xref: /freebsd/share/examples/jails/jng (revision faf25f48d601ae39f5752602f3020e2e92605625)
1#!/bin/sh
2#-
3# Copyright (c) 2016 Devin Teske
4# All rights reserved.
5#
6# Redistribution and use in source and binary forms, with or without
7# modification, are permitted provided that the following conditions
8# are met:
9# 1. Redistributions of source code must retain the above copyright
10#    notice, this list of conditions and the following disclaimer.
11# 2. Redistributions in binary form must reproduce the above copyright
12#    notice, this list of conditions and the following disclaimer in the
13#    documentation and/or other materials provided with the distribution.
14#
15# THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18# ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21# OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22# HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24# OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25# SUCH DAMAGE.
26#
27# $FreeBSD$
28#
29############################################################ IDENT(1)
30#
31# $Title: netgraph(4) management script for vnet jails $
32#
33############################################################ INFORMATION
34#
35# Use this tool with jail.conf(5) (or rc.conf(5) ``legacy'' configuration) to
36# manage `vnet' interfaces for jails. Designed to automate the creation of vnet
37# interface(s) during jail `prestart' and destroy said interface(s) during jail
38# `poststop'.
39#
40# In jail.conf(5) format:
41#
42# ### BEGIN EXCERPT ###
43#
44# xxx {
45# 	host.hostname = "xxx.yyy";
46# 	path = "/vm/xxx";
47#
48# 	#
49# 	# NB: Below 2-lines required
50# 	# NB: The number of ngN_xxx interfaces should match the number of
51# 	#     arguments given to `jng bridge xxx' in exec.prestart value.
52# 	#
53# 	vnet;
54# 	vnet.interface = ng0_xxx, ng1_xxx, ...;
55#
56# 	exec.clean;
57# 	exec.system_user = "root";
58# 	exec.jail_user = "root";
59#
60# 	#
61# 	# NB: Below 2-lines required
62# 	# NB: The number of arguments after `jng bridge xxx' should match
63# 	#     the number of ngN_xxx arguments in vnet.interface value.
64# 	#
65# 	exec.prestart += "jng bridge xxx em0 em1 ...";
66# 	exec.poststop += "jng shutdown xxx";
67#
68# 	# Standard recipe
69# 	exec.start += "/bin/sh /etc/rc";
70# 	exec.stop = "/bin/sh /etc/rc.shutdown jail";
71# 	exec.consolelog = "/var/log/jail_xxx_console.log";
72# 	mount.devfs;
73#
74# 	# Optional (default off)
75# 	#allow.mount;
76# 	#allow.set_hostname = 1;
77# 	#allow.sysvipc = 1;
78# 	#devfs_ruleset = "11"; # rule to unhide bpf for DHCP
79# }
80#
81# ### END EXCERPT ###
82#
83# In rc.conf(5) ``legacy'' format (used when /etc/jail.conf does not exist):
84#
85# ### BEGIN EXCERPT ###
86#
87# jail_enable="YES"
88# jail_list="xxx"
89#
90# #
91# # Global presets for all jails
92# #
93# jail_devfs_enable="YES"	# mount devfs
94#
95# #
96# # Global options (default off)
97# #
98# #jail_mount_enable="YES"		# mount /etc/fstab.{name}
99# #jail_set_hostname_allow="YES"	# Allow hostname to change
100# #jail_sysvipc_allow="YES"		# Allow SysV Interprocess Comm.
101#
102# # xxx
103# jail_xxx_hostname="xxx.shxd.cx"		# hostname
104# jail_xxx_rootdir="/vm/xxx"			# root directory
105# jail_xxx_vnet_interfaces="ng0_xxx ng1xxx ..."	# vnet interface(s)
106# jail_xxx_exec_prestart0="jng bridge xxx em0 em1 ..."	# bridge interface(s)
107# jail_xxx_exec_poststop0="jng shutdown xxx"	# destroy interface(s)
108# #jail_xxx_mount_enable="YES"			# mount /etc/fstab.xxx
109# #jail_xxx_devfs_ruleset="11"			# rule to unhide bpf for DHCP
110#
111# ### END EXCERPT ###
112#
113# Note that the legacy rc.conf(5) format is converted to
114# /var/run/jail.{name}.conf by /etc/rc.d/jail if jail.conf(5) is missing.
115#
116# ASIDE: dhclient(8) inside a vnet jail...
