1TCPDUMP(8) System Manager's Manual TCPDUMP(8)
2
6 tcpdump - dump traffic on a network
7
9 tcpdump [ -AbdDefIKlLnNOpqRStuUvxX ] [ -B buffer_size ] [ -c count ]
10 [ -C file_size ] [ -G rotate_seconds ] [ -F file ]
11 [ -i interface ] [ -m module ] [ -M secret ]
12 [ -r file ] [ -s snaplen ] [ -T type ] [ -w file ]
13 [ -W filecount ]
14 [ -E spi@ipaddr algo:secret,... ]
15 [ -y datalinktype ] [ -z postrotate-command ] [ -Z user ]
16 [ expression ]
17
19 Tcpdump prints out a description of the contents of packets on a net‐
20 work interface that match the boolean expression. It can also be run
21 with the -w flag, which causes it to save the packet data to a file for
22 later analysis, and/or with the -r flag, which causes it to read from a
23 saved packet file rather than to read packets from a network interface.
24 In all cases, only packets that match expression will be processed by
25 tcpdump.
26 Tcpdump will, if not run with the -c flag, continue capturing packets
28 until it is interrupted by a SIGINT signal (generated, for example, by
29 typing your interrupt character, typically control-C) or a SIGTERM sig‐
30 nal (typically generated with the kill(1) command); if run with the -c
31 flag, it will capture packets until it is interrupted by a SIGINT or
32 SIGTERM signal or the specified number of packets have been processed.
33 When tcpdump finishes capturing packets, it will report counts of:
35 packets ``captured'' (this is the number of packets that tcpdump
37 has received and processed);
38 packets ``received by filter'' (the meaning of this depends on
40 the OS on which you're running tcpdump, and possibly on the way
41 the OS was configured - if a filter was specified on the command
42 line, on some OSes it counts packets regardless of whether they
43 were matched by the filter expression and, even if they were
44 matched by the filter expression, regardless of whether tcpdump
45 has read and processed them yet, on other OSes it counts only
46 packets that were matched by the filter expression regardless of
47 whether tcpdump has read and processed them yet, and on other
48 OSes it counts only packets that were matched by the filter
49 expression and were processed by tcpdump);
50 packets ``dropped by kernel'' (this is the number of packets
52 that were dropped, due to a lack of buffer space, by the packet
53 capture mechanism in the OS on which tcpdump is running, if the
54 OS reports that information to applications; if not, it will be
55 reported as 0).
56 On platforms that support the SIGINFO signal, such as most BSDs
58 (including Mac OS X) and Digital/Tru64 UNIX, it will report those
59 counts when it receives a SIGINFO signal (generated, for example, by
60 typing your ``status'' character, typically control-T, although on some
61 platforms, such as Mac OS X, the ``status'' character is not set by
62 default, so you must set it with stty(1) in order to use it) and will
63 continue capturing packets.
64 Reading packets from a network interface may require that you have spe‐
66 cial privileges; see the pcap (3PCAP) man page for details. Reading a
67 saved packet file doesn't require special privileges.
68
70 -A Print each packet (minus its link level header) in ASCII. Handy
71 for capturing web pages.
72 -b Print the AS number in BGP packets in ASDOT notation rather than
74 ASPLAIN notation.
75 -B Set the operating system capture buffer size to buffer_size.
77 -c Exit after receiving count packets.
79 -C Before writing a raw packet to a savefile, check whether the
81 file is currently larger than file_size and, if so, close the
82 current savefile and open a new one. Savefiles after the first
83 savefile will have the name specified with the -w flag, with a
84 number after it, starting at 1 and continuing upward. The units
85 of file_size are millions of bytes (1,000,000 bytes, not
86 1,048,576 bytes).
87 Note that when used with -Z option (enabled by default), privi‐
89 leges are dropped before opening first savefile.
90 -d Dump the compiled packet-matching code in a human readable form
92 to standard output and stop.
93 -dd Dump packet-matching code as a C program fragment.
95 -ddd Dump packet-matching code as decimal numbers (preceded with a
97 count).
98 -D Print the list of the network interfaces available on the system
100 and on which tcpdump can capture packets. For each network
101 interface, a number and an interface name, possibly followed by
102 a text description of the interface, is printed. The interface
103 name or the number can be supplied to the -i flag to specify an
104 interface on which to capture.
105 This can be useful on systems that don't have a command to list
107 them (e.g., Windows systems, or UNIX systems lacking ifconfig
108 -a); the number can be useful on Windows 2000 and later systems,
109 where the interface name is a somewhat complex string.
110 The -D flag will not be supported if tcpdump was built with an
112 older version of libpcap that lacks the pcap_findalldevs() func‐
113 tion.
114 -e Print the link-level header on each dump line.
116 -E Use spi@ipaddr algo:secret for decrypting IPsec ESP packets that
118 are addressed to addr and contain Security Parameter Index value
119 spi. This combination may be repeated with comma or newline
120 seperation.
121 Note that setting the secret for IPv4 ESP packets is supported
123 at this time.
124 Algorithms may be des-cbc, 3des-cbc, blowfish-cbc, rc3-cbc,
126 cast128-cbc, or none. The default is des-cbc. The ability to
127 decrypt packets is only present if tcpdump was compiled with
128 cryptography enabled.
129 secret is the ASCII text for ESP secret key. If preceeded by
131 0x, then a hex value will be read.
