $addr`; do
		    if [ -n "$loglevel" ]; then
			log_rule $loglevel $chain $logtarget -t nat -d `fix_bang $adr`
		    fi

		    addnatrule $chain -d $adr -j $target1
		done
	    else
		for adr in `separate_list $addr`; do
		    if [ -n "$loglevel" ]; then
			ensurenatchain $chain
			log_rule $loglevel $chain $logtarget -t nat \
			    `fix_bang $proto $cli $sports -d $adr $multiport $dports`
		    fi
		    
		    addnatrule $chain $proto $cli $sports \
			-d $adr $multiport $dports -j $target1
		done
	    fi
	fi
    fi

    # Replace destination port by the new destination port

    if [ -n "$servport" ]; then
	if [ -z "$multiport" ]; then
	    dports="--dport ${servport#*:}"
	else
	    dports="--dports ${servport#*:}"
	fi
    fi

    # Handle SNAT

    if [ -n "$snat" ]; then
	if [ -n "$cli" ]; then
	    [ $command = check ] || addnatrule `snat_chain $dest` $proto $cli $multiport \
		$sports -d $serv $dports -j SNAT --to-source $snat
	else
	    for source_host in $source_hosts; do
		[ "x${source_host#*:}" = "x0.0.0.0/0" ] && \
		    error_message "Warning: SNAT will occur on all connections to this server and port - rule \"$rule\""

		[ $command = check ] || addnatrule `snat_chain $dest` \
		    -s ${source_host#*:} $proto $sports $multiport \
		    -d $serv $dports -j SNAT --to-source $snat
	    done
	fi
    fi
}

#
# Add one Filter Rule -- Helper function for the rules file processor
#
# The caller has established the following variables:
#    check       = current command. If 'check', we're executing a 'check'
#                  which only goes through the motions.
#    client	 = SOURCE IP or MAC
#    server	 = DESTINATION IP or interface
#    protocol	 = Protocol
#    address	 = Original Destination Address
#    port	 = Destination Port
#    cport	 = Source Port
#    multioption = String to invoke multiport match if appropriate
#    servport	 = Port the server listens on
#    chain	 = The canonical chain for this rule
#
add_a_rule()
{
    local natrule=

    # Set source variables. The 'cli' variable will hold the client match predicate(s).

    cli=

    case "$client" in
    -)
	;;
    *:*)
	cli="-i ${client%:*} -s ${client#*:}"
	;;
    *.*.*)
	cli="-s $client"
	;;
    ~*)
	cli=`mac_match $client`
	;;
    *)
	[ -n "$client" ] && cli="-i $client"
	;;
    esac

    # Set destination variables - 'serv' and 'dest_interface' hold the server match predicate(s).

    dest_interface=
    serv=

    case "$server" in
    -)
	;;
    *.*.*)
	serv=$server
	;;
    ~*)
	fatal_error "Rule \"$rule\" - Destination may not be specified by MAC Address"
	;;
    *)
	[ -n "$server" ] && dest_interface="-o $server"
	;;
    esac

    # Setup protocol and port variables

    sports=
    dports=
    state="-m state --state NEW"
    proto=$protocol
    addr=$address
    servport=$serverport
    multiport=

    [ x$port  = x- ] && port=
    [ x$cport = x- ] && cport=

    case $proto in
	tcp|udp|TCP|UDP|6|17)
	    if [ -n "$port" ]; then
		dports="--dport"
		if [ -n "$multioption" -a "$port" != "${port%,*}" ]; then
		    multiport="$multioption"
		    dports="--dports"
		fi
		dports="$dports $port"
	    fi

	    if [ -n "$cport" ]; then
		sports="--sport"
		if [ -n "$multioption" -a "$cport" != "${cport%,*}" ]; then
		    multiport="$multioption"
		    sports="--sports"
		fi
		sports="$sports $cport"
	    fi
		;;
	icmp|ICMP|1)
	    [ -n "$port" ]  && 	dports="--icmp-type $port"
	    state=
	    ;;
	all|ALL)
	    [ -n "$port" ] && \
		fatal_error "Port number not allowed with \"all\"; rule: \"$rule\""
	    proto=
	    ;;
	*)
	    state=
	    [ -n "$port" ] && \
		fatal_error "Port number not allowed with protocol \"$proto\"; rule: \"$rule\""
	    ;;
    esac

    proto="${proto:+-p $proto}"

    # Some misc. setup

    case "$logtarget" in
	REJECT)
	    target=reject
	    [ -n "$servport" ] && \
		fatal_error "Server port may not be specified in a REJECT rule;"\
                            "rule: \"$rule\""
	    ;;
	REDIRECT)
	    [ -n "$serv" ] && startup_error "REDIRECT rules cannot"\
		" specify a server IP; rule: \"$rule\""
	    servport=${servport:=$port}
	    natrule=Yes
	    ;;
	DNAT)
	    [ -n "$serv" ] || fatal_error "DNAT rules require a" \
		" server address; rule: \"$rule\""
	    natrule=Yes
	    ;;
	LOG)
	    [ -z "$loglevel" ] && fatal_error "LOG requires log level"
	    ;;
    esac

    # Complain if the rule is really a policy

    if [ -z "$proto" -a -z "$cli" -a -z "$serv" -a -z "$servport" -a "$logtarget" != LOG ]; then
	error_message "Warning -- Rule \"$rule\" is a POLICY"
	error_message "	       -- and should be moved to the policy file"
    fi

    if [ -n "${serv}${servport}" ]; then
	if [ $command != check ]; then

	    # A specific server or server port given

	    if [ -n "$natrule" ]; then
		add_nat_rule
	    elif [ -n "$addr" -a "$addr" != "$serv" ] || [ -n "$servport" -a "$servport" != "$port" ]; then
		fatal_error "Only DNAT and REDIRECT rules may specify destination mapping; rule \"$rule\""
	    fi

	    if [ -z "$dnat_only" -a $chain != ${FW}2${FW} ]; then
		if [ -n "$serv" ]; then
		    for serv1 in `separate_list $serv`; do
			for srv in `ip_range $serv1`; do
			    if [ -n "$addr" -a -n "$CONNTRACK_MATCH" ]; then
				for adr in $addr; do
				    if [ -n "$loglevel" -a -z "$natrule" ]; then
					log_rule $loglevel $chain $logtarget -m conntrack --ctorigdst $adr \
					    `fix_bang $proto $sports $multiport $state $cli -d $srv $dports`
				    fi

				    run_iptables2 -A $chain $proto $multiport $state $cli $sports \
					-d $srv $dports -m conntrack --ctorigdst $adr -j $target
				done
			    else
				if [ -n "$loglevel" -a -z "$natrule" ]; then
				    log_rule $loglevel $chain $logtarget \
					`fix_bang $proto $sports $multiport $state $cli -d $srv $dports`
				fi

				run_iptables2 -A $chain $proto $multiport $state $cli $sports \
				    -d $srv $dports -j $target
			    fi
			done
		    done
		else
		    if [ -n "$loglevel" -a -z "$natrule" ]; then
		        log_rule $loglevel $chain $logtarget \
			    `fix_bang $proto $sports $multiport $state $cli $dports`
		    fi

		    run_iptables2 -A $chain $proto $multiport $state $cli $sports \
			$dports -j $target
		fi
	    fi
	fi
    else

	# Destination is a simple zone

	[ -n "$addr" ] && fatal_error \
	    "An ORIGINAL DESTINATION ($addr) is only allowed in" \
	    " a DNAT or REDIRECT: \"$rule\""

	if [ $command != check ]; then
	    if [ -n "$loglevel" ]; then
		log_rule $loglevel $chain $logtarget \
		    `fix_bang $proto $multiport $dest_interface $state $cli $sports $dports`
	    fi

	    if [ $logtarget != LOG ]; then
		run_iptables2 -A $chain $proto $multiport $dest_interface $state \
		    $cli $sports $dports -j $target
	    fi
	fi
    fi
}

#
# Process a record from the rules file for the 'start', 'restart' or 'check' commands
#
process_rule() # $1 = target
               # $2 = clients
               # $3 = servers
               # $4 = protocol
               # $5 = ports
               # $6 = cports
               # $7 = address
{
    local target="$1"
    local clients="$2"
    local servers="$3"
    local protocol="$4"
    local ports="$5"
    local cports="$6"
    local address="$7"
    local rule="`echo $target $clients $servers $protocol $ports $cports $address`"

    # Function Body -- isolate log level

    if [ "$target" = "${target%:*}" ]; then
	loglevel=
    else
	loglevel="${target#*:}"
	target="${target%:*}"
	expandv loglevel
    fi

    logtarget="$target"
    dnat_only=

    # Tranform the rule:
    #
    # - set 'target' to the filter table target.
    # - make $FW the destination for REDIRECT
    # - remove '-' suffix from logtargets while setting 'dnat_only'
    # - clear 'address' if it has been set to '-'

    [ "x$address" = "x-" ] && address=

    case $target in
    DNAT)
	target=ACCEPT
	address=${address:=detect}
	;;
    DNAT-)
	target=ACCEPT
	address=${address:=detect}
	dnat_only=Yes
	logtarget=DNAT
	;;
    REDIRECT)
	target=ACCEPT
	address=${address:=all}
	if [ "x-" = "x$servers" ]; then
	    servers=$FW
	else
	    servers="$FW::$servers"
	fi
	;;
    REDIRECT-)
	target=ACCEPT
	logtarget=REDIRECT
	dnat_only=Yes
	address=${address:=all}
	if [ "x-" = "x$servers" ]; then
	    servers=$FW
	else
	    servers="$FW::$servers"
	fi
	;;
    esac

    # Parse and validate source

    if [ "$clients" = "${clients%:*}" ]; then
	clientzone="$clients"
	clients=
    else
	clientzone="${clients%%:*}"
	clients="${clients#*:}"
	[ -z "$clientzone" -o -z "$clients" ] && \
	   fatal_error "Empty source zone or qualifier: rule \"$rule\""
    fi

    if [ "$clientzone" = "${clientzone%\!*}" ]; then
	excludezones=
    else
	excludezones="${clientzone#*\!}"
	clientzone="${clientzone%\!*}"

	[ "$logtarget" = DNAT ] || [ "$logtarget" = REDIRECT ] ||\
	    fatal_error "Exclude list only allowed with DNAT or REDIRECT"
    fi

    validate_zone $clientzone || fatal_error "Undefined Client Zone in rule \"$rule\""

