xref: /freebsd/share/examples/jails/jng (revision de121c54781b92d4287277eefa8fd3eb783c8290)
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#
28############################################################ IDENT(1)
29#
30# $Title: netgraph(4) management script for vnet jails $
31#
32############################################################ INFORMATION
33#
34# Use this tool with jail.conf(5) (or rc.conf(5) ``legacy'' configuration) to
35# manage `vnet' interfaces for jails. Designed to automate the creation of vnet
36# interface(s) during jail `prestart' and destroy said interface(s) during jail
37# `poststop'.
38#
39# In jail.conf(5) format:
40#
41# ### BEGIN EXCERPT ###
42#
43# xxx {
44# 	host.hostname = "xxx.yyy";
45# 	path = "/vm/xxx";
46#
47# 	#
48# 	# NB: Below 2-lines required
49# 	# NB: The number of ngN_xxx interfaces should match the number of
50# 	#     arguments given to `jng bridge xxx' in exec.prestart value.
51# 	#
52# 	vnet;
53# 	vnet.interface = ng0_xxx, ng1_xxx, ...;
54#
55# 	exec.clean;
56# 	exec.system_user = "root";
57# 	exec.jail_user = "root";
58#
59# 	#
60# 	# NB: Below 2-lines required
61# 	# NB: The number of arguments after `jng bridge xxx' should match
62# 	#     the number of ngN_xxx arguments in vnet.interface value.
63# 	#
64# 	exec.prestart += "jng bridge xxx em0 em1 ...";
65# 	exec.poststop += "jng shutdown xxx";
66#
67# 	# Standard recipe
68# 	exec.start += "/bin/sh /etc/rc";
69# 	exec.stop = "/bin/sh /etc/rc.shutdown jail";
70# 	exec.consolelog = "/var/log/jail_xxx_console.log";
71# 	mount.devfs;
72#
73# 	# Optional (default off)
74# 	#allow.mount;
75# 	#allow.set_hostname = 1;
76# 	#allow.sysvipc = 1;
77# 	#devfs_ruleset = "11"; # rule to unhide bpf for DHCP
78# }
79#
80# ### END EXCERPT ###
81#
82# In rc.conf(5) ``legacy'' format (used when /etc/jail.conf does not exist):
83#
84# ### BEGIN EXCERPT ###
85#
86# jail_enable="YES"
87# jail_list="xxx"
88#
89# #
90# # Global presets for all jails
91# #
92# jail_devfs_enable="YES"	# mount devfs
93#
94# #
95# # Global options (default off)
96# #
97# #jail_mount_enable="YES"		# mount /etc/fstab.{name}
98# #jail_set_hostname_allow="YES"	# Allow hostname to change
99# #jail_sysvipc_allow="YES"		# Allow SysV Interprocess Comm.
100#
101# # xxx
102# jail_xxx_hostname="xxx.shxd.cx"		# hostname
103# jail_xxx_rootdir="/vm/xxx"			# root directory
104# jail_xxx_vnet_interfaces="ng0_xxx ng1xxx ..."	# vnet interface(s)
105# jail_xxx_exec_prestart0="jng bridge xxx em0 em1 ..."	# bridge interface(s)
106# jail_xxx_exec_poststop0="jng shutdown xxx"	# destroy interface(s)
107# #jail_xxx_mount_enable="YES"			# mount /etc/fstab.xxx
108# #jail_xxx_devfs_ruleset="11"			# rule to unhide bpf for DHCP
109#
110# ### END EXCERPT ###
111#
112# Note that the legacy rc.conf(5) format is converted to
113# /var/run/jail.{name}.conf by /etc/rc.d/jail if jail.conf(5) is missing.
114#
115# ASIDE: dhclient(8) inside a vnet jail...
116#
117# To allow dhclient(8) to work inside a vnet jail, make sure the following
118# appears in /etc/devfs.rules (which should be created if it doesn't exist):
119#
120# 	[devfsrules_jail=11]
121# 	add include $devfsrules_hide_all
122# 	add include $devfsrules_unhide_basic
123# 	add include $devfsrules_unhide_login
124# 	add path 'bpf*' unhide
125#
126# And set ether devfs.ruleset="11" (jail.conf(5)) or
127# jail_{name}_devfs_ruleset="11" (rc.conf(5)).
128#
129# NB: While this tool can't create every type of desirable topology, it should
130# handle most setups, minus some which considered exotic or purpose-built.