132 The option assumes RFC2406 ESP, not RFC1827 ESP. The option is
134 only for debugging purposes, and the use of this option with a
135 true `secret' key is discouraged. By presenting IPsec secret
136 key onto command line you make it visible to others, via ps(1)
137 and other occasions.
138 In addition to the above syntax, the syntax file name may be
140 used to have tcpdump read the provided file in. The file is
141 opened upon receiving the first ESP packet, so any special per‐
142 missions that tcpdump may have been given should already have
143 been given up.
144 -f Print `foreign' IPv4 addresses numerically rather than symboli‐
146 cally (this option is intended to get around serious brain dam‐
147 age in Sun's NIS server — usually it hangs forever translating
148 non-local internet numbers).
149 The test for `foreign' IPv4 addresses is done using the IPv4
151 address and netmask of the interface on which capture is being
152 done. If that address or netmask are not available, available,
153 either because the interface on which capture is being done has
154 no address or netmask or because the capture is being done on
155 the Linux "any" interface, which can capture on more than one
156 interface, this option will not work correctly.
157 -F Use file as input for the filter expression. An additional
159 expression given on the command line is ignored.
160 -G If specified, rotates the dump file specified with the -w option
162 every rotate_seconds seconds. Savefiles will have the name
163 specified by -w which should include a time format as defined by
164 strftime(3). If no time format is specified, each new file will
165 overwrite the previous.
166 If used in conjunction with the -C option, filenames will take
168 the form of `file<count>'.
169 -i Listen on interface. If unspecified, tcpdump searches the sys‐
171 tem interface list for the lowest numbered, configured up inter‐
172 face (excluding loopback). Ties are broken by choosing the ear‐
173 liest match.
174 On Linux systems with 2.2 or later kernels, an interface argu‐
176 ment of ``any'' can be used to capture packets from all inter‐
177 faces. Note that captures on the ``any'' device will not be
178 done in promiscuous mode.
179 If the -D flag is supported, an interface number as printed by
181 that flag can be used as the interface argument.
182 -I Put the interface in "monitor mode"; this is supported only on
184 IEEE 802.11 Wi-Fi interfaces, and supported only on some operat‐
185 ing systems.
186 Note that in monitor mode the adapter might disassociate from
188 the network with which it's associated, so that you will not be
189 able to use any wireless networks with that adapter. This could
190 prevent accessing files on a network server, or resolving host
191 names or network addresses, if you are capturing in monitor mode
192 and are not connected to another network with another adapter.
193 This flag will affect the output of the -L flag. If -I isn't
195 specified, only those link-layer types available when not in
196 monitor mode will be shown; if -I is specified, only those link-
197 layer types available when in monitor mode will be shown.
198 -K Don't attempt to verify IP, TCP, or UDP checksums. This is use‐
200 ful for interfaces that perform some or all of those checksum
201 calculation in hardware; otherwise, all outgoing TCP checksums
202 will be flagged as bad.
203 -l Make stdout line buffered. Useful if you want to see the data
205 while capturing it. E.g.,
206 ``tcpdump -l | tee dat'' or ``tcpdump -l >
207 dat & tail -f dat''.
208 -L List the known data link types for the interface, in the speci‐
210 fied mode, and exit. The list of known data link types may be
211 dependent on the specified mode; for example, on some platforms,
212 a Wi-Fi interface might support one set of data link types when
213 not in monitor mode (for example, it might support only fake
214 Ethernet headers, or might support 802.11 headers but not sup‐
215 port 802.11 headers with radio information) and another set of
216 data link types when in monitor mode (for example, it might sup‐
217 port 802.11 headers, or 802.11 headers with radio information,
218 only in monitor mode).
219 -m Load SMI MIB module definitions from file module. This option
221 can be used several times to load several MIB modules into tcp‐
222 dump.
223 -M Use secret as a shared secret for validating the digests found
225 in TCP segments with the TCP-MD5 option (RFC 2385), if present.
226 -n Don't convert host addresses to names. This can be used to
228 avoid DNS lookups.
229 -nn Don't convert protocol and port numbers etc. to names either.
231 -N Don't print domain name qualification of host names. E.g., if
233 you give this flag then tcpdump will print ``nic'' instead of
234 ``nic.ddn.mil''.
235 -O Do not run the packet-matching code optimizer. This is useful
237 only if you suspect a bug in the optimizer.
238 -p Don't put the interface into promiscuous mode. Note that the
240 interface might be in promiscuous mode for some other reason;
241 hence, `-p' cannot be used as an abbreviation for `ether host
242 {local-hw-addr} or ether broadcast'.
243 -q Quick (quiet?) output. Print less protocol information so out‐
245 put lines are shorter.
246 -R Assume ESP/AH packets to be based on old specification (RFC1825
248 to RFC1829). If specified, tcpdump will not print replay pre‐
249 vention field. Since there is no protocol version field in
250 ESP/AH specification, tcpdump cannot deduce the version of
251 ESP/AH protocol.
252 -r Read packets from file (which was created with the -w option).
254 Standard input is used if file is ``-''.
255 -S Print absolute, rather than relative, TCP sequence numbers.
257 -s Snarf snaplen bytes of data from each packet rather than the
259 default of 65535 bytes. Packets truncated because of a limited
260 snapshot are indicated in the output with ``[|proto]'', where
261 proto is the name of the protocol level at which the truncation
262 has occurred. Note that taking larger snapshots both increases
263 the amount of time it takes to process packets and, effectively,
264 decreases the amount of packet buffering. This may cause pack‐
265 ets to be lost. You should limit snaplen to the smallest number
266 that will capture the protocol information you're interested in.