    # Parse and validate destination

    source=$clientzone

    [ $source = $FW ] && source_hosts= || eval source_hosts=\"\$${source}_hosts\"

    if [ "$servers" = "${servers%:*}" ] ; then
	serverzone="$servers"
	servers=
	serverport=
    else
	serverzone="${servers%%:*}"
	servers="${servers#*:}"
	if [ "$servers" != "${servers%:*}" ] ; then
	    serverport="${servers#*:}"
	    servers="${servers%:*}"
	    [ -z "$serverzone" -o -z "$serverport" ] && \
	        fatal_error "Empty destination zone or server port: rule \"$rule\""
	else
	    serverport=
	    [ -z "$serverzone" -o -z "$servers" ] && \
	        fatal_error "Empty destination zone or qualifier: rule \"$rule\""
	fi
    fi

    if ! validate_zone $serverzone; then
	fatal_error "Undefined Server Zone in rule \"$rule\""
    fi

    dest=$serverzone

    # Ensure that this rule doesn't apply to a NONE policy pair of zones

    chain=${source}2${dest}

    eval policy=\$${chain}_policy

    [ $policy = NONE ] && \
	fatal_error "Rules may not override a NONE policy: rule \"$rule\""

    # Be sure that this isn't a fw->fw rule.

    if [ "x$chain" = x${FW}2${FW} ]; then
	case $logtarget in
	    REDIRECT|DNAT)
		#
		# Redirect rules that have the firewall as the source are fw->fw rules
		#
		;;
	    *)
		error_message "WARNING: fw -> fw rules are not supported; rule \"$rule\" ignored"
		return
		;;
	esac
    else

	# Create the canonical chain if it doesn't already exist
    
	[ $command = check ] || ensurechain $chain
    fi

    # Generate Netfilter rule(s)

    case $logtarget in
	DNAT*)
	    if [ -n "$MULTIPORT" -a \
		"$ports" = "${ports%:*}" -a \
		"$cports" = "${cports%:*}" -a \
		`list_count $ports` -le 15 -a \
		`list_count $cports` -le 15 ]
		then
	        #
	        # MULTIPORT is enabled, there are no port ranges in the rule and less than
	        # 16 ports are listed - use multiport match.
	        #
		multioption="-m multiport"
		for client in `separate_list ${clients:=-}`; do
		    #
		    # add_a_rule() modifies these so we must set their values each time
		    #
		    server=${servers:=-}
		    port=${ports:=-}
		    cport=${cports:=-}
		    add_a_rule
		done
	    else
	        #
	        # MULTIPORT is disabled or the rule isn't compatible with multiport match
	        #
		multioption=
		for client in `separate_list ${clients:=-}`; do
		    for port in `separate_list ${ports:=-}`; do
			for cport in `separate_list ${cports:=-}`; do
			    server=${servers:=-}
			    add_a_rule
			done
		    done
		done
	    fi
	    ;;
	*)

	    if [ -n "$MULTIPORT" -a \
		"$ports" = "${ports%:*}" -a \
		"$cports" = "${cports%:*}" -a \
		`list_count $ports` -le 15 -a \
		`list_count $cports` -le 15 ]
		then
	        #
	        # MULTIPORT is enabled, there are no port ranges in the rule and less than
	        # 16 ports are listed - use multiport match.
	        #
		multioption="-m multiport"
		for client in `separate_list ${clients:=-}`; do
		    for server in `separate_list ${servers:=-}`; do
		        #
		        # add_a_rule() modifies these so we must set their values each time
		        #
			port=${ports:=-}
			cport=${cports:=-}
			add_a_rule
		    done
		done
	    else
	        #
	        # MULTIPORT is disabled or the rule isn't compatible with multiport match
	        #
		multioption=
		for client in `separate_list ${clients:=-}`; do
		    for server in `separate_list ${servers:=-}`; do
			for port in `separate_list ${ports:=-}`; do
			    for cport in `separate_list ${cports:=-}`; do
				add_a_rule
			    done
			done
		    done
		done
	    fi
	    ;;
    esac
    #
    # Report Result
    #
    if [ $command = check ]; then
	echo "   Rule \"$rule\" checked."
    else
	echo "   Rule \"$rule\" added."
    fi
}

#
# Process the rules file for the 'start', 'restart' or 'check' command.
#
process_rules() # $1 = name of rules file
{
    #
    # Process a rule where the source or destination is "all"
    #
    process_wildcard_rule() {
	for yclients in $xclients; do
	    for yservers in $xservers; do
		if [ "${yclients}" != "${yservers}" ] ; then
		    process_rule $xtarget $yclients $yservers $xprotocol $xports $xcports $xaddress
		fi
	    done
	done
    }

    while read xtarget xclients xservers xprotocol xports xcports xaddress; do
	case "${xtarget%:*}" in

	    ACCEPT|DROP|REJECT|DNAT|DNAT|DNAT-|REDIRECT|REDIRECT-|LOG|CONTINUE)
		expandv xclients xservers xprotocol xports xcports xaddress

		if [ "x$xclients" = xall ]; then
		    xclients="$zones $FW"
		    if [ "x$xservers" = xall ]; then
			xservers="$zones $FW"
		    fi
		    process_wildcard_rule
		    continue
		fi

		if [ "x$xservers" = xall ]; then
		    xservers="$zones $FW"
		    process_wildcard_rule
		    continue
		fi

		process_rule $xtarget $xclients $xservers $xprotocol $xports $xcports $xaddress
		;;
	    *)
		rule="`echo $xtarget $xclients $xservers $xprotocol $xports $xcports $xaddress`"
		fatal_error "Invalid Target in rule \"$rule\""
		;;

	esac
    done < $TMP_DIR/rules
}

#
# Process a record from the tos file
#
# The caller has loaded the column contents from the record into the following
# variables:
#
#    src dst protocol sport dport tos
#
# and has loaded a space-separated list of their values in "rule".
#
process_tos_rule() {
    #
    # Parse the contents of the 'src' variable
    #
    if [ "$src" = "${src%:*}" ]; then
	srczone="$src"
	src=
    else
	srczone="${src%:*}"
	src="${src#*:}"
    fi

    source=
    #
    # Validate the source zone
    #
    if validate_zone $srczone; then
	source=$srczone
    elif [ "$srczone" = "all" ]; then
	source="all"
    else
	error_message "Warning: Undefined Source Zone - rule \"$rule\" ignored"
	return
    fi

    [ -n "$src" ] && case "$src" in
	*.*.*)
	    #
	    # IP Address or subnet
	    #
	    src="-s $src"
	    ;;
	~*)
	    src=`mac_match $src`
	    ;;
	*)
	    #
	    # Assume that this is a device name
	    #
	    src="-i $src"
	    ;;
    esac

    #
    # Parse the contents of the 'dst' variable
    #
    if [ "$dst" = "${dst%:*}" ]; then
	dstzone="$dst"
	dst=
    else
	dstzone="${dst%:*}"
	dst="${dst#*:}"
    fi

    dest=
    #
    # Validate the destination zone
    #
    if validate_zone $dstzone; then
	dest=$dstzone
    elif [ "$dstzone" = "all" ]; then
	dest="all"
    else
	error_message \
	 "Warning: Undefined Destination Zone - rule \"$rule\" ignored"
	return
    fi

    [ -n "$dst" ] && case "$dst" in
	*.*.*)
	    #
	    # IP Address or subnet
	    #
	    ;;
	*)
	    #
	    # Assume that this is a device name
	    #
	    error_message \
	     "Warning: Invalid Destination - rule \"$rule\" ignored"
	    return
	    ;;
    esac

    #
    # Setup PROTOCOL and PORT variables
    #
    sports=""
    dports=""

    case $protocol in
	tcp|udp|TCP|UDP|6|17)
	    [ -n "$sport" ] && [ "x${sport}" != "x-" ] && \
		sports="--sport $sport"
	    [ -n "$dport" ] && [ "x${dport}" != "x-" ] && \
		dports="--dport $dport"
	    ;;
	icmp|ICMP|0)
	    [ -n "$dport" ] && [ "x${dport}" != "x-" ] && \
		dports="--icmp-type $dport"
	    ;;
	all|ALL)
	    protocol=
	    ;;
	*)
	    ;;
    esac

    protocol="${protocol:+-p $protocol}"

    tos="-j TOS --set-tos $tos"

    case "$dstzone" in
    all|ALL)
	dst=0.0.0.0/0
	;;
    *)
	[ -z "$dst" ] && eval dst=\$${dstzone}_hosts
	;;
    esac

    for dest in $dst; do
	dest="-d $dest"

	case $srczone in
	$FW)
	    run_iptables2 -t mangle -A outtos \
		$protocol $dest $dports $sports $tos
	    ;;
	all|ALL)
	    run_iptables2 -t mangle -A outtos \
		$protocol $dest $dports $sports $tos
	    run_iptables2 -t mangle -A pretos \
		$protocol $dest $dports $sports $tos
	    ;;
	*)
	    if [ -n "$src" ]; then
		run_iptables2 -t mangle -A pretos $src \
		    $protocol $dest $dports $sports $tos
	    else
		eval interfaces=\$${srczone}_interfaces

		for interface in $interfaces; do
		    run_iptables2 -t mangle -A pretos -i $interface \
			$protocol $dest $dports $sports $tos
		done
	    fi
	    ;;
	esac
    done

    echo "   Rule \"$rule\" added."
}

#
# Process the tos file
#
process_tos() # $1 = name of tos file
{
    echo "Processing $1..."

    run_iptables -t mangle -N pretos
    run_iptables -t mangle -N outtos

    strip_file tos $1

    while read src dst protocol sport dport tos; do
	expandv src dst protocol sport dport tos
	rule="`echo $src $dst $protocol $sport $dport $tos`"
	process_tos_rule
    done < $TMP_DIR/tos

    run_iptables -t mangle -A PREROUTING -j pretos
    run_iptables -t mangle -A OUTPUT	 -j outtos
}

#
# Load a Kernel Module
#
loadmodule() # $1 = module name, $2 - * arguments
{
    local modulename=$1
    local modulefile

    if [ -z "`lsmod | grep $modulename`" ]; then
	shift
	modulefile=$MODULESDIR/${modulename}.o

	if [ -f $modulefile ]; then
	    insmod $modulefile $*
	    return
	fi
	#
	# If the modules directory contains compressed modules then we'll
	# assume that insmod can load them
	#
	modulefile=${modulefile}.gz

	if [ -f $modulefile ]; then
	    insmod $modulefile $*
	fi
    fi
}

#
# Display elements of a list with leading white space
#
display_list() # $1 = List Title, rest of $* = list to display
{
    [ $# -gt 1 ] && echo "   $*"
}