131#
132############################################################ GLOBALS
133
134pgm="${0##*/}" # Program basename
135
136#
137# Global exit status
138#
139SUCCESS=0
140FAILURE=1
141
142############################################################ FUNCTIONS
143
144usage()
145{
146	local action usage descr
147	exec >&2
148	echo "Usage: $pgm action [arguments]"
149	echo "Actions:"
150	for action in \
151		bridge		\
152		graph		\
153		show		\
154		show1		\
155		shutdown	\
156		stats		\
157	; do
158		eval usage=\"\$jng_${action}_usage\"
159		[ "$usage" ] || continue
160		eval descr=\"\$jng_${action}_descr\"
161		printf "\t%s\n\t\t%s\n" "$usage" "$descr"
162	done
163	exit $FAILURE
164}
165
166action_usage()
167{
168	local usage descr action="$1"
169	eval usage=\"\$jng_${action}_usage\"
170	echo "Usage: $pgm $usage" >&2
171	eval descr=\"\$jng_${action}_descr\"
172	printf "\t%s\n" "$descr"
173	exit $FAILURE
174}
175
176derive_mac()
177{
178	local OPTIND=1 OPTARG __flag
179	local __mac_num= __make_pair=
180	while getopts 2n: __flag; do
181		case "$__flag" in
182		2) __make_pair=1 ;;
183		n) __mac_num=${OPTARG%%[^0-9]*} ;;
184		esac
185	done
186	shift $(( $OPTIND - 1 ))
187
188	if [ ! "$__mac_num" ]; then
189		eval __mac_num=\${_${iface}_num:--1}
190		__mac_num=$(( $__mac_num + 1 ))
191		eval _${iface}_num=\$__mac_num
192	fi
193
194	local __iface="$1" __name="$2" __var_to_set="$3" __var_to_set_b="$4"
195	local __iface_devid __new_devid __num __new_devid_b
196	#
197	# Calculate MAC address derived from given iface.
198	#
199	# The formula I'm using is ``NP:SS:SS:II:II:II'' where:
200	# + N denotes 4 bits used as a counter to support branching
201	#   each parent interface up to 15 times under the same jail
202	#   name (see S below).
203	# + P denotes the special nibble whose value, if one of
204	#   2, 6, A, or E (but usually 2) denotes a privately
205	#   administered MAC address (while remaining routable).
206	# + S denotes 16 bits, the sum(1) value of the jail name.
207	# + I denotes bits that are inherited from parent interface.
208	#
209	# The S bits are a CRC-16 checksum of NAME, allowing the jail
210	# to change link numbers in ng_bridge(4) without affecting the
211	# MAC address. Meanwhile, if...
212	#   + the jail NAME changes (e.g., it was duplicated and given
213	#     a new name with no other changes)
214	#   + the underlying network interface changes
215	#   + the jail is moved to another host
216	# the MAC address will be recalculated to a new, similarly
217	# unique value preventing conflict.
218	#
219	__iface_devid=$( ifconfig $__iface ether | awk '/ether/,$0=$2' )
220	# ??:??:??:II:II:II
221	__new_devid=${__iface_devid#??:??:??} # => :II:II:II
222	# => :SS:SS:II:II:II
223	__num=$( set -- `echo -n "$__name" | sum` && echo $1 )
224	__new_devid=$( printf :%02x:%02x \
225		$(( $__num >> 8 & 255 )) $(( $__num & 255 )) )$__new_devid
226	# => P:SS:SS:II:II:II
227	case "$__iface_devid" in
228	   ?2:*) __new_devid=a$__new_devid __new_devid_b=e$__new_devid ;;
229	?[Ee]:*) __new_devid=2$__new_devid __new_devid_b=6$__new_devid ;;
230	      *) __new_devid=2$__new_devid __new_devid_b=e$__new_devid
231	esac
232	# => NP:SS:SS:II:II:II
233	__new_devid=$( printf %x $(( $__mac_num & 15 )) )$__new_devid
234	__new_devid_b=$( printf %x $(( $__mac_num & 15 )) )$__new_devid_b
235
236	#
237	# Return derivative MAC address(es)
238	#
239	if [ "$__make_pair" ]; then
240		if [ "$__var_to_set" -a "$__var_to_set_b" ]; then
241			eval $__var_to_set=\$__new_devid
242			eval $__var_to_set_b=\$__new_devid_b
243		else
244			echo $__new_devid $__new_devid_b
245		fi
246	else
247		if [ "$__var_to_set" ]; then
248			eval $__var_to_set=\$__new_devid
249		else
250			echo $__new_devid
251		fi
252	fi
253}
254
255mustberoot_to_continue()
256{
257	if [ "$( id -u )" -ne 0 ]; then
258		echo "Must run as root!" >&2
259		exit $FAILURE
260	fi
261}
262
263jng_bridge_usage="bridge [-b BRIDGE_NAME] NAME [!|=]iface0 [[!|=]iface1 ...]"