267 Setting snaplen to 0 sets it to the default of 65535, for back‐
268 wards compatibility with recent older versions of tcpdump.
269 -T Force packets selected by "expression" to be interpreted the
271 specified type. Currently known types are aodv (Ad-hoc On-
272 demand Distance Vector protocol), cnfp (Cisco NetFlow protocol),
273 rpc (Remote Procedure Call), rtp (Real-Time Applications proto‐
274 col), rtcp (Real-Time Applications control protocol), snmp (Sim‐
275 ple Network Management Protocol), tftp (Trivial File Transfer
276 Protocol), vat (Visual Audio Tool), and wb (distributed White
277 Board).
278 -t Don't print a timestamp on each dump line.
280 -tt Print an unformatted timestamp on each dump line.
282 -ttt Print a delta (micro-second resolution) between current and pre‐
284 vious line on each dump line.
285 -tttt Print a timestamp in default format proceeded by date on each
287 dump line.
288 -ttttt Print a delta (micro-second resolution) between current and
290 first line on each dump line.
291 -u Print undecoded NFS handles.
293 -U Make output saved via the -w option ``packet-buffered''; i.e.,
295 as each packet is saved, it will be written to the output file,
296 rather than being written only when the output buffer fills.
297 The -U flag will not be supported if tcpdump was built with an
299 older version of libpcap that lacks the pcap_dump_flush() func‐
300 tion.
301 -v When parsing and printing, produce (slightly more) verbose out‐
303 put. For example, the time to live, identification, total
304 length and options in an IP packet are printed. Also enables
305 additional packet integrity checks such as verifying the IP and
306 ICMP header checksum.
307 When writing to a file with the -w option, report, every 10 sec‐
309 onds, the number of packets captured.
310 -vv Even more verbose output. For example, additional fields are
312 printed from NFS reply packets, and SMB packets are fully
313 decoded.
314 -vvv Even more verbose output. For example, telnet SB ... SE options
316 are printed in full. With -X Telnet options are printed in hex
317 as well.
318 -w Write the raw packets to file rather than parsing and printing
320 them out. They can later be printed with the -r option. Stan‐
321 dard output is used if file is ``-''. See pcap-savefile(5) for
322 a description of the file format.
323 -W Used in conjunction with the -C option, this will limit the num‐
325 ber of files created to the specified number, and begin over‐
326 writing files from the beginning, thus creating a 'rotating'
327 buffer. In addition, it will name the files with enough leading
328 0s to support the maximum number of files, allowing them to sort
329 correctly.
330 Used in conjunction with the -G option, this will limit the num‐
332 ber of rotated dump files that get created, exiting with status
333 0 when reaching the limit. If used with -C as well, the behavior
334 will result in cyclical files per timeslice.
335 -x When parsing and printing, in addition to printing the headers
337 of each packet, print the data of each packet (minus its link
338 level header) in hex. The smaller of the entire packet or
339 snaplen bytes will be printed. Note that this is the entire
340 link-layer packet, so for link layers that pad (e.g. Ethernet),
341 the padding bytes will also be printed when the higher layer
342 packet is shorter than the required padding.
343 -xx When parsing and printing, in addition to printing the headers
345 of each packet, print the data of each packet, including its
346 link level header, in hex.
347 -X When parsing and printing, in addition to printing the headers
349 of each packet, print the data of each packet (minus its link
350 level header) in hex and ASCII. This is very handy for
351 analysing new protocols.
352 -XX When parsing and printing, in addition to printing the headers
354 of each packet, print the data of each packet, including its
355 link level header, in hex and ASCII.
356 -y Set the data link type to use while capturing packets to
358 datalinktype.
359 -z Used in conjunction with the -C or -G options, this will make
361 tcpdump run " command file " where file is the savefile being
362 closed after each rotation. For example, specifying -z gzip or
363 -z bzip2 will compress each savefile using gzip or bzip2.
364 Note that tcpdump will run the command in parallel to the cap‐
366 ture, using the lowest priority so that this doesn't disturb the
367 capture process.
368 And in case you would like to use a command that itself takes
370 flags or different arguments, you can always write a shell
371 script that will take the savefile name as the only argument,
372 make the flags & arguments arrangements and execute the command
373 that you want.
374 -Z Drops privileges (if root) and changes user ID to user and the
376 group ID to the primary group of user.
377 This behavior is enabled by default (-Z tcpdump), and can be
379 disabled by -Z root.
380 expression
383 selects which packets will be dumped. If no expression is
384 given, all packets on the net will be dumped. Otherwise, only
385 packets for which expression is `true' will be dumped.
386 For the expression syntax, see pcap-filter(7).
388 Expression arguments can be passed to tcpdump as either a single
390 argument or as multiple arguments, whichever is more convenient.
391 Generally, if the expression contains Shell metacharacters, it
392 is easier to pass it as a single, quoted argument. Multiple
393 arguments are concatenated with spaces before being parsed.