117#
118# To allow dhclient(8) to work inside a vnet jail, make sure the following
119# appears in /etc/devfs.rules (which should be created if it doesn't exist):
120#
121# 	[devfsrules_jail=11]
122# 	add include $devfsrules_hide_all
123# 	add include $devfsrules_unhide_basic
124# 	add include $devfsrules_unhide_login
125# 	add path 'bpf*' unhide
126#
127# And set ether devfs.ruleset="11" (jail.conf(5)) or
128# jail_{name}_devfs_ruleset="11" (rc.conf(5)).
129#
130# NB: While this tool can't create every type of desirable topology, it should
131# handle most setups, minus some which considered exotic or purpose-built.
132#
133############################################################ GLOBALS
134
135pgm="${0##*/}" # Program basename
136
137#
138# Global exit status
139#
140SUCCESS=0
141FAILURE=1
142
143############################################################ FUNCTIONS
144
145usage()
146{
147	local action usage descr
148	exec >&2
149	echo "Usage: $pgm action [arguments]"
150	echo "Actions:"
151	for action in \
152		bridge		\
153		graph		\
154		show		\
155		show1		\
156		shutdown	\
157		stats		\
158	; do
159		eval usage=\"\$jng_${action}_usage\"
160		[ "$usage" ] || continue
161		eval descr=\"\$jng_${action}_descr\"
162		printf "\t%s\n\t\t%s\n" "$usage" "$descr"
163	done
164	exit $FAILURE
165}
166
167action_usage()
168{
169	local usage descr action="$1"
170	eval usage=\"\$jng_${action}_usage\"
171	echo "Usage: $pgm $usage" >&2
172	eval descr=\"\$jng_${action}_descr\"
173	printf "\t%s\n" "$descr"
174	exit $FAILURE
175}
176
177derive_mac()
178{
179	local OPTIND=1 OPTARG __flag
180	local __mac_num= __make_pair=
181	while getopts 2n: __flag; do
182		case "$__flag" in
183		2) __make_pair=1 ;;
184		n) __mac_num=${OPTARG%%[^0-9]*} ;;
185		esac
186	done
187	shift $(( $OPTIND - 1 ))
188
189	if [ ! "$__mac_num" ]; then
190		eval __mac_num=\${_${iface}_num:--1}
191		__mac_num=$(( $__mac_num + 1 ))
192		eval _${iface}_num=\$__mac_num
193	fi
194
195	local __iface="$1" __name="$2" __var_to_set="$3" __var_to_set_b="$4"
196	local __iface_devid __new_devid __num __new_devid_b
197	#
198	# Calculate MAC address derived from given iface.
199	#
200	# The formula I'm using is ``NP:SS:SS:II:II:II'' where:
201	# + N denotes 4 bits used as a counter to support branching
202	#   each parent interface up to 15 times under the same jail
203	#   name (see S below).
204	# + P denotes the special nibble whose value, if one of
205	#   2, 6, A, or E (but usually 2) denotes a privately
206	#   administered MAC address (while remaining routable).
207	# + S denotes 16 bits, the sum(1) value of the jail name.
208	# + I denotes bits that are inherited from parent interface.
209	#
210	# The S bits are a CRC-16 checksum of NAME, allowing the jail
211	# to change link numbers in ng_bridge(4) without affecting the
212	# MAC address. Meanwhile, if...
213	#   + the jail NAME changes (e.g., it was duplicated and given
214	#     a new name with no other changes)
215	#   + the underlying network interface changes
216	#   + the jail is moved to another host
217	# the MAC address will be recalculated to a new, similarly
218	# unique value preventing conflict.
219	#
220	__iface_devid=$( ifconfig $__iface ether | awk '/ether/,$0=$2' )
221	# ??:??:??:II:II:II
222	__new_devid=${__iface_devid#??:??:??} # => :II:II:II
223	# => :SS:SS:II:II:II
224	__num=$( set -- `echo -n "$__name" | sum` && echo $1 )
225	__new_devid=$( printf :%02x:%02x \
226		$(( $__num >> 8 & 255 )) $(( $__num & 255 )) )$__new_devid
227	# => P:SS:SS:II:II:II
228	case "$__iface_devid" in
229	   ?2:*) __new_devid=a$__new_devid __new_devid_b=e$__new_devid ;;
230	?[Ee]:*) __new_devid=2$__new_devid __new_devid_b=6$__new_devid ;;
231	      *) __new_devid=2$__new_devid __new_devid_b=e$__new_devid
232	esac
233	# => NP:SS:SS:II:II:II
234	__new_devid=$( printf %x $(( $__mac_num & 15 )) )$__new_devid
235	__new_devid_b=$( printf %x $(( $__mac_num & 15 )) )$__new_devid_b
236
237	#
238	# Return derivative MAC address(es)
239	#
240	if [ "$__make_pair" ]; then
241		if [ "$__var_to_set" -a "$__var_to_set_b" ]; then
242			eval $__var_to_set=\$__new_devid
243			eval $__var_to_set_b=\$__new_devid_b
244		else
245			echo $__new_devid $__new_devid_b
246		fi
247	else
248		if [ "$__var_to_set" ]; then
249			eval $__var_to_set=\$__new_devid
250		else
251			echo $__new_devid
252		fi
253	fi
254}
255
256mustberoot_to_continue()
257{
258	if [ "$( id -u )" -ne 0 ]; then
259		echo "Must run as root!" >&2
260		exit $FAILURE
261	fi
262}
263
264jng_bridge_usage="bridge [-b BRIDGE_NAME] NAME [!|=]iface0 [[!|=]iface1 ...]"