#
# Add rules to the "common" chain to silently drop packets addressed to any of
# the passed addresses
#
drop_broadcasts() # $* = broadcast addresses
{
    while [ $# -gt 0 ]; do
	run_iptables -A common -d $1 -j DROP
	shift
    done
}

#
# Add policy rule ( and possibly logging rule) to the passed chain
#
policy_rules() # $1 = chain to add rules to
	       # $2 = policy
	       # $3 = loglevel
{
    local target="$2"

    case "$target" in
    ACCEPT)
	;;

    DROP)
	run_iptables -A $1 -j common
	;;
    REJECT)
	run_iptables -A $1 -j common
	target=reject
	;;
    CONTINUE)
	target=
	;;
    *)
	fatal_error "Invalid policy ($policy) for $1"
	;;

    esac

    if [ $# -eq 3 -a "x${3}" != "x-" ]; then
	log_rule $3 $1 $2 
    fi

    [ -n "$target" ] && run_iptables -A $1 -j $target
}

#
# Generate default policy & log level rules for the passed client & server
# zones
#
# This function is only called when the canonical chain for this client/server
# pair is known to exist. If the default policy for this pair specifies the
# same chain then we add the policy (and logging) rule to the canonical chain;
# otherwise add a rule to the canonical chain to jump to the appropriate
# policy chain.
#
default_policy() # $1 = client $2 = server
{
    local chain="${1}2${2}"
    local policy=
    local loglevel=
    local chain1

    jump_to_policy_chain() {
	#
	# Add a jump to from the canonical chain to the policy chain. On return,
	# $chain is set to the name of the policy chain
	#
	run_iptables -A $chain -j $chain1
	chain=$chain1
    }

    apply_default()
    {
	#
	# Generate policy file column values from the policy chain
	#
	eval policy=\$${chain1}_policy
	eval loglevel=\$${chain1}_loglevel
	eval synparams=\$${chain1}_synparams
	#
	# Add the appropriate rules to the canonical chain ($chain) to enforce
	# the specified policy

	if [ "$chain" = "$chain1" ]; then
	    #
	    # The policy chain is the canonical chain; add policy rule to it
	    # The syn flood jump has already been added if required.
	    #
	    policy_rules $chain $policy $loglevel
	else
	    #
	    # The policy chain is different from the canonical chain -- approach
	    # depends on the policy
	    #
	    case $policy in
	    ACCEPT)
		if [ -n "$synparams" ]; then
		    #
		    # To avoid double-counting SYN packets, enforce the policy
		    # in this chain.
		    #
		    enable_syn_flood_protection $chain $chain1
		    policy_rules $chain $policy $loglevel
		else
		    #
		    # No problem with double-counting so just jump to the
		    # policy chain.
		    #
		    jump_to_policy_chain
		fi
		;;
	    CONTINUE)
		#
		# Silly to jump to the policy chain -- add any logging
		# rules and enable SYN flood protection if requested
		#
		[ -n "$synparams" ] && \
		    enable_syn_flood_protection $chain $chain1
		policy_rules $chain $policy $loglevel
		;;
	    *)
		#
		# DROP or REJECT policy -- enforce in the policy chain and
		# enable SYN flood protection if requested.
		#
		[ -n "$synparams" ] && \
		    enable_syn_flood_protection $chain $chain1
		jump_to_policy_chain
		;;
	    esac
	fi

	echo "   Policy $policy for $1 to $2 using chain $chain"
    }

    eval chain1=\$${1}2${2}_policychain

    if [ -n "$chain1" ]; then
	apply_default $1 $2
    else
	fatal_error "No default policy for zone $1 to zone $2"
    fi
}

#
# Complete a standard chain
#
#	- run any supplied user exit
#	- search the policy file for an applicable policy and add rules as
#	  appropriate
#	- If no applicable policy is found, add rules for an assummed
#	  policy of DROP INFO
#
complete_standard_chain() # $1 = chain, $2 = source zone, $3 = destination zone
{
    local policy=
    local loglevel=
    local policychain=

    run_user_exit $1

    eval policychain=\$${2}2${3}_policychain

    if [ -n "$policychain" ]; then
	eval policy=\$${policychain}_policy
	eval loglevel=\$${policychain}_loglevel

	policy_rules $1 $policy $loglevel
    else
	policy_rules $1 DROP INFO
    fi
}

#
# Find the appropriate chain to pass packets from a source zone to a
# destination zone
#
# If the canonical chain for this zone pair exists, echo it's name; otherwise
# locate and echo the name of the appropriate policy chain
#
rules_chain() # $1 = source zone, $2 = destination zone
{
    local chain=${1}2${2}

    havechain $chain && { echo $chain; return; }

    [ "$1" = "$2" ] && { echo ACCEPT; return; }

    eval chain=\$${chain}_policychain

    [ -n "$chain" ] && { echo $chain; return; }

    fatal_error "No appropriate chain for zone $1 to zone $2"
}

#
#  echo the list of subnets routed out of a given interface
#
get_routed_subnets() # $1 = interface name
{
    local address
    local rest

    ip route show dev $1 2> /dev/null |
	while read address rest; do
	    if [ "x$address" = xdefault ]; then
		error_message "Warning: default route ignored on interface $1"
	    else
		[ "$address" = "${address%/*}" ] && address="${address}/32"
		echo $address
	    fi
        done
}
#
# Convert an IP address in quad format to an integer
#
decodeaddr() {
    local x
    local temp=0
    local ifs=$IFS
    
    IFS=.

    for x in $1; do
	temp=$(( $(( $temp $LEFTSHIFT 8 )) | $x ))
    done

    echo $temp

    IFS=$ifs
}
#
# convert an integer to quad format
#
encodeaddr() {
    addr=$1
    local x
    local y=$(($addr & 255))

    for x in 1 2 3 ; do
	addr=$(($addr >> 8))
	y=$(($addr & 255)).$y
    done

    echo $y
}
#
# Enumerate the members of an IP range
#
ip_range() {
    local first
    local last

    case $1 in 
    *-*)
	;;
    *)
	echo $1
	return
	;;
    esac

    first=`decodeaddr ${1%-*}`
    last=`decodeaddr ${1#*-}`

    if [ $first -gt $last -o $(($last - $first)) -gt 256 ]; then
	fatal_error "Invalid IP address range: $1"
    fi

    while [ $first -le $last ]; do
	echo `encodeaddr $first`
	first=$(($first + 1))
    done
}
#
# Netmask from CIDR
#
ip_netmask() {
    echo $(( -1 $LEFTSHIFT $((32 - ${1#*/})) ))
}

#
# Network address from CIDR
#
networkaddress() {
    local decodedaddr=`decodeaddr ${1%/*}`
    local netmask=`ip_netmask $1`

    echo `encodeaddr $(($decodedaddr & $netmask))`
}
#
# Calculate broadcast address from CIDR
#
broadcastaddress() {
    local decodedaddr=`decodeaddr ${1%/*}`
    local netmask=`ip_netmask $1`

    echo `encodeaddr $(( $(($decodedaddr & $netmask)) | $(( $netmask ^ -1 )) ))`
}
#
# Test for subnet membership
#
in_subnet() # $1 = IP address, $2 = CIDR network
{
    local netmask=`ip_netmask $2`

    test $(( `decodeaddr $1` & $netmask)) -eq $(( `decodeaddr ${2%/*}` & $netmask ))
}
#
# Set up Source NAT (including masquerading)
#
setup_masq()
{
    setup_one() {
	local using

	case $fullinterface in
	    *:*:*)
		# Both alias name and subnet
		destnet="${fullinterface##*:}"
		fullinterface="${fullinterface%:*}"
		;;
	    *:*)
		# Alias name OR subnet
		case ${fullinterface#*:} in
		    *.*)
			# It's a subnet
			destnet="${fullinterface#*:}"
			fullinterface="${fullinterface%:*}"
			;;
		    *)
			#it's an alias name
			destnet="0.0.0.0/0"
			;;
		esac
		;;
	    *)
		destnet="0.0.0.0/0"
		;;
	esac

	interface=${fullinterface%:*}

	if ! list_search $interface $all_interfaces; then
	    fatal_error "Unknown interface $interface"
	fi

	if [ "$subnet" = "${subnet%!*}" ]; then
	    nomasq=
	else
	    nomasq="${subnet#*!}"
	    subnet="${subnet%!*}"
	fi

	chain=`masq_chain $interface`
	iface=

	source="$subnet"

	case $subnet in
	*.*.*)
	    ;;
	-)
	    #
	    # Note: This only works if you have the LOCAL NAT patches in the
	    #	    kernel and in the iptables utility
	    #
	    chain=OUTPUT
	    subnet=
	    source=$FW
	    iface="-o $interface"
	    ;;
	*)
	    subnets=`get_routed_subnets $subnet`
	    [ -z "$subnets" ] && startup_error "Unable to determine the routes through interface $subnet"
	    subnet="$subnets"
	    ;;
	esac

	if [ -n "$address" -a -n "$ADD_SNAT_ALIASES" ]; then
	    for addr in `ip_range $address` ; do
		if ! list_search $addr $aliases_to_add; then
		    aliases_to_add="$aliases_to_add $addr $fullinterface"
		    case $fullinterface in
		    *:*)
			fullinterface=${fullinterface%:*}:$((${fullinterface#*:} + 1 ))
			;;
		    esac
		fi
	    done
	fi

	destination=$destnet

	if [ -n "$nomasq" ]; then
	    newchain=masq${masq_seq}
	    createnatchain $newchain

	    if [ -n "$subnet" ]; then
		for s in $subnet; do
		    addnatrule $chain -d $destnet $iface -s $s -j $newchain
		done
	    else
		addnatrule $chain -d $destnet $iface -j $newchain
	    fi

	    masq_seq=$(($masq_seq + 1))
	    chain=$newchain
	    subnet=
	    iface=
	    destnet=

	    for addr in `separate_list $nomasq`; do
		addnatrule $chain -s $addr -j RETURN
	    done

	    source="$source except $nomasq"
	else
	    destnet="-d $destnet"
	fi

	if [ -n "$subnet" ]; then
	    for s in $subnet; do
		if [ -n "$address" ]; then
		    addnatrule $chain -s $s $destnet $iface -j SNAT --to-source $address
		    echo "   To $destination from $s through ${interface} using $address"
		else
		    addnatrule $chain -s $s $destnet $iface -j MASQUERADE
		    echo "   To $destination from $s through ${interface}"
		fi
	    done
	elif [ -n "$address" ]; then
	    addnatrule $chain $destnet $iface -j SNAT --to-source $address
	    echo "   To $destination from $source through ${interface} using $address"
	else
	    addnatrule $chain $destnet $iface -j MASQUERADE
	    echo "   To $destination from $source through ${interface}"
	fi