394
396 To print all packets arriving at or departing from sundown:
397 tcpdump host sundown
398 To print traffic between helios and either hot or ace:
400 tcpdump host helios and \( hot or ace \)
401 To print all IP packets between ace and any host except helios:
403 tcpdump ip host ace and not helios
404 To print all traffic between local hosts and hosts at Berkeley:
406 tcpdump net ucb-ether
407 To print all ftp traffic through internet gateway snup: (note that the
409 expression is quoted to prevent the shell from (mis-)interpreting the
410 parentheses):
411 tcpdump 'gateway snup and (port ftp or ftp-data)'
412 To print traffic neither sourced from nor destined for local hosts (if
414 you gateway to one other net, this stuff should never make it onto your
415 local net).
416 tcpdump ip and not net localnet
417 To print the start and end packets (the SYN and FIN packets) of each
419 TCP conversation that involves a non-local host.
420 tcpdump 'tcp[tcpflags] & (tcp-syn|tcp-fin) != 0 and not src and dst net localnet'
421 To print all IPv4 HTTP packets to and from port 80, i.e. print only
423 packets that contain data, not, for example, SYN and FIN packets and
424 ACK-only packets. (IPv6 is left as an exercise for the reader.)
425 tcpdump 'tcp port 80 and (((ip[2:2] - ((ip[0]&0xf)<<2)) - ((tcp[12]&0xf0)>>2)) != 0)'
426 To print IP packets longer than 576 bytes sent through gateway snup:
428 tcpdump 'gateway snup and ip[2:2] > 576'
429 To print IP broadcast or multicast packets that were not sent via Eth‐
431 ernet broadcast or multicast:
432 tcpdump 'ether[0] & 1 = 0 and ip[16] >= 224'
433 To print all ICMP packets that are not echo requests/replies (i.e., not
435 ping packets):
436 tcpdump 'icmp[icmptype] != icmp-echo and icmp[icmptype] != icmp-echoreply'
437
439 The output of tcpdump is protocol dependent. The following gives a
440 brief description and examples of most of the formats.
441 Link Level Headers
443 If the '-e' option is given, the link level header is printed out. On
445 Ethernets, the source and destination addresses, protocol, and packet
446 length are printed.
447 On FDDI networks, the '-e' option causes tcpdump to print the `frame
449 control' field, the source and destination addresses, and the packet
450 length. (The `frame control' field governs the interpretation of the
451 rest of the packet. Normal packets (such as those containing IP data‐
452 grams) are `async' packets, with a priority value between 0 and 7; for
453 example, `async4'. Such packets are assumed to contain an 802.2 Logi‐
454 cal Link Control (LLC) packet; the LLC header is printed if it is not
455 an ISO datagram or a so-called SNAP packet.
456 On Token Ring networks, the '-e' option causes tcpdump to print the
458 `access control' and `frame control' fields, the source and destination
459 addresses, and the packet length. As on FDDI networks, packets are
460 assumed to contain an LLC packet. Regardless of whether the '-e'
461 option is specified or not, the source routing information is printed
462 for source-routed packets.
463 On 802.11 networks, the '-e' option causes tcpdump to print the `frame
465 control' fields, all of the addresses in the 802.11 header, and the
466 packet length. As on FDDI networks, packets are assumed to contain an
467 LLC packet.
468 (N.B.: The following description assumes familiarity with the SLIP com‐
470 pression algorithm described in RFC-1144.)
471 On SLIP links, a direction indicator (``I'' for inbound, ``O'' for out‐
473 bound), packet type, and compression information are printed out. The
474 packet type is printed first. The three types are ip, utcp, and ctcp.
475 No further link information is printed for ip packets. For TCP pack‐
476 ets, the connection identifier is printed following the type. If the
477 packet is compressed, its encoded header is printed out. The special
478 cases are printed out as *S+n and *SA+n, where n is the amount by which
479 the sequence number (or sequence number and ack) has changed. If it is
480 not a special case, zero or more changes are printed. A change is
481 indicated by U (urgent pointer), W (window), A (ack), S (sequence num‐
482 ber), and I (packet ID), followed by a delta (+n or -n), or a new value
483 (=n). Finally, the amount of data in the packet and compressed header
484 length are printed.
485 For example, the following line shows an outbound compressed TCP
487 packet, with an implicit connection identifier; the ack has changed by
488 6, the sequence number by 49, and the packet ID by 6; there are 3 bytes
489 of data and 6 bytes of compressed header:
490 O ctcp * A+6 S+49 I+6 3 (6)
491 ARP/RARP Packets
493 Arp/rarp output shows the type of request and its arguments. The for‐
495 mat is intended to be self explanatory. Here is a short sample taken
496 from the start of an `rlogin' from host rtsg to host csam:
497 arp who-has csam tell rtsg
498 arp reply csam is-at CSAM
499 The first line says that rtsg sent an arp packet asking for the Ether‐
500 net address of internet host csam. Csam replies with its Ethernet
501 address (in this example, Ethernet addresses are in caps and internet
502 addresses in lower case).
503 This would look less redundant if we had done tcpdump -n:
505 arp who-has 128.3.254.6 tell 128.3.254.68
506 arp reply 128.3.254.6 is-at 02:07:01:00:01:c4
507 If we had done tcpdump -e, the fact that the first packet is broadcast
509 and the second is point-to-point would be visible:
510 RTSG Broadcast 0806 64: arp who-has csam tell rtsg
511 CSAM RTSG 0806 64: arp reply csam is-at CSAM
512 For the first packet this says the Ethernet source address is RTSG, the
513 destination is the Ethernet broadcast address, the type field contained
514 hex 0806 (type ETHER_ARP) and the total length was 64 bytes.