265jng_bridge_descr="Create ng0_NAME [ng1_NAME ...]"
266jng_bridge()
267{
268	local OPTIND=1 OPTARG flag bridge=bridge
269	while getopts b: flag; do
270		case "$flag" in
271		b) bridge="$OPTARG"
272		   [ "$bridge" ] || action_usage bridge ;; # NOTREACHED
273		*) action_usage bridge # NOTREACHED
274		esac
275	done
276	shift $(( $OPTIND - 1 ))
277
278	local name="$1"
279	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -gt 1 ] ||
280		action_usage bridge # NOTREACHED
281	shift 1 # name
282
283	mustberoot_to_continue
284
285	local iface parent eiface eiface_devid
286	local new clone_mac no_derive num quad i=0
287	for iface in $*; do
288
289		clone_mac=
290		no_derive=
291		case "$iface" in
292		=*) iface=${iface#=} clone_mac=1 ;;
293		!*) iface=${iface#!} no_derive=1 ;;
294		esac
295
296		# Make sure the interface doesn't exist already
297		eiface=ng${i}_$name
298		if ngctl msg "$eiface:" getifname > /dev/null 2>&1; then
299			i=$(( $i + 1 ))
300			continue
301		fi
302
303		# Bring the interface up
304		ifconfig $iface up || return
305
306		# Set promiscuous mode and don't overwrite src addr
307		ngctl msg $iface: setpromisc 1 || return
308		ngctl msg $iface: setautosrc 0 || return
309
310		# Make sure the interface has been bridged
311		if ! ngctl info ${iface}bridge: > /dev/null 2>&1; then
312			ngctl mkpeer $iface: bridge lower link0 || return
313			ngctl connect $iface: $iface:lower upper link1 ||
314				return
315			ngctl name $iface:lower ${iface}bridge || return
316		fi
317
318		# Optionally create a secondary bridge
319		if [ "$bridge" != "bridge" ] &&
320		   ! ngctl info "$iface$bridge:" > /dev/null 2>&1
321		then
322			num=2
323			while ngctl msg ${iface}bridge: getstats $num \
324				> /dev/null 2>&1
325			do
326				num=$(( $num + 1 ))
327			done
328			ngctl mkpeer $iface:lower bridge link$num link1 ||
329				return
330			ngctl name ${iface}bridge:link$num "$iface$bridge" ||
331				return
332		fi
333
334		# Create a new interface to the bridge
335		num=2
336		while ngctl msg "$iface$bridge:" getstats $num > /dev/null 2>&1
337		do
338			num=$(( $num + 1 ))
339		done
340		ngctl mkpeer "$iface$bridge:" eiface link$num ether || return
341
342		# Rename the new interface
343		while [ ${#eiface} -gt 15 ]; do # OS limitation
344			eiface=${eiface%?}
345		done
346		new=$( set -- `ngctl show -n "$iface$bridge:link$num"` &&
347			echo $2 ) || return
348		ngctl name "$iface$bridge:link$num" $eiface || return
349		ifconfig $new name $eiface || return
350		ifconfig $eiface up || return
351
352		#
353		# Set the MAC address of the new interface using a sensible
354		# algorithm to prevent conflicts on the network.