    }

    strip_file masq $1

    [ -n "$NAT_ENABLED" ] && echo "Masqueraded Subnets and Hosts:"

    while read fullinterface subnet address; do
	expandv fullinterface subnet address
	[ -n "$NAT_ENABLED" ] && setup_one || \
	    error_message "Warning: NAT disabled; masq rule ignored"
    done < $TMP_DIR/masq
}

#
# Add a record to the blacklst chain
#
#   $source      = address match
#   $proto       = protocol selector
#   $dport       = destination port selector
#
add_blacklist_rule() {
    if [ -n "$BLACKLIST_LOGLEVEL" ]; then
	log_rule $BLACKLIST_LOGLEVEL blacklst $BLACKLIST_DISPOSITION  `fix_bang $source $proto $dport`
    fi

    run_iptables2 -A blacklst $source $proto $dport -j $disposition
}

#
# Process a record from the blacklist file
#
#   $subnet	 = address/subnet
#   $protocol    = Protocol Number/Name
#   $port        = Port Number/Name
#
process_blacklist_rec() {
    local source
    local addr
    local proto
    local dport

    for addr in `separate_list $subnet`; do
	case $addr in
	~*)
	    addr=`echo $addr | sed 's/~//;s/-/:/g'`
	    source="--match mac --mac-source $addr"
	    ;;
	*)
	    source="-s $addr"
	    ;;
	esac

	if [ -n "$protocol" ]; then
	    proto=" -p $protocol "

	    case $protocol in
		tcp|TCP|6|udp|UDP|17)
		    if [ -n "$ports" ]; then
			if [ -n "$MULTIPORT" -a \
			    "$ports" != "${ports%,*}" -a \
			    "$ports" = "${ports%:*}" -a \
			    `list_count $ports` -le 15 ]
			then
			    dport="-m multiport --dports $ports"
			    add_blacklist_rule
			else
			    for dport in `separate_list $ports`; do
				dport="--dport $dport"
				add_blacklist_rule
			    done
			fi
		    else
			add_blacklist_rule
		    fi
		    ;;
		icmp|ICMP|0)
		    if [ -n "$ports" ]; then
			for dport in `separate_list $ports`; do
			    dport="--icmp-type $dport"
			    add_blacklist_rule
			done
		    else
			add_blacklist_rule
		    fi
		    ;;
		*)
		    add_blacklist_rule
		    ;;
	    esac
	else
	    add_blacklist_rule
	fi

	if [ -n "$ports" ]; then
	    addr="$addr $protocol $ports"
	elif [ -n "$protocol" ]; then
	    addr="$addr $protocol"
	fi

	echo "   $addr added to Black List"
    done
}

#
# Setup the Black List
#
setup_blacklist() {
    local interfaces=`find_interfaces_by_option blacklist`
    local f=`find_file blacklist`
    local disposition=$BLACKLIST_DISPOSITION

    if [ -n "$interfaces" -a -f $f ]; then
	echo "Setting up Blacklisting..."

	strip_file blacklist $f

	createchain blacklst no

	for interface in $interfaces; do
	    for chain in `first_chains $interface`; do
		run_iptables -A $chain -j blacklst
	    done

	    echo "   Blacklisting enabled on $interface"
	done

	[ "$disposition" = REJECT ] && disposition=reject

	while read subnet protocol ports; do
	    expandv subnet protocol ports
	    process_blacklist_rec
	done < $TMP_DIR/blacklist

    fi
}

#
# Refresh the Black List
#
refresh_blacklist() {
    local f=`find_file blacklist`
    local disposition=$BLACKLIST_DISPOSITION

    if qt iptables -L blacklst -n ; then
	echo "Refreshing Black List..."

	strip_file blacklist $f

	[ "$disposition" = REJECT ] && disposition=reject

	run_iptables -F blacklst

	while read subnet protocol ports; do
	    expandv subnet protocol ports
	    process_blacklist_rec
	done < $TMP_DIR/blacklist
    fi
}

#
# Verify that kernel has netfilter support
#
verify_os_version() {

    osversion=`uname -r`

    case $osversion in
    2.4.*|2.5.*)
	;;
    *)
	startup_error "Shorewall version $version does not work with kernel version $osversion"
	;;
    esac

    [ $command = start -a -n "`lsmod 2> /dev/null | grep '^ipchains'`" ] && \
	startup_error "Shorewall can't start with the ipchains kernel module loaded - see FAQ #8"
}

#
# Add IP Aliases
#
add_ip_aliases()
{
    local addresses external interface inet cidr rest val

    address_details() 
    {
	#
	# Folks feel uneasy if they don't see all of the same
	# decoration on these IP addresses that they see when their
	# distro's net config tool adds them. In an attempt to reduce
	# the anxiety level, we have the following code which sets
	# the VLSM and BRD from an existing address in the same subnet
	#
	# Get all of the lines that contain inet addresses 
	#
	ip addr show $interface 2> /dev/null | grep 'inet' | while read inet cidr rest ; do
	    if in_subnet $external $cidr; then
		echo "/${cidr#*/} brd `broadcastaddress $cidr`"
		break
	    fi
	done
    }

    do_one()
    {
	val=`address_details`
	run_ip addr add ${external}${val} dev $interface $label
	echo "$external $interface" >> ${STATEDIR}/nat
	[ -n "$label" ] && label="with $label"
	echo "   IP Address $external added to interface $interface $label"
    }

    set -- $aliases_to_add

    while [ $# -gt 0 ]; do
	external=$1
	interface=$2
	label=

	if [ "$interface" != "${interface%:*}" ]; then
	    label="${interface#*:}"
	    interface="${interface%:*}"
	    label="label $interface:$label"
	fi

	shift;shift

	list_search $external `find_interface_addresses $interface` || do_one
    done
}

#
# Load kernel modules required for Shorewall
#
load_kernel_modules() {

    [ -z "$MODULESDIR" ] && \
	MODULESDIR=/lib/modules/$osversion/kernel/net/ipv4/netfilter

    modules=`find_file modules`

    if [ -f $modules -a -d $MODULESDIR ]; then
	echo "Loading Modules..."
	. $modules
    fi
}

# Verify that the 'ip' program is installed

verify_ip() {
    qt ip link ls ||\
	startup_error "Shorewall $version requires the iproute package ('ip' utility)"
}

# 
# Determine which optional facilities are supported by iptables/netfilter
#
determine_capabilities() {
    qt iptables -t nat    -L -n && NAT_ENABLED=Yes    || NAT_ENABLED=
    qt iptables -t mangle -L -n && MANGLE_ENABLED=Yes || MANGLE_ENABLED=

    CONNTRACK_MATCH=
    MULTIPORT=

    if qt iptables -N fooX1234 ; then
	qt iptables -A fooX1234 -m conntrack --ctorigdst 192.168.1.1 -j ACCEPT && CONNTRACK_MATCH=Yes
        qt iptables -A fooX1234 -p tcp -m multiport --dports 21,22 -j ACCEPT   && MULTIPORT=Yes

	qt iptables -F fooX1234
	qt iptables -X fooX1234
    fi
}

report_capability() # $1 = Capability Name, $2 Capability Setting (if any)
{
    local setting=

    [ x$1 = "xYes" ] && { setting="Available"; shift; } || setting="Not available"

    echo "  " $@: $setting
}

report_capabilities() {
    echo "Shorewall has detected the following iptables/netfilter capabilities:"
    report_capability $NAT_ENABLED "NAT"
    report_capability $MANGLE_ENABLED "Packet Mangling"
    report_capability $MULTIPORT "Multi-port Match"
    report_capability $CONNTRACK_MATCH "Connection Tracking Match"
}

#
# Perform Initialization
#	- Delete all old rules
#	- Delete all user chains
#	- Set the POLICY on all standard chains and add a rule to allow packets
#	  that are part of established connections
#	- Determine the zones
#
initialize_netfilter () {

    report_capabilities

    echo "Determining Zones..."

    determine_zones

    [ -z "$zones" ] && startup_error "No Zones Defined"

    display_list "Zones:" $zones

    echo "Validating interfaces file..."

    validate_interfaces_file

    echo "Validating hosts file..."

    validate_hosts_file

    echo "Validating Policy file..."

    validate_policy

    echo "Determining Hosts in Zones..."

    determine_interfaces
    determine_hosts

    run_user_exit init

    #
    # The some files might be large so strip them while the firewall is still running
    # (restart command). This reduces the length of time that the firewall isn't
    # accepting new connections.
    #

    strip_file rules
    strip_file proxyarp
    strip_file maclist
    strip_file nat

    terminator=fatal_error

    deletechain shorewall

    [ -n "$NAT_ENABLED" ] && delete_nat

    delete_proxy_arp

    [ -n "$MANGLE_ENABLED" ] && \
	run_iptables -t mangle -F && \
	run_iptables -t mangle -X

    [ -n "$CLEAR_TC" ] && delete_tc

    echo "Deleting user chains..."

    setpolicy INPUT DROP
    setpolicy OUTPUT DROP
    setpolicy FORWARD DROP

    deleteallchains

    setcontinue FORWARD
    setcontinue INPUT
    setcontinue OUTPUT

    #
    # Enable the Loopback interface
    #
    run_iptables -A INPUT  -i lo -j ACCEPT
    run_iptables -A OUTPUT -o lo -j ACCEPT

    #
    # Allow DNS lookups during startup for FQDNs and deep-six INVALID packets
    #

    for chain in INPUT OUTPUT FORWARD; do
	run_iptables -A $chain -p udp --dport 53 -j ACCEPT
	run_iptables -A $chain -p ! icmp -m state --state INVALID -j DROP
    done