515 TCP Packets
517 (N.B.:The following description assumes familiarity with the TCP proto‐
519 col described in RFC-793. If you are not familiar with the protocol,
520 neither this description nor tcpdump will be of much use to you.)
521 The general format of a tcp protocol line is:
523 src > dst: flags data-seqno ack window urgent options
524 Src and dst are the source and destination IP addresses and ports.
525 Flags are some combination of S (SYN), F (FIN), P (PUSH), R (RST), W
526 (ECN CWR) or E (ECN-Echo), or a single `.' (no flags). Data-seqno
527 describes the portion of sequence space covered by the data in this
528 packet (see example below). Ack is sequence number of the next data
529 expected the other direction on this connection. Window is the number
530 of bytes of receive buffer space available the other direction on this
531 connection. Urg indicates there is `urgent' data in the packet.
532 Options are tcp options enclosed in angle brackets (e.g., <mss 1024>).
533 Src, dst and flags are always present. The other fields depend on the
535 contents of the packet's tcp protocol header and are output only if
536 appropriate.
537 Here is the opening portion of an rlogin from host rtsg to host csam.
539 rtsg.1023 > csam.login: S 768512:768512(0) win 4096 <mss 1024>
540 csam.login > rtsg.1023: S 947648:947648(0) ack 768513 win 4096 <mss 1024>
541 rtsg.1023 > csam.login: . ack 1 win 4096
542 rtsg.1023 > csam.login: P 1:2(1) ack 1 win 4096
543 csam.login > rtsg.1023: . ack 2 win 4096
544 rtsg.1023 > csam.login: P 2:21(19) ack 1 win 4096
545 csam.login > rtsg.1023: P 1:2(1) ack 21 win 4077
546 csam.login > rtsg.1023: P 2:3(1) ack 21 win 4077 urg 1
547 csam.login > rtsg.1023: P 3:4(1) ack 21 win 4077 urg 1
548 The first line says that tcp port 1023 on rtsg sent a packet to port
549 login on csam. The S indicates that the SYN flag was set. The packet
550 sequence number was 768512 and it contained no data. (The notation is
551 `first:last(nbytes)' which means `sequence numbers first up to but not
552 including last which is nbytes bytes of user data'.) There was no
553 piggy-backed ack, the available receive window was 4096 bytes and there
554 was a max-segment-size option requesting an mss of 1024 bytes.
555 Csam replies with a similar packet except it includes a piggy-backed
557 ack for rtsg's SYN. Rtsg then acks csam's SYN. The `.' means no flags
558 were set. The packet contained no data so there is no data sequence
559 number. Note that the ack sequence number is a small integer (1). The
560 first time tcpdump sees a tcp `conversation', it prints the sequence
561 number from the packet. On subsequent packets of the conversation, the
562 difference between the current packet's sequence number and this ini‐
563 tial sequence number is printed. This means that sequence numbers
564 after the first can be interpreted as relative byte positions in the
565 conversation's data stream (with the first data byte each direction
566 being `1'). `-S' will override this feature, causing the original
567 sequence numbers to be output.
568 On the 6th line, rtsg sends csam 19 bytes of data (bytes 2 through 20
570 in the rtsg → csam side of the conversation). The PUSH flag is set in
571 the packet. On the 7th line, csam says it's received data sent by rtsg
572 up to but not including byte 21. Most of this data is apparently sit‐
573 ting in the socket buffer since csam's receive window has gotten 19
574 bytes smaller. Csam also sends one byte of data to rtsg in this
575 packet. On the 8th and 9th lines, csam sends two bytes of urgent,
576 pushed data to rtsg.
577 If the snapshot was small enough that tcpdump didn't capture the full
579 TCP header, it interprets as much of the header as it can and then
580 reports ``[|tcp]'' to indicate the remainder could not be interpreted.
581 If the header contains a bogus option (one with a length that's either
582 too small or beyond the end of the header), tcpdump reports it as
583 ``[bad opt]'' and does not interpret any further options (since it's
584 impossible to tell where they start). If the header length indicates
585 options are present but the IP datagram length is not long enough for
586 the options to actually be there, tcpdump reports it as ``[bad hdr
587 length]''.
588 Capturing TCP packets with particular flag combinations (SYN-ACK, URG-
590 ACK, etc.)
591 There are 8 bits in the control bits section of the TCP header:
593 CWR | ECE | URG | ACK | PSH | RST | SYN | FIN
595 Let's assume that we want to watch packets used in establishing a TCP
597 connection. Recall that TCP uses a 3-way handshake protocol when it
598 initializes a new connection; the connection sequence with regard to
599 the TCP control bits is
600 1) Caller sends SYN
602 2) Recipient responds with SYN, ACK
603 3) Caller sends ACK
604 Now we're interested in capturing packets that have only the SYN bit
606 set (Step 1). Note that we don't want packets from step 2 (SYN-ACK),
607 just a plain initial SYN. What we need is a correct filter expression
608 for tcpdump.