264jng_bridge_descr="Create ng0_NAME [ng1_NAME ...]"
265jng_bridge()
266{
267	local OPTIND=1 OPTARG flag bridge=bridge
268	while getopts b: flag; do
269		case "$flag" in
270		b) bridge="$OPTARG"
271		   [ "$bridge" ] || action_usage bridge ;; # NOTREACHED
272		*) action_usage bridge # NOTREACHED
273		esac
274	done
275	shift $(( $OPTIND - 1 ))
276
277	local name="$1"
278	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -gt 1 ] ||
279		action_usage bridge # NOTREACHED
280	shift 1 # name
281
282	mustberoot_to_continue
283
284	local iface parent eiface eiface_devid
285	local new clone_mac no_derive num quad i=0
286	for iface in $*; do
287
288		clone_mac=
289		no_derive=
290		case "$iface" in
291		=*) iface=${iface#=} clone_mac=1 ;;
292		!*) iface=${iface#!} no_derive=1 ;;
293		esac
294
295		# Make sure the interface doesn't exist already
296		eiface=ng${i}_$name
297		if ngctl msg "$eiface:" getifname > /dev/null 2>&1; then
298			i=$(( $i + 1 ))
299			continue
300		fi
301
302		# Bring the interface up
303		ifconfig $iface up || return
304
305		# Set promiscuous mode and don't overwrite src addr
306		ngctl msg $iface: setpromisc 1 || return
307		ngctl msg $iface: setautosrc 0 || return
308
309		# Make sure the interface has been bridged
310		if ! ngctl info ${iface}bridge: > /dev/null 2>&1; then
311			ngctl mkpeer $iface: bridge lower link0 || return
312			ngctl connect $iface: $iface:lower upper link1 ||
313				return
314			ngctl name $iface:lower ${iface}bridge || return
315		fi
316
317		mtu=$(ifconfig ${iface} | sed -n '1s/^.*mtu //p;') || return
318
319		# Optionally create a secondary bridge
320		if [ "$bridge" != "bridge" ] &&
321		   ! ngctl info "$iface$bridge:" > /dev/null 2>&1
322		then
323			num=2
324			while ngctl msg ${iface}bridge: getstats $num \
325				> /dev/null 2>&1
326			do
327				num=$(( $num + 1 ))
328			done
329			ngctl mkpeer $iface:lower bridge link$num link1 ||
330				return
331			ngctl name ${iface}bridge:link$num "$iface$bridge" ||
332				return
333		fi
334
335		# Create a new interface to the bridge
336		num=2
337		while ngctl msg "$iface$bridge:" getstats $num > /dev/null 2>&1
338		do
339			num=$(( $num + 1 ))
340		done
341		ngctl mkpeer "$iface$bridge:" eiface link$num ether || return
342
343		# Rename the new interface
344		while [ ${#eiface} -gt 15 ]; do # OS limitation
345			eiface=${eiface%?}
346		done
347		new=$( set -- `ngctl show -n "$iface$bridge:link$num"` &&
348			echo $2 ) || return
349		ngctl name "$iface$bridge:link$num" $eiface || return
350		ifconfig $new name $eiface || return
351		ifconfig $eiface mtu $mtu || return
352		ifconfig $eiface up || return
353
354		#
355		# Set the MAC address of the new interface using a sensible
356		# algorithm to prevent conflicts on the network.