    [ -n "$CLAMPMSS" ] && \
	run_iptables -A FORWARD -p tcp \
	    --tcp-flags SYN,RST SYN -j TCPMSS --clamp-mss-to-pmtu


    if [ -z "$NEWNOTSYN" ]; then
	createchain newnotsyn no

	for interface in `find_interfaces_by_option newnotsyn`; do
	    run_iptables -A newnotsyn -i $interface -p tcp --tcp-flags ACK ACK -j ACCEPT
	    run_iptables -A newnotsyn -i $interface -p tcp --tcp-flags RST RST -j ACCEPT
	    run_iptables -A newnotsyn -i $interface -p tcp --tcp-flags FIN FIN -j ACCEPT
	    run_iptables -A newnotsyn -i $interface -j RETURN
	done

	run_user_exit newnotsyn

	if [ -n "$LOGNEWNOTSYN" ]; then
	    log_rule $LOGNEWNOTSYN newnotsyn DROP
	fi

	run_iptables -A newnotsyn -j DROP
    fi

    createchain icmpdef no
    createchain common	no
    createchain reject	no
    createchain dynamic no

    if [ -f /var/lib/shorewall/save ]; then
	echo "Restoring dynamic rules..."

	while read target ignore1 ignore2 address rest; do
	    case $target in
	    DROP|reject)
		run_iptables2 -A dynamic -s $address -j $target
		;;
	    *)
		;;
	esac
	done < /var/lib/shorewall/save
    fi

    echo "Creating input Chains..."

    for interface in $all_interfaces; do
	createchain `forward_chain $interface` no
	run_iptables -A `forward_chain $interface` -j dynamic
	createchain `input_chain $interface`   no
	run_iptables -A `input_chain $interface`   -j dynamic
    done
}

#
# Build the common chain -- called during [re]start and refresh
#
build_common_chain() {

    #
    # Common ICMP rules
    #
    run_user_exit icmpdef
    #
    # Common rules in each chain
    #
    common=`find_file common`

    if [ -f $common ]; then
	. $common
    else
	. `find_file common.def`
    fi
    #
    # New Not Syn Stuff
    #
    if [ -n "$NEWNOTSYN" ]; then
	run_iptables -A common -p tcp --tcp-flags ACK ACK -j ACCEPT
	run_iptables -A common -p tcp --tcp-flags RST RST -j ACCEPT
	run_iptables -A common -p tcp --tcp-flags FIN FIN -j ACCEPT
    fi
    #
    # BROADCASTS
    #
    drop_broadcasts `find_broadcasts`
}

#
# Construct zone-independent rules
#
add_common_rules() {
    local savelogparms="$LOGPARMS"
    #
    # Reject Rules
    #
    run_iptables -A reject -p tcp  -j REJECT --reject-with tcp-reset
    run_iptables -A reject -p udp  -j REJECT
    #
    # Not all versions of iptables support these so don't complain if they don't work
    #
    qt iptables  -A reject -p icmp -j REJECT --reject-with icmp-host-unreachable
    if ! qt iptables -A reject -j REJECT --reject-with icmp-host-prohibited; then
	#
	# In case the above doesn't work
	#
	run_iptables -A reject -j REJECT
    fi
    #
    # dropunclean rules
    #
    interfaces="`find_interfaces_by_option dropunclean`"

    if [ -n "$interfaces" ]; then
	createchain badpkt no

	if [ -n "$LOGUNCLEAN" ]; then
	    
	    LOGPARMS="$LOGPARMS --log-ip-options"

	    log_rule $LOGUNCLEAN badpkt DROP -p ! tcp 

	    LOGPARMS="$LOGPARMS --log-tcp-options"

	    log_rule $LOGUNCLEAN badpkt DROP -p tcp 

	    LOGPARMS="$savelogparms"
	fi

	run_iptables -A badpkt -j DROP
	echo "Mangled/Invalid Packet filtering enabled on:"

	for interface in $interfaces; do
	    for chain in `first_chains $interface`; do
		run_iptables -A $chain --match unclean -j badpkt
	    done
	    echo "   $interface"
	done
    fi
    #
    # logunclean rules
    #
    interfaces="`find_interfaces_by_option logunclean`"

    if [ -n "$interfaces" ]; then
	createchain logpkt no

	[ -z"$LOGUNCLEAN" ] && LOGUNCLEAN=info

	LOGPARMS="$LOGPARMS --log-ip-options"

	log_rule $LOGUNCLEAN logpkt DROP -p ! tcp 

	LOGPARMS="$LOGPARMS --log-tcp-options"
	
	log_rule $LOGUNCLEAN logpkt DROP -p tcp 

	LOGPARMS="$savelogparms"

	echo "Mangled/Invalid Packet Logging enabled on:"

	for interface in $interfaces; do
	    for chain in `first_chains $interface`; do
		run_iptables -A $chain --match unclean -j logpkt
	    done
	    echo "   $interface"
	done
    fi

    build_common_chain

    #
    # DHCP
    #
    echo "Adding rules for DHCP"

    for interface in `find_interfaces_by_option dhcp`; do
	run_iptables -A `input_chain $interface` -p udp --dport 67:68 -j ACCEPT
	run_iptables -A OUTPUT	  -o $interface	 -p udp --dport 67:68 -j ACCEPT
    done

    #
    # RFC 1918
    #
    norfc1918_interfaces="`find_interfaces_by_option norfc1918`"

    if [ -n "$norfc1918_interfaces" ]; then
	echo "Enabling RFC1918 Filtering"

	strip_file rfc1918

	createchain rfc1918 no

	createchain logdrop no
	
	log_rule $RFC1918_LOG_LEVEL logdrop DROP
	
	run_iptables -A logdrop -j DROP

	if [ -n "$MANGLE_ENABLED" -a -z "$CONNTRACK_MATCH" ]; then
	    #
	    # Mangling is enabled but conntrack match isn't available -- 
	    # create a chain in the mangle table to filter RFC1918 destination
            # addresses. This must be done in the mangle table before we apply
	    # any DNAT rules in the nat table
	    #
	    # Also add a chain to log and drop any RFC1918 packets that we find
	    #
	    run_iptables -t mangle -N man1918
	    run_iptables -t mangle -N logdrop
	    log_rule $RFC1918_LOG_LEVEL logdrop DROP -t mangle
	    run_iptables -t mangle -A logdrop -j DROP
	fi

	while read subnet target; do
	    case $target in
		logdrop|DROP|RETURN)
		    ;;
		*)
		    fatal_error "Illegal target ($target) for $subnet"
		    ;;
	    esac

	    run_iptables2 -A rfc1918 -s $subnet -j $target

	    if [ -n "$CONNTRACK_MATCH" ]; then
		#
		# We have connection tracking match -- match on the original destination
		#
		run_iptables2 -A rfc1918 -m conntrack --ctorigdst $subnet -j $target
	    elif [ -n "$MANGLE_ENABLED" ]; then
		#
		# No connection tracking match but we have mangling -- add a rule to
		# the mangle table
		#
		run_iptables2 -t mangle -A man1918 -d $subnet -j $target
	    fi
	done < $TMP_DIR/rfc1918

	for interface in $norfc1918_interfaces; do
	    for chain in `first_chains $interface`; do
		run_iptables -A $chain -m state --state NEW -j rfc1918
	    done

	    [ -n "$MANGLE_ENABLED" -a -z "$CONNTRACK_MATCH" ] && \
		run_iptables -t mangle -A PREROUTING -m state --state NEW -i $interface -j man1918
	done

    fi

    interfaces=`find_interfaces_by_option tcpflags`

    if [ -n "$interfaces" ]; then
	echo "Setting up TCP Flags checking..."

	createchain tcpflags no

	if [ -n "$TCP_FLAGS_LOG_LEVEL" ]; then
	    createchain logflags no

	    savelogparms="$LOGPARMS"

	    LOGPARMS="$LOGPARMS --log-ip-options"

	    log_rule $TCP_FLAGS_LOG_LEVEL logflags $TCP_FLAGS_DISPOSITION 

	    LOGPARMS="$savelogparms"
	    
	    case $TCP_FLAGS_DISPOSITION in
		REJECT)
		    run_iptables -A logflags -j REJECT --reject-with tcp-reset
		    ;;
		*)
		    run_iptables -A logflags -j $TCP_FLAGS_DISPOSITION
		    ;;
	    esac

	    disposition="-j logflags"
	else
	    disposition="-j $TCP_FLAGS_DISPOSITION"
	fi

	run_iptables -A tcpflags -p tcp --tcp-flags ALL FIN,URG,PSH $disposition
	run_iptables -A tcpflags -p tcp --tcp-flags ALL NONE        $disposition
	run_iptables -A tcpflags -p tcp --tcp-flags SYN,RST SYN,RST $disposition
	run_iptables -A tcpflags -p tcp --tcp-flags SYN,FIN SYN,FIN $disposition
	#
	# There are a lot of probes to ports 80, 3128 and 8080 that use a source
	# port of 0. This catches them even if they are directed at an IP that
	# hosts a web server.
	#
	run_iptables -A tcpflags -p tcp --syn --sport 0             $disposition

	for interface in $interfaces; do
	    for chain in `first_chains $interface`; do
		run_iptables -A $chain -p tcp -j tcpflags
	    done
	done
    fi
    #
    # Process Black List
    #
    setup_blacklist

    #
    # Route Filtering
    #
    for f in /proc/sys/net/ipv4/conf/*/rp_filter; do
	echo 0 > $f
    done

    interfaces="`find_interfaces_by_option routefilter`"

    if [ -n "$interfaces" -o -n "$ROUTE_FILTER" ]; then
	echo "Setting up Kernel Route Filtering..."

	if [ -n "$ROUTE_FILTER" ]; then
	    echo 1 > /proc/sys/net/ipv4/conf/all/rp_filter
	else
	    echo 0 > /proc/sys/net/ipv4/conf/all/rp_filter

	    for interface in $interfaces; do
		file=/proc/sys/net/ipv4/conf/$interface/rp_filter
		if [ -f $file ]; then
		    echo 1 > $file
		else
		    error_message \
			"Warning: Cannot set route filtering on $interface"
		fi
	    done
	fi
    fi
    #
    # IP Forwarding
    #
    case "$IP_FORWARDING" in
    [Oo][Nn])
	echo 1 > /proc/sys/net/ipv4/ip_forward
	echo "IP Forwarding Enabled"
	;;
    [Oo][Ff][Ff])
	echo 0 > /proc/sys/net/ipv4/ip_forward
	echo "IP Forwarding Disabled!"
	;;
    esac
}