609 Recall the structure of a TCP header without options:
611 0 15 31
613 -----------------------------------------------------------------
614 | source port | destination port |
615 -----------------------------------------------------------------
616 | sequence number |
617 -----------------------------------------------------------------
618 | acknowledgment number |
619 -----------------------------------------------------------------
620 | HL | rsvd |C|E|U|A|P|R|S|F| window size |
621 -----------------------------------------------------------------
622 | TCP checksum | urgent pointer |
623 -----------------------------------------------------------------
624 A TCP header usually holds 20 octets of data, unless options are
626 present. The first line of the graph contains octets 0 - 3, the second
627 line shows octets 4 - 7 etc.
628 Starting to count with 0, the relevant TCP control bits are contained
630 in octet 13:
631 0 7| 15| 23| 31
633 ----------------|---------------|---------------|----------------
634 | HL | rsvd |C|E|U|A|P|R|S|F| window size |
635 ----------------|---------------|---------------|----------------
636 | | 13th octet | | |
637 Let's have a closer look at octet no. 13:
639 | |
641 |---------------|
642 |C|E|U|A|P|R|S|F|
643 |---------------|
644 |7 5 3 0|
645 These are the TCP control bits we are interested in. We have numbered
647 the bits in this octet from 0 to 7, right to left, so the PSH bit is
648 bit number 3, while the URG bit is number 5.
649 Recall that we want to capture packets with only SYN set. Let's see
651 what happens to octet 13 if a TCP datagram arrives with the SYN bit set
652 in its header:
653 |C|E|U|A|P|R|S|F|
655 |---------------|
656 |0 0 0 0 0 0 1 0|
657 |---------------|
658 |7 6 5 4 3 2 1 0|
659 Looking at the control bits section we see that only bit number 1 (SYN)
661 is set.
662 Assuming that octet number 13 is an 8-bit unsigned integer in network
664 byte order, the binary value of this octet is
665 00000010
667 and its decimal representation is
669 7 6 5 4 3 2 1 0
671 0*2 + 0*2 + 0*2 + 0*2 + 0*2 + 0*2 + 1*2 + 0*2 = 2
672 We're almost done, because now we know that if only SYN is set, the
674 value of the 13th octet in the TCP header, when interpreted as a 8-bit
675 unsigned integer in network byte order, must be exactly 2.
676 This relationship can be expressed as
678 tcp[13] == 2
679 We can use this expression as the filter for tcpdump in order to watch
681 packets which have only SYN set:
682 tcpdump -i xl0 tcp[13] == 2
683 The expression says "let the 13th octet of a TCP datagram have the dec‐
685 imal value 2", which is exactly what we want.
686 Now, let's assume that we need to capture SYN packets, but we don't
688 care if ACK or any other TCP control bit is set at the same time.
689 Let's see what happens to octet 13 when a TCP datagram with SYN-ACK set
690 arrives:
691 |C|E|U|A|P|R|S|F|
693 |---------------|
694 |0 0 0 1 0 0 1 0|
695 |---------------|
696 |7 6 5 4 3 2 1 0|
697 Now bits 1 and 4 are set in the 13th octet. The binary value of octet
699 13 is
700 00010010
702 which translates to decimal
704 7 6 5 4 3 2 1 0
706 0*2 + 0*2 + 0*2 + 1*2 + 0*2 + 0*2 + 1*2 + 0*2 = 18
707 Now we can't just use 'tcp[13] == 18' in the tcpdump filter expression,
709 because that would select only those packets that have SYN-ACK set, but
710 not those with only SYN set. Remember that we don't care if ACK or any
711 other control bit is set as long as SYN is set.
712 In order to achieve our goal, we need to logically AND the binary value
714 of octet 13 with some other value to preserve the SYN bit. We know
715 that we want SYN to be set in any case, so we'll logically AND the
716 value in the 13th octet with the binary value of a SYN:
717 00010010 SYN-ACK 00000010 SYN
719 AND 00000010 (we want SYN) AND 00000010 (we want SYN)
720 -------- --------
721 = 00000010 = 00000010
722 We see that this AND operation delivers the same result regardless
724 whether ACK or another TCP control bit is set. The decimal representa‐
725 tion of the AND value as well as the result of this operation is 2
726 (binary 00000010), so we know that for packets with SYN set the follow‐
727 ing relation must hold true:
728 ( ( value of octet 13 ) AND ( 2 ) ) == ( 2 )
730 This points us to the tcpdump filter expression
732 tcpdump -i xl0 'tcp[13] & 2 == 2'
733 Some offsets and field values may be expressed as names rather than as
735 numeric values. For example tcp[13] may be replaced with tcp[tcpflags].
736 The following TCP flag field values are also available: tcp-fin, tcp-
737 syn, tcp-rst, tcp-push, tcp-act, tcp-urg.
738 This can be demonstrated as:
740 tcpdump -i xl0 'tcp[tcpflags] & tcp-push != 0'
741 Note that you should use single quotes or a backslash in the expression
743 to hide the AND ('&') special character from the shell.
744 UDP Packets
746 UDP format is illustrated by this rwho packet:
748 actinide.who > broadcast.who: udp 84
749 This says that port who on host actinide sent a udp datagram to port
750 who on host broadcast, the Internet broadcast address. The packet con‐
751 tained 84 bytes of user data.
752 Some UDP services are recognized (from the source or destination port
754 number) and the higher level protocol information printed. In particu‐
755 lar, Domain Name service requests (RFC-1034/1035) and Sun RPC calls
756 (RFC-1050) to NFS.
757 UDP Name Server Requests
759 (N.B.:The following description assumes familiarity with the Domain
761 Service protocol described in RFC-1035. If you are not familiar with
762 the protocol, the following description will appear to be written in
763 greek.)