357		#
358		eiface_devid=
359		if [ "$clone_mac" ]; then
360			eiface_devid=$( ifconfig $iface ether |
361				awk '/ether/,$0=$2' )
362		elif [ ! "$no_derive" ]; then
363			derive_mac $iface "$name" eiface_devid
364		fi
365		[ "$eiface_devid" ] &&
366			ifconfig $eiface ether $eiface_devid > /dev/null 2>&1
367
368		i=$(( $i + 1 ))
369	done # for iface
370}
371
372jng_graph_usage="graph [-f] [-T type] [-o output]"
373jng_graph_descr="Generate network graph (default output is \`jng.svg')"
374jng_graph()
375{
376	local OPTIND=1 OPTARG flag
377	local output=jng.svg output_type= force=
378	while getopts fo:T: flag; do
379		case "$flag" in
380		f) force=1 ;;
381		o) output="$OPTARG" ;;
382		T) output_type="$OPTARG" ;;
383		*) action_usage graph # NOTREACHED
384		esac
385	done
386	shift $(( $OPTIND - 1 ))
387	[ $# -eq 0 -a "$output" ] || action_usage graph # NOTREACHED
388	mustberoot_to_continue
389	if [ -e "$output" -a ! "$force" ]; then
390		echo "$output: Already exists (use \`-f' to overwrite)" >&2
391		return $FAILURE
392	fi
393	if [ ! "$output_type" ]; then
394		local valid suffix
395		valid=$( dot -Txxx 2>&1 )
396		for suffix in ${valid##*:}; do
397			[ "$output" != "${output%.$suffix}" ] || continue
398			output_type=$suffix
399			break
400		done
401	fi
402	ngctl dot | dot ${output_type:+-T "$output_type"} -o "$output"
403}
404
405jng_show_usage="show"
406jng_show_descr="List possible NAME values for \`show NAME'"
407jng_show1_usage="show NAME"
408jng_show1_descr="Lists ng0_NAME [ng1_NAME ...]"
409jng_show2_usage="show [NAME]"
410jng_show()
411{
412	local OPTIND=1 OPTARG flag
413	while getopts "" flag; do
414		case "$flag" in
415		*) action_usage show2 # NOTREACHED
416		esac
417	done
418	shift $(( $OPTIND - 1 ))
419	mustberoot_to_continue
420	if [ $# -eq 0 ]; then
421		ngctl ls | awk '$4=="bridge",$0=$2' |
422			xargs -rn1 -Ibridge ngctl show bridge: |
423			awk 'sub(/^ng[[:digit:]]+_/, "", $2), $0 = $2' |
424			sort -u
425		return
426	fi
427	ngctl ls | awk -v name="$1" '
428		match($2, /^ng[[:digit:]]+_/) &&
429			substr($2, RSTART + RLENGTH) == name &&
430			$4 == "eiface", $0 = $2
431	' | sort
432}
433
434jng_shutdown_usage="shutdown NAME"
435jng_shutdown_descr="Shutdown ng0_NAME [ng1_NAME ...]"
436jng_shutdown()
437{
438	local OPTIND=1 OPTARG flag
439	while getopts "" flag; do
440		case "$flag" in
441		*) action_usage shutdown # NOTREACHED
442		esac
443	done
444	shift $(( $OPTIND -1 ))
445	local name="$1"
446	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -eq 1 ] ||
447		action_usage shutdown # NOTREACHED
448	mustberoot_to_continue
449	jng_show "$name" | xargs -rn1 -I eiface ngctl shutdown eiface:
450}
451
452jng_stats_usage="stats NAME"
453jng_stats_descr="Show ng_bridge link statistics for NAME interfaces"
454jng_stats()
455{
456	local OPTIND=1 OPTARG flag
457	while getopts "" flag; do
458		case "$flag" in
459		*) action_usage stats # NOTREACHED
460		esac
461	done
462	shift $(( $OPTIND -1 ))
463	local name="$1"
464	[ "${name:-x}" = "${name#*[!0-9a-zA-Z_]}" -a $# -eq 1 ] ||
465		action_usage stats # NOTREACHED
466	mustberoot_to_continue
467	for eiface in $( jng_show "$name" ); do
468		echo "$eiface:"
469		ngctl show $eiface: | awk '
470		$3 == "bridge" && $5 ~ /^link/ {
471			bridge = $2
472			link = substr($5, 5)
473			system(sprintf("ngctl msg %s: getstats %u",
474				bridge, link))
475		}' | fmt 2 | awk '
476			/=/ && fl = index($0, "=") {
477				printf "%20s = %s\n",
478					substr($0, 0, fl-1),
479					substr($0, 0, fl+1)
480			}
481		' # END-QUOTE
482	done
483}
484
485############################################################ MAIN
486
487#
488# Command-line arguments
489#
490action="$1"
491[ "$action" ] || usage # NOTREACHED
492
493#
494# Validate action argument
495#
496if [ "$BASH_VERSION" ]; then
497	type="$( type -t "jng_$action" )" || usage # NOTREACHED
498else
499	type="$( type "jng_$action" 2> /dev/null )" || usage # NOTREACHED
500fi
501case "$type" in
502*function)
503	shift 1 # action
504	eval "jng_$action" \"\$@\"
505	;;
506*) usage # NOTREACHED
507esac
508
509################################################################################
510# END
511################################################################################
512