#
# Scan the policy file defining the necessary chains
# Add the appropriate policy rule(s) to the end of each canonical chain
#
apply_policy_rules() {
    #
    # Create policy chains
    #
    for chain in $all_policy_chains; do
	eval policy=\$${chain}_policy
	eval loglevel=\$${chain}_loglevel
	eval synparams=\$${chain}_synparams

	[ -n "$synparams" ] && setup_syn_flood_chain $chain $synparams

	if havechain $chain; then
	    [ -n "$synparams" ] && \
		run_iptables -I $chain 2 -p tcp --syn -j @$chain
	else
	    #
	    # The chain doesn't exist. Create the chain and add policy
	    # rules
	    #
	    # We must include the ESTABLISHED and RELATED state
	    # rule here to account for replys and reverse
	    # related sessions associated with sessions going
	    # in the other direction
	    #
	    createchain $chain yes

	    #
	    # If either client or server is 'all' then this MUST be
	    # a policy chain and we must apply the appropriate policy rules
	    #
	    # Otherwise, this is a canonical chain which will be handled in
	    # the for loop below
	    #
	    case $chain in
		all2*|*2all)
		    policy_rules $chain $policy $loglevel
		    ;;
	    esac

	    [ -n "$synparams" ] && \
		[ $policy = ACCEPT -o $policy = CONTINUE ] && \
		run_iptables -I $chain 2 -p tcp --syn -j @$chain
	fi

    done

    #
    # Add policy rules to canonical chains
    #
    for zone in $FW $zones; do
	for zone1 in $FW $zones; do
	    chain=${zone}2${zone1}
	    if havechain $chain; then
		run_user_exit $chain
		default_policy $zone $zone1
	    fi
	done
    done
}

#
# Activate the rules
#
activate_rules()
{
    local PREROUTING_rule=1
    local POSTROUTING_rule=1
    #
    # Jump to a NAT chain from one of the builtin nat chains
    #
    addnatjump() # $1 = BUILTIN chain, $2 = user chain, $3 - * other arguments
    {
	local sourcechain=$1 destchain=$2
	shift
	shift

	havenatchain $destchain && \
	    run_iptables -t nat -A $sourcechain $@ -j $destchain
    }

    #
    # Jump to a RULES chain from one of the builtin nat chains
    #
    # If NAT_BEFORE_RULES then append the rule to the chain; otherwise, insert
    # the jump near the front of the builtin chain
    #
    addrulejump() # $1 = BUILTIN chain, $2 = user chain, $3 - * other arguments
    {
	local sourcechain=$1 destchain=$2
	shift
	shift

	if havenatchain $destchain; then
	    if [ -n "$NAT_BEFORE_RULES" ]; then
		run_iptables -t nat -A $sourcechain $@ -j $destchain
	    else
		eval run_iptables -t nat -I $sourcechain \
		    \$${sourcechain}_rule $@ -j $destchain
		eval ${sourcechain}_rule=\$\(\(\$${sourcechain}_rule + 1\)\)
	    fi
	fi
    }

    #
    # Add jumps from the builtin chains to the nat chains
    #
    addnatjump PREROUTING  nat_in
    addnatjump POSTROUTING nat_out

    for interface in $all_interfaces; do
	addnatjump PREROUTING  `input_chain $interface`	 -i $interface
	addnatjump POSTROUTING `output_chain $interface` -o $interface
    done

    > ${STATEDIR}/chains
    > ${STATEDIR}/zones

    for zone in $zones; do
	eval source_hosts=\$${zone}_hosts

	echo $zone $source_hosts >> ${STATEDIR}/zones

	chain1=`rules_chain $FW $zone`
	chain2=`rules_chain $zone $FW`

	eval complex=\$${zone}_is_complex

	if [ -n "$complex" ]; then
	    frwd_chain=${zone}_frwd
	    createchain $frwd_chain No
	fi

	echo "$FW $zone $chain1" >> ${STATEDIR}/chains
	echo "$zone $FW $chain2" >> ${STATEDIR}/chains

	for host in $source_hosts; do
	    interface=${host%:*}
	    subnet=${host#*:}

	    run_iptables -A OUTPUT -o $interface -d $subnet -j $chain1

	    #
	    # Add jumps from the builtin chains for DNAT and SNAT rules
	    #
	    addrulejump PREROUTING  `dnat_chain $zone` -i $interface -s $subnet
	    addrulejump POSTROUTING `snat_chain $zone` -o $interface -d $subnet

	    run_iptables -A `input_chain $interface` -s $subnet -j $chain2

	    [ -n "$complex" ] && \
		run_iptables -A `forward_chain $interface` -s $subnet -j $frwd_chain
	done

	for zone1 in $zones; do

	    eval policy=\$${zone}2${zone1}_policy

	    [ "$policy" = NONE ] && continue

	    eval dest_hosts=\$${zone1}_hosts

	    chain="`rules_chain $zone $zone1`"

	    echo "$zone $zone1 $chain" >> ${STATEDIR}/chains

	    if [ $zone = $zone1 ]; then
		eval routeback=\"\$${zone}_routeback\"
	    else
		routeback=
	    fi
	    
	    for host in $source_hosts; do
		interface=${host%:*}

		if [ -n "$complex" ]; then
		    chain1=$frwd_chain
		else
		    chain1=`forward_chain $interface`
		fi

		for host1 in $dest_hosts; do
		    interface1=${host1%:*}
		    subnet1=${host1#*:}

		    if [ "$host" != "$host1" ] || list_search $host $routeback; then
			run_iptables -A $chain1 -o $interface1 -d $subnet1 -j $chain
		    fi
		done
	    done
	done
    done

    for interface in $all_interfaces; do
       run_iptables -A FORWARD -i $interface -j `forward_chain $interface`
       run_iptables -A INPUT   -i $interface -j `input_chain $interface`
       addnatjump POSTROUTING `masq_chain $interface` -o $interface
    done

    complete_standard_chain INPUT all $FW
    complete_standard_chain OUTPUT $FW all
    complete_standard_chain FORWARD all all
    #
    # Remove rules added to keep the firewall alive during [re]start"
    #
    for chain in INPUT OUTPUT FORWARD; do
	run_iptables -D $chain -m state --state ESTABLISHED,RELATED -j ACCEPT
	run_iptables -D $chain -p udp --dport 53 -j ACCEPT
    done
}

#
# Start/Restart the Firewall
#
define_firewall() # $1 = Command (Start or Restart)
{
    if [ -f /etc/shorewall/startup_disabled ]; then
	echo "   Shorewall Startup is disabled -- to enable startup"
	echo "   after you have completed Shorewall configuration,"
	echo "   remove the file /etc/shorewall/startup_disabled"

	[ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	my_mutex_off
	exit 2
    fi

    echo "${1}ing Shorewall..."

    verify_os_version

    verify_ip

    load_kernel_modules

    echo "Initializing..."

    initialize_netfilter

    echo "Configuring Proxy ARP"

    setup_proxy_arp

    setup_nat

    echo "Adding Common Rules"

    add_common_rules

    tunnels=`find_file tunnels`

    [ -f $tunnels ] && \
	echo "Processing $tunnels..." && setup_tunnels $tunnels

    maclist_hosts=`find_hosts_by_option maclist`

    if [ -n "$maclist_hosts" ] ; then
	setup_mac_lists
    fi

    rules=`find_file rules`

    echo "Processing $rules..."

    process_rules $rules

    policy=`find_file policy`

    echo "Processing $policy..."

    apply_policy_rules

    masq=`find_file masq`

    [ -f $masq ] && setup_masq $masq

    tos=`find_file tos`

    [ -f $tos ] && [ -n "$MANGLE_ENABLED" ] && process_tos $tos

    ecn=`find_file ecn`

    [ -f $ecn ] && [ -n "$MANGLE_ENABLED" ] && setup_ecn $ecn

    [ -n "$TC_ENABLED" ] && setup_tc

    echo "Activating Rules..."

    activate_rules

    [ -n "$aliases_to_add" ] && \
	echo "Adding IP Addresses..." && \
	add_ip_aliases

    run_user_exit start

    createchain shorewall no

    date > $STATEDIR/restarted

    report "Shorewall ${1}ed"

    rm -rf $TMP_DIR
}

#
# Rebuild the common chain
#
refresh_firewall()
{
    echo "Refreshing Shorewall..."

    echo "Determining Zones and Interfaces..."

    determine_zones

    validate_interfaces_file

    [ -z "$zones" ] && startup_error "No Zones Defined"

    determine_interfaces

    run_user_exit refresh

    run_iptables -F common

    echo "Adding Common Rules"

    build_common_chain

    #
    # Blacklist
    #
    refresh_blacklist

    ecn=`find_file ecn`

    [ -f $ecn ] && [ -n "$MANGLE_ENABLED" ] && setup_ecn $ecn
    #
    # Refresh Traffic Control
    #
    [ -n "$TC_ENABLED" ] && refresh_tc

    report "Shorewall Refreshed"

    rm -rf $TMP_DIR
}

#
# Add a host or subnet to a zone
#
add_to_zone() # $1 = <interface>[:<hosts>] $2 = zone
{
    local base

    nat_chain_exists() # $1 = chain name
    {
	qt iptables -t nat -L $1 -n
    }

    do_iptables() # $@ = command
    {
	if ! iptables $@ ; then
	    startup_error "Can't add $1 to zone $2"
	fi
    }

    output_rule_num() {
	local num=`iptables -L OUTPUT -n --line-numbers | grep icmp | cut -d' ' -f1 | head -n1`

	[ -n "$num" ] && echo $(($num+1))
    }
    #
    # Isolate interface and host parts
    #
    interface=${1%:*}
    host=${1#*:}

    [ -z "$host" ] && host="0.0.0.0/0"
    #
    # Load $zones
    #
    determine_zones
    #
    # Validate Zone
    #
    zone=$2

    validate_zone $zone || startup_error "Unknown zone: $zone"