764 Name server requests are formatted as
766 src > dst: id op? flags qtype qclass name (len)
767 h2opolo.1538 > helios.domain: 3+ A? ucbvax.berkeley.edu. (37)
768 Host h2opolo asked the domain server on helios for an address record
769 (qtype=A) associated with the name ucbvax.berkeley.edu. The query id
770 was `3'. The `+' indicates the recursion desired flag was set. The
771 query length was 37 bytes, not including the UDP and IP protocol head‐
772 ers. The query operation was the normal one, Query, so the op field
773 was omitted. If the op had been anything else, it would have been
774 printed between the `3' and the `+'. Similarly, the qclass was the
775 normal one, C_IN, and omitted. Any other qclass would have been
776 printed immediately after the `A'.
777 A few anomalies are checked and may result in extra fields enclosed in
779 square brackets: If a query contains an answer, authority records or
780 additional records section, ancount, nscount, or arcount are printed as
781 `[na]', `[nn]' or `[nau]' where n is the appropriate count. If any of
782 the response bits are set (AA, RA or rcode) or any of the `must be
783 zero' bits are set in bytes two and three, `[b2&3=x]' is printed, where
784 x is the hex value of header bytes two and three.
785 UDP Name Server Responses
787 Name server responses are formatted as
789 src > dst: id op rcode flags a/n/au type class data (len)
790 helios.domain > h2opolo.1538: 3 3/3/7 A 128.32.137.3 (273)
791 helios.domain > h2opolo.1537: 2 NXDomain* 0/1/0 (97)
792 In the first example, helios responds to query id 3 from h2opolo with 3
793 answer records, 3 name server records and 7 additional records. The
794 first answer record is type A (address) and its data is internet
795 address 128.32.137.3. The total size of the response was 273 bytes,
796 excluding UDP and IP headers. The op (Query) and response code (NoEr‐
797 ror) were omitted, as was the class (C_IN) of the A record.
798 In the second example, helios responds to query 2 with a response code
800 of non-existent domain (NXDomain) with no answers, one name server and
801 no authority records. The `*' indicates that the authoritative answer
802 bit was set. Since there were no answers, no type, class or data were
803 printed.
804 Other flag characters that might appear are `-' (recursion available,
806 RA, not set) and `|' (truncated message, TC, set). If the `question'
807 section doesn't contain exactly one entry, `[nq]' is printed.
808 SMB/CIFS decoding
811 tcpdump now includes fairly extensive SMB/CIFS/NBT decoding for data on
813 UDP/137, UDP/138 and TCP/139. Some primitive decoding of IPX and Net‐
814 BEUI SMB data is also done.
815 By default a fairly minimal decode is done, with a much more detailed
817 decode done if -v is used. Be warned that with -v a single SMB packet
818 may take up a page or more, so only use -v if you really want all the
819 gory details.
820 For information on SMB packet formats and what all te fields mean see
822 www.cifs.org or the pub/samba/specs/ directory on your favorite
823 samba.org mirror site. The SMB patches were written by Andrew Tridgell
824 (tridge@samba.org).
825 NFS Requests and Replies
828 Sun NFS (Network File System) requests and replies are printed as:
830 src.xid > dst.nfs: len op args
831 src.nfs > dst.xid: reply stat len op results
832 sushi.6709 > wrl.nfs: 112 readlink fh 21,24/10.73165
833 wrl.nfs > sushi.6709: reply ok 40 readlink "../var"
834 sushi.201b > wrl.nfs:
835 144 lookup fh 9,74/4096.6878 "xcolors"
836 wrl.nfs > sushi.201b:
837 reply ok 128 lookup fh 9,74/4134.3150
838 In the first line, host sushi sends a transaction with id 6709 to wrl
839 (note that the number following the src host is a transaction id, not
840 the source port). The request was 112 bytes, excluding the UDP and IP
841 headers. The operation was a readlink (read symbolic link) on file
842 handle (fh) 21,24/10.731657119. (If one is lucky, as in this case, the
843 file handle can be interpreted as a major,minor device number pair,
844 followed by the inode number and generation number.) Wrl replies `ok'
845 with the contents of the link.
846 In the third line, sushi asks wrl to lookup the name `xcolors' in
848 directory file 9,74/4096.6878. Note that the data printed depends on
849 the operation type. The format is intended to be self explanatory if
850 read in conjunction with an NFS protocol spec.
851 If the -v (verbose) flag is given, additional information is printed.
853 For example:
854 sushi.1372a > wrl.nfs:
855 148 read fh 21,11/12.195 8192 bytes @ 24576
856 wrl.nfs > sushi.1372a:
857 reply ok 1472 read REG 100664 ids 417/0 sz 29388
858 (-v also prints the IP header TTL, ID, length, and fragmentation
859 fields, which have been omitted from this example.) In the first line,
860 sushi asks wrl to read 8192 bytes from file 21,11/12.195, at byte off‐
861 set 24576. Wrl replies `ok'; the packet shown on the second line is
862 the first fragment of the reply, and hence is only 1472 bytes long (the
863 other bytes will follow in subsequent fragments, but these fragments do
864 not have NFS or even UDP headers and so might not be printed, depending
865 on the filter expression used). Because the -v flag is given, some of
866 the file attributes (which are returned in addition to the file data)
867 are printed: the file type (``REG'', for regular file), the file mode
868 (in octal), the uid and gid, and the file size.