355		#
356		eiface_devid=
357		if [ "$clone_mac" ]; then
358			eiface_devid=$( ifconfig $iface ether |
359				awk '/ether/,$0=$2' )
360		elif [ ! "$no_derive" ]; then
361			derive_mac $iface "$name" eiface_devid
362		fi
363		[ "$eiface_devid" ] &&
364			ifconfig $eiface ether $eiface_devid > /dev/null 2>&1
365
366		i=$(( $i + 1 ))
367	done # for iface
368}
369
370jng_graph_usage="graph [-f] [-T type] [-o output]"
371jng_graph_descr="Generate network graph (default output is \`jng.svg')"
372jng_graph()
373{
374	local OPTIND=1 OPTARG flag
375	local output=jng.svg output_type= force=
376	while getopts fo:T: flag; do
377		case "$flag" in
378		f) force=1 ;;
379		o) output="$OPTARG" ;;
380		T) output_type="$OPTARG" ;;
381		*) action_usage graph # NOTREACHED
382		esac
383	done
384	shift $(( $OPTIND - 1 ))
385	[ $# -eq 0 -a "$output" ] || action_usage graph # NOTREACHED
386	mustberoot_to_continue
387	if [ -e "$output" -a ! "$force" ]; then
388		echo "$output: Already exists (use \`-f' to overwrite)" >&2
389		return $FAILURE
390	fi
391	if [ ! "$output_type" ]; then
392		local valid suffix
393		valid=$( dot -Txxx 2>&1 )
394		for suffix in ${valid##*:}; do
395			[ "$output" != "${output%.$suffix}" ] || continue
396			output_type=$suffix
397			break
398		done
399	fi
400	ngctl dot | dot ${output_type:+-T "$output_type"} -o "$output"
401}
402
403jng_show_usage="show"
404jng_show_descr="List possible NAME values for \`show NAME'"
405jng_show1_usage="show NAME"
406jng_show1_descr="Lists ng0_NAME [ng1_NAME ...]"
407jng_show2_usage="show [NAME]"
408jng_show()
409{
410	local OPTIND=1 OPTARG flag
411	while getopts "" flag; do
412		case "$flag" in
413		*) action_usage show2 # NOTREACHED
414		esac
415	done
416	shift $(( $OPTIND - 1 ))
417	mustberoot_to_continue
418	if [ $# -eq 0 ]; then
419		ngctl ls | awk '$4=="bridge",$0=$2' |
420			xargs -rn1 -Ibridge ngctl show bridge: |
421			awk 'sub(/^ng[[:digit:]]+_/, "", $2), $0 = $2' |
422			sort -u
423		return
424	fi
425	ngctl ls | awk -v name="$1" '
426		match($2, /^ng[[:digit:]]+_/) &&
427			substr($2, RSTART + RLENGTH) == name &&
428			$4 == "eiface", $0 = $2
429	' | sort
430}
431
432jng_shutdown_usage="shutdown NAME"
433jng_shutdown_descr="Shutdown ng0_NAME [ng1_NAME ...]"
434jng_shutdown()
435{
436	local OPTIND=1 OPTARG flag
437	while getopts "" flag; do
438		case "$flag" in
439		*) action_usage shutdown # NOTREACHED
440		esac
441	done
442	shift $(( $OPTIND -1 ))
443	local name="$1"
444	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -eq 1 ] ||
445		action_usage shutdown # NOTREACHED
446	mustberoot_to_continue
447	jng_show "$name" | xargs -rn1 -I eiface ngctl shutdown eiface:
448}
449
450jng_stats_usage="stats NAME"
451jng_stats_descr="Show ng_bridge link statistics for NAME interfaces"
452jng_stats()
453{
454	local OPTIND=1 OPTARG flag
455	while getopts "" flag; do
456		case "$flag" in
457		*) action_usage stats # NOTREACHED
458		esac
459	done
460	shift $(( $OPTIND -1 ))
461	local name="$1"
462	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -eq 1 ] ||
463		action_usage stats # NOTREACHED
464	mustberoot_to_continue
465	for eiface in $( jng_show "$name" ); do
466		echo "$eiface:"
467		ngctl show $eiface: | awk '
468		$3 == "bridge" && $5 ~ /^link/ {
469			bridge = $2
470			link = substr($5, 5)
471			system(sprintf("ngctl msg %s: getstats %u",
472				bridge, link))
473		}' | fmt 2 | awk '
474			/=/ && fl = index($0, "=") {
475				printf "%20s = %s\n",
476					substr($0, 0, fl-1),
477					substr($0, 0, fl+1)
478			}
479		' # END-QUOTE
480	done
481}
482
483############################################################ MAIN
484
485#
486# Command-line arguments
487#
488action="$1"
489[ "$action" ] || usage # NOTREACHED
490
491#
492# Validate action argument
493#
494if [ "$BASH_VERSION" ]; then
495	type="$( type -t "jng_$action" )" || usage # NOTREACHED
496else
497	type="$( type "jng_$action" 2> /dev/null )" || usage # NOTREACHED
498fi
499case "$type" in
500*function)
501	shift 1 # action
502	eval "jng_$action" \"\$@\"
503	;;
504*) usage # NOTREACHED
505esac
506
507################################################################################
508# END
509################################################################################
510