    [ "$zone" = $FW ] && startup_error "Can't add $1 to firewall zone"
    #
    # Be sure that Shorewall has been restarted using a DZ-aware version of the code
    #
    [ -f ${STATEDIR}/chains ] || startup_error "${STATEDIR}/chains -- file not found"
    [ -f ${STATEDIR}/zones ]  || startup_error "${STATEDIR}/zones -- file not found"
    #
    # Be sure that the interface was present at last [re]start
    #
    if ! chain_exists `input_chain $interface` ; then
	startup_error "Unknown interface $interface"
    fi
    #
    # Build lists of interfaces with special rules
    #
    dhcp_interfaces=`find_interfaces_by_option dhcp`
    blacklist_interfaces=`find_interfaces_by_option blacklist`
    maclist_interfaces=`find_interfaces_by_option maclist`
    tcpflags_interfaces=`find_interfaces_by_option tcpflags`
    #
    # Normalize the first argument to this function
    #
    newhost="$interface:$host"

    terminator=fatal_error
    #
    # Create a new Zone state file
    #
    > ${STATEDIR}/zones_$$
    #
    # Add $1 to the Zone state file
    #
    while read z hosts; do
	if [ "$z" = "$zone" ]; then
	    for h in $hosts; do
		if [ "$h" = "$newhost" ]; then
		    rm -f ${STATEDIR}/zones_$$
		    startup_error "$1 already in zone $zone"
		fi
	    done

	    [ -z "$hosts" ] && hosts=$newhost || hosts="$hosts $newhost"
	fi

	eval ${z}_hosts=\"$hosts\"

	echo "$z $hosts" >> ${STATEDIR}/zones_$$
    done < ${STATEDIR}/zones

    mv -f ${STATEDIR}/zones_$$ ${STATEDIR}/zones
    #
    # If the zone passed in the command has a dnat chain then insert a rule in
    # the nat table PREROUTING chain to jump to that chain when the source
    # matches the new host(s)#
    #
    chain=${zone}_dnat

    if nat_chain_exists $chain; then
	do_iptables -t nat -I PREROUTING -i $interface -s $host -j $chain
    fi
    #
    # Insert new rules into the input chains for the passed interface
    #
    while read z1 z2 chain; do
	if [ "$z1" = "$zone" ]; then
	    if [ "$z2" = "$FW" ]; then
		#
		# We will insert the rule right after the DHCP, 'ping' and
                # MAC rules (if any)
		#
		if list_search $interface $dhcp_interfaces; then
		    rulenum=3
		else
		    rulenum=2
		fi

		if list_search $interface $maclist_interfaces; then
		    rulenum=$(($rulenum + 1))
		fi

		if list_search $interface $tcpflags_interfaces; then
		    rulenum=$(($rulenum + 1))
		fi

		do_iptables -I `input_chain $interface` $rulenum -s $host -j $chain
	    else
		#
		# Insert rules into the passed interface's forward chain
		#
		# We insert them after any blacklist/MAC verification rules
		#
		source_chain=`forward_chain $interface`
		eval dest_hosts=\"\$${z2}_hosts\"

		base=`chain_base $interface`

		eval rulenum=\$${base}_rulenum

		if [ -z "$rulenum" ]; then
		    if list_search $interface $blacklist_interfaces; then
			rulenum=3
		    else
			rulenum=2
		    fi

		    if list_search $interface $maclist_interfaces; then
			rulenum=$(($rulenum + 1))
		    fi

		    if list_search $interface $tcpflags_interfaces; then
			rulenum=$(($rulenum + 1))
		    fi
		fi

		for h in $dest_hosts; do
		    iface=${h%:*}
		    hosts=${h#*:}

		    if [ "$iface" != "$interface" -o "$hosts" != "$host" ]; then
			do_iptables -I $source_chain $rulenum -s $host -o $iface -d $hosts -j $chain
			rulenum=$(($rulenum + 1))
		    fi
		done

		eval ${base}_rulenum=$rulenum

	    fi
	elif [ "$z2" = "$zone" ]; then
	    if [ "$z1" = "$FW" ]; then
		#
		# Add a rule to the OUTPUT chain -- always after the icmp * ACCEPT rule
		#
		do_iptables -I OUTPUT `output_rule_num` -o $interface -d $host -j $chain
	    else
		#
		# Insert rules into the source interface's forward chain
		#
		# We insert them after any blacklist rules
		#
		eval source_hosts=\"\$${z1}_hosts\"

		for h in $source_hosts; do
		    iface=${h%:*}
		    hosts=${h#*:}

		    base=`chain_base $iface`

		    eval rulenum=\$${base}_rulenum

		    if [ -z "$rulenum" ]; then
			if list_search $iface $blacklist_interfaces; then
			    rulenum=3
			else
			    rulenum=2
			fi
		    fi

		    if [ "$iface" != "$interface" -o "$hosts" != "$host" ]; then
			do_iptables -I `forward_chain $iface` $rulenum -s $hosts -o $interface -d $host -j $chain
			rulenum=$(($rulenum + 1))
		    fi

		    eval ${base}_rulenum=$rulenum
		done
	    fi
	fi
    done < ${STATEDIR}/chains

    rm -rf $TMP_DIR
    
    echo "$1 added to zone $2"
}

#
# Delete a host or subnet from a zone
#
delete_from_zone() # $1 = <interface>[:<hosts>] $2 = zone
{
    #
    # Delete the subnect host(s) from the zone state file
    #
    delete_from_zones_file()
    {
	> ${STATEDIR}/zones_$$

	while read z hosts; do
	    if [ "$z" = "$zone" ]; then
		temp=$hosts
		hosts=

		for h in $temp; do
		    if [ "$h" = "$delhost" ]; then
			echo Yes
		    else
			hosts="$hosts $h"
		    fi
		done
	    fi

	    echo "$z $hosts" >> ${STATEDIR}/zones_$$
	done < ${STATEDIR}/zones

	mv -f ${STATEDIR}/zones_$$ ${STATEDIR}/zones
    }
    #
    # Isolate interface and host parts
    #
    interface=${1%:*}
    host=${1#*:}

    [ -z "$host" ] && host="0.0.0.0/0"
    #
    # Load $zones
    #
    determine_zones

    zone=$2

    validate_zone $zone || startup_error "Unknown zone: $zone"

    [ "$zone" = $FW ] && startup_error "Can't remove $1 from firewall zone"
    #
    # Be sure that Shorewall has been restarted using a DZ-aware version of the code
    #
    [ -f ${STATEDIR}/chains ] || startup_error "${STATEDIR}/chains -- file not found"
    [ -f ${STATEDIR}/zones ]  || startup_error "${STATEDIR}/zones -- file not found"
    #
    # Be sure that the interface was present at last [re]start
    #
    if ! chain_exists `input_chain $interface` ; then
	startup_error "Unknown interface $interface"
    fi
    #
    # Normalize the first argument to this function
    #
    delhost="$interface:$host"
    #
    # Delete the passed hosts from the zone state file
    #
    [ -z "`delete_from_zones_file`" ] && \
	error_message "Warning: $1 does not appear to be in zone $2"
    #
    # Construct the zone host maps
    #
    while read z hosts; do
	eval ${z}_hosts=\"$hosts\"
    done < ${STATEDIR}/zones

    terminator=fatal_error
    #
    # Delete any nat table entries for the host(s)
    #
    qt iptables -t nat -D PREROUTING -i $interface -s $host -j ${zone}_dnat
    #
    # Delete rules rules the input chains for the passed interface
    #
    while read z1 z2 chain; do
	if [ "$z1" = "$zone" ]; then
	    if [ "$z2" = "$FW" ]; then
		qt iptables -D `input_chain $interface` -s $host -j $chain
	    else
		source_chain=`forward_chain $interface`
		eval dest_hosts=\"\$${z2}_hosts\"

		for h in $dest_hosts $delhost; do
		    iface=${h%:*}
		    hosts=${h#*:}

		    if [ "$iface" != "$interface" -o "$hosts" != "$host" ]; then
			qt iptables -D $source_chain -s $host -o $iface -d $hosts -j $chain
		    fi
		done
	    fi
	elif [ "$z2" = "$zone" ]; then
	    if [ "$z1" = "$FW" ]; then
		qt iptables -D OUTPUT -o $interface -d $host -j $chain
	    else
		eval source_hosts=\"\$${z1}_hosts\"
		
		for h in $source_hosts; do
		    iface=${h%:*}
		    hosts=${h#*:}

		    if [ "$iface" != "$interface" -o "$hosts" != "$host" ]; then
			qt iptables -D `forward_chain $iface` -s $hosts -o $interface -d $host -j $chain
		    fi
		done
	    fi
	fi
    done < ${STATEDIR}/chains

    rm -rf $TMP_DIR

    echo "$1 removed from zone $2"
}

#
# Determine the value for a parameter that defaults to Yes
#
added_param_value_yes() # $1 = Parameter Name, $2 = Parameter value
{
    local val="$2"

    if [ -z "$val" ]; then
	echo "Yes"
    else case $val in
	[Yy][Ee][Ss])
	    echo "Yes"
	    ;;
	[Nn][Oo])
	    echo ""
	    ;;
	*)
	    startup_error "Invalid value ($val) for $1"
	    ;;
	esac
    fi
}

#
# Determine the value for a parameter that defaults to No
#
added_param_value_no() # $1 = Parameter Name, $2 = Parameter value
{
    local val="$2"

    if [ -z "$val" ]; then
	echo ""
    else case $val in
	[Yy][Ee][Ss])
	    echo "Yes"
	    ;;
	[Nn][Oo])
	    echo ""
	    ;;
	*)
	    startup_error "Invalid value ($val) for $1"
	    ;;
	esac
    fi
}

#
# Initialize this program
#
do_initialize() {
    #
    # Code to see if shell can support xor
    #
    check_xor() {
	echo $(( 256 ^ -1 ))
    }
    
    check_xor1() {
	local x=`check_xor 2> /dev/null`
	[ -n "$x" ] && echo $x || echo 0
    }

    # Run all utility programs using the C locale
    #
    # Thanks to Vincent Planchenault for this tip #

    export LC_ALL=C

    PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin
    #
    # Establish termination function
    #
    terminator=startup_error
    #
    # So that emacs doesn't get lost
    #
    LEFTSHIFT='<<'
    #
    # Clear all configuration variables
    #
    version=
    FW=
    SUBSYSLOCK=
    STATEDIR=
    ALLOWRELATED=Yes
    LOGRATE=
    LOGBURST=
    LOGPARMS=
    ADD_IP_ALIASES=
    ADD_SNAT_ALIASES=
    TC_ENABLED=
    LOGUNCLEAN=
    BLACKLIST_DISPOSITION=
    BLACKLIST_LOGLEVEL=
    CLAMPMSS=
    ROUTE_FILTER=
    NAT_BEFORE_RULES=
    DETECT_DNAT_IPADDRS=
    MUTEX_TIMEOUT=
    NEWNOTSYN=
    LOGNEWNOTSYN=
    FORWARDPING=
    MACLIST_DISPOSITION=
    MACLIST_LOG_LEVEL=
    TCP_FLAGS_DISPOSITION=
    TCP_FLAGS_LOG_LEVEL=
    RFC1918_LOG_LEVEL=
    MARK_IN_FORWARD_CHAIN=
    SHARED_DIR=/usr/share/shorewall
    FUNCTIONS=
    VERSION_FILE=
    LOGFORMAT=
    LOGRULENUMBERS=

    stopping=
    have_mutex=
    masq_seq=1
    nonat_seq=1
    aliases_to_add=

    TMP_DIR=/tmp/shorewall-$$
    rm -rf $TMP_DIR
    mkdir -p $TMP_DIR && chmod 700 $TMP_DIR || \
       startup_error "Can't create $TMP_DIR"

    trap "rm -rf $TMP_DIR; my_mutex_off; exit 2" 1 2 3 4 5 6 9

    FUNCTIONS=$SHARED_DIR/functions

    if [ -f $FUNCTIONS ]; then
	. $FUNCTIONS
    else
	startup_error "$FUNCTIONS does not exist!"
    fi

    VERSION_FILE=$SHARED_DIR/version

    [ -f $VERSION_FILE ] && version=`cat $VERSION_FILE`

    run_user_exit params

    config=`find_file shorewall.conf`

    if [ -f $config ]; then
	echo "Processing $config..."
	. $config
    else
	echo "$config does not exist!" >&2
	exit 2
    fi

    [ -z "${STATEDIR}" ] && STATEDIR=/var/state/shorewall

    [ -d $STATEDIR ] || mkdir -p $STATEDIR

    [ -z "$FW" ] && FW=fw

    ALLOWRELATED="`added_param_value_yes ALLOWRELATED $ALLOWRELATED`"
    [ -n "$ALLOWRELATED" ] || \
	startup_error "ALLOWRELATED=No is not supported"
    ADD_IP_ALIASES="`added_param_value_yes ADD_IP_ALIASES $ADD_IP_ALIASES`"
    TC_ENABLED="`added_param_value_yes TC_ENABLED $TC_ENABLED`"

    if [ -n "${LOGRATE}${LOGBURST}" ]; then
	LOGPARMS="--match limit"
	[ -n "$LOGRATE" ]  && LOGPARMS="$LOGPARMS --limit $LOGRATE"
	[ -n "$LOGBURST" ] && LOGPARMS="$LOGPARMS --limit-burst $LOGBURST"
    fi

    if [ -n "$IP_FORWARDING" ]; then
	case "$IP_FORWARDING" in
	[Oo][Nn]|[Oo][Ff][Ff]|[Kk][Ee][Ee][Pp])
	    ;;
	*)
	    startup_error "Invalid value ($IP_FORWARDING) for IP_FORWARDING"
	    ;;
	esac
    else
	IP_FORWARDING=On
    fi

    if [ -n "$TC_ENABLED" -a -z "$MANGLE_ENABLED" ]; then
	startup_error "Traffic Control requires Mangle"
    fi

    [ -z "$BLACKLIST_DISPOSITION" ] && BLACKLIST_DISPOSITION=DROP

    CLAMPMSS=`added_param_value_no CLAMPMSS $CLAMPMSS`
    ADD_SNAT_ALIASES=`added_param_value_no ADD_SNAT_ALIASES $ADD_SNAT_ALIASES`
    ROUTE_FILTER=`added_param_value_no ROUTE_FILTER $ROUTE_FILTER`
    NAT_BEFORE_RULES=`added_param_value_yes NAT_BEFORE_RULES $NAT_BEFORE_RULES`
    DETECT_DNAT_IPADDRS=`added_param_value_no DETECT_DNAT_IPADDRS $DETECT_DNAT_IPADDRS`
    FORWARDPING=`added_param_value_no FORWARDPING $FORWARDPING`
    [ -n "$FORWARDPING" ] && \
	startup_error "FORWARDPING=Yes is no longer supported"

    NEWNOTSYN=`added_param_value_yes NEWNOTSYN $NEWNOTSYN`

    maclist_target=reject

    if [ -n "$MACLIST_DISPOSITION" ] ; then
	case $MACLIST_DISPOSITION in
	    REJECT)
		;;
	    ACCEPT|DROP)
		maclist_target=$MACLIST_DISPOSITION
		;;
	    *)
		startup_error "Invalid value ($MACLIST_DISPOSITION) for MACLIST_DISPOSITION"
		;;
	esac
    else
	MACLIST_DISPOSITION=REJECT
    fi

    if [ -n "$TCP_FLAGS_DISPOSITION" ] ; then
	case $TCP_FLAGS_DISPOSITION in
	    REJECT|ACCEPT|DROP)
		;;
	    *)
		startup_error "Invalid value ($TCP_FLAGS_DISPOSITION) for TCP_FLAGS_DISPOSITION"
		;;
	esac
    else
	TCP_FLAGS_DISPOSITION=DROP
    fi

    [ -z "$RFC1918_LOG_LEVEL" ] && RFC1918_LOG_LEVEL=info
    MARK_IN_FORWARD_CHAIN=`added_param_value_no MARK_IN_FORWARD_CHAIN $MARK_IN_FORWARD_CHAIN`
    [ -n "$MARK_IN_FORWARD_CHAIN" ] && marking_chain=tcfor || marking_chain=tcpre
    if [ -n "$TC_ENABLED" ]; then
	CLEAR_TC=`added_param_value_yes CLEAR_TC $CLEAR_TC`
    else
	CLEAR_TC=
    fi
    
    if [ -n "$LOGFORMAT" ]; then
	if [ -n "`echo $LOGFORMAT | grep '%d'`" ]; then
	    LOGRULENUMBERS=Yes
	    temp=`printf "$LOGFORMAT" fooxx 1 barxx 2> /dev/null`
	    if [ $? -ne 0 ]; then
		startup_error "Invalid LOGFORMAT string: \"$LOGFORMAT\""
	    fi
	else
	    temp=`printf "$LOGFORMAT" fooxx barxx 2> /dev/null`
	    if [ $? -ne 0 ]; then
		startup_error "Invalid LOGFORMAT string: \"$LOGFORMAT\""
	    fi
	fi

	if [ ${#temp} -gt 29 ]; then
	    startup_error "LOGFORMAT string is too long: \"$LOGFORMAT\""
	fi
    else
	LOGFORMAT="Shorewall:%s:%s:"
    fi
    #
    # Strip the files that we use often
    #
    strip_file interfaces
    strip_file hosts
    #
    # Determine the capabilities of the installed iptables/netfilter
    #
    determine_capabilities
    #
    # Check out the user's shell
    #
    [ -n "$SHOREWALL_SHELL" ] || SHOREWALL_SHELL=/bin/sh

    temp=`decodeaddr 192.168.1.1` 
    if [ `encodeaddr $temp` != 192.168.1.1 -o `check_xor1` -ne -257 ]; then
	startup_error "Shell $SHOREWALL_SHELL is broken and may not be used with Shorewall"
    fi
}

#
# Give Usage Information
#
usage() {
    echo "Usage: $0 [debug] {start|stop|reset|restart|status|refresh|clear|{add|delete} <interface>[:hosts] zone}}"
    exit 1
}

#
# E X E C U T I O N    B E G I N S   H E R E
#
#
# Start trace if first arg is "debug"
#
[ $# -gt 1 ] && [ "$1" = "debug" ] && { set -x ; shift ; }

nolock=

[ $# -gt 1 ] && [ "$1" = "nolock" ] && { nolock=Yes; shift ; }

trap "my_mutex_off; exit 2" 1 2 3 4 5 6 9

command="$1"

case "$command" in
    stop)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	echo -n "Stopping Shorewall..."
	stop_firewall
	[ -n "$SUBSYSLOCK" ] && rm -f $SUBSYSLOCK
	echo "done."
	my_mutex_off
	;;

    start)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	if qt iptables -L shorewall -n ; then
	    [ -n "$SUBSYSLOCK" ] && touch $SUBSYSLOCK
	    echo "Shorewall Already Started"
	    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	    my_mutex_off
	    exit 0;
	fi
	define_firewall "Start" && [ -n "$SUBSYSLOCK" ] && touch $SUBSYSLOCK
	my_mutex_off
	;;

    restart)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	if qt iptables -L shorewall -n ; then
	    define_firewall "Restart"
	else
	    echo "Shorewall Not Currently Running"
	    define_firewall "Start"
	fi

	[ $? -eq 0 ] && [ -n "$SUBSYSLOCK" ] && touch $SUBSYSLOCK
	my_mutex_off
	;;

    status)
	[ $# -ne 1 ] && usage
	echo "Shorewall-$version Status at $HOSTNAME - `date`"
	echo
	iptables -L -n -v
	;;

    reset)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	if ! qt iptables -L shorewall -n ; then
	    echo "Shorewall Not Started"
	    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	    my_mutex_off
	    exit 2;
	fi
	iptables -Z
	iptables -t nat -Z
	iptables -t mangle -Z
	report "Shorewall Counters Reset"
	date > $STATEDIR/restarted
	my_mutex_off
	;;

    refresh)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	if ! qt iptables -L shorewall -n ; then
	    echo "Shorewall Not Started"
	    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	    my_mutex_off
	    exit 2;
	fi
	refresh_firewall;
	my_mutex_off
	;;

    clear)
	[ $# -ne 1 ] && usage
	do_initialize
	my_mutex_on
	echo -n "Clearing Shorewall..."
	clear_firewall
	[ -n "$SUBSYSLOCK" ] && rm -f $SUBSYSLOCK
	echo "done."
	my_mutex_off
	;;

    check)
	[ $# -ne 1 ] && usage
	do_initialize
	check_config
	;;

    add)
	[ $# -ne 3 ] && usage
	do_initialize
	my_mutex_on
	if ! qt iptables -L shorewall -n ; then
	    echo "Shorewall Not Started"
	    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	    my_mutex_off
	    exit 2;
	fi
	add_to_zone $2 $3
	my_mutex_off
	;;

    delete)
	[ $# -ne 3 ] && usage
	do_initialize
	my_mutex_on
	if ! qt iptables -L shorewall -n ; then
	    echo "Shorewall Not Started"
	    [ -n "$TMP_DIR" ] && rm -rf $TMP_DIR
	    my_mutex_off
	    exit 2;
	fi
	delete_from_zone $2 $3
	my_mutex_off
	;;

    *)
	usage
	;;

esac