869 If the -v flag is given more than once, even more details are printed.
871 Note that NFS requests are very large and much of the detail won't be
873 printed unless snaplen is increased. Try using `-s 192' to watch NFS
874 traffic.
875 NFS reply packets do not explicitly identify the RPC operation.
877 Instead, tcpdump keeps track of ``recent'' requests, and matches them
878 to the replies using the transaction ID. If a reply does not closely
879 follow the corresponding request, it might not be parsable.
880 AFS Requests and Replies
882 Transarc AFS (Andrew File System) requests and replies are printed as:
884 src.sport > dst.dport: rx packet-type
886 src.sport > dst.dport: rx packet-type service call call-name args
887 src.sport > dst.dport: rx packet-type service reply call-name args
888 elvis.7001 > pike.afsfs:
889 rx data fs call rename old fid 536876964/1/1 ".newsrc.new"
890 new fid 536876964/1/1 ".newsrc"
891 pike.afsfs > elvis.7001: rx data fs reply rename
892 In the first line, host elvis sends a RX packet to pike. This was a RX
893 data packet to the fs (fileserver) service, and is the start of an RPC
894 call. The RPC call was a rename, with the old directory file id of
895 536876964/1/1 and an old filename of `.newsrc.new', and a new directory
896 file id of 536876964/1/1 and a new filename of `.newsrc'. The host
897 pike responds with a RPC reply to the rename call (which was success‐
898 ful, because it was a data packet and not an abort packet).
899 In general, all AFS RPCs are decoded at least by RPC call name. Most
901 AFS RPCs have at least some of the arguments decoded (generally only
902 the `interesting' arguments, for some definition of interesting).
903 The format is intended to be self-describing, but it will probably not
905 be useful to people who are not familiar with the workings of AFS and
906 RX.
907 If the -v (verbose) flag is given twice, acknowledgement packets and
909 additional header information is printed, such as the the RX call ID,
910 call number, sequence number, serial number, and the RX packet flags.
911 If the -v flag is given twice, additional information is printed, such
913 as the the RX call ID, serial number, and the RX packet flags. The MTU
914 negotiation information is also printed from RX ack packets.
915 If the -v flag is given three times, the security index and service id
917 are printed.
918 Error codes are printed for abort packets, with the exception of Ubik
920 beacon packets (because abort packets are used to signify a yes vote
921 for the Ubik protocol).
922 Note that AFS requests are very large and many of the arguments won't
924 be printed unless snaplen is increased. Try using `-s 256' to watch
925 AFS traffic.
926 AFS reply packets do not explicitly identify the RPC operation.
928 Instead, tcpdump keeps track of ``recent'' requests, and matches them
929 to the replies using the call number and service ID. If a reply does
930 not closely follow the corresponding request, it might not be parsable.
931 KIP AppleTalk (DDP in UDP)
934 AppleTalk DDP packets encapsulated in UDP datagrams are de-encapsulated
936 and dumped as DDP packets (i.e., all the UDP header information is dis‐
937 carded). The file /etc/atalk.names is used to translate AppleTalk net
938 and node numbers to names. Lines in this file have the form
939 number name
940 1.254 ether
942 16.1 icsd-net
943 1.254.110 ace
944 The first two lines give the names of AppleTalk networks. The third
945 line gives the name of a particular host (a host is distinguished from
946 a net by the 3rd octet in the number - a net number must have two
947 octets and a host number must have three octets.) The number and name
948 should be separated by whitespace (blanks or tabs). The
949 /etc/atalk.names file may contain blank lines or comment lines (lines
950 starting with a `#').
951 AppleTalk addresses are printed in the form
953 net.host.port
954 144.1.209.2 > icsd-net.112.220
956 office.2 > icsd-net.112.220
957 jssmag.149.235 > icsd-net.2
958 (If the /etc/atalk.names doesn't exist or doesn't contain an entry for
959 some AppleTalk host/net number, addresses are printed in numeric form.)
960 In the first example, NBP (DDP port 2) on net 144.1 node 209 is sending
961 to whatever is listening on port 220 of net icsd node 112. The second
962 line is the same except the full name of the source node is known
963 (`office'). The third line is a send from port 235 on net jssmag node
964 149 to broadcast on the icsd-net NBP port (note that the broadcast
965 address (255) is indicated by a net name with no host number - for this
966 reason it's a good idea to keep node names and net names distinct in
967 /etc/atalk.names).
968 NBP (name binding protocol) and ATP (AppleTalk transaction protocol)
970 packets have their contents interpreted. Other protocols just dump the
971 protocol name (or number if no name is registered for the protocol) and
972 packet size.
973 NBP packets are formatted like the following examples:
975 icsd-net.112.220 > jssmag.2: nbp-lkup 190: "=:LaserWriter@*"
976 jssmag.209.2 > icsd-net.112.220: nbp-reply 190: "RM1140:LaserWriter@*" 250
977 techpit.2 > icsd-net.112.220: nbp-reply 190: "techpit:LaserWriter@*" 186
978 The first line is a name lookup request for laserwriters sent by net
979 icsd host 112 and broadcast on net jssmag. The nbp id for the lookup
980 is 190. The second line shows a reply for this request (note that it
981 has the same id) from host jssmag.209 saying that it has a laserwriter
982 resource named "RM1140" registered on port 250. The third line is
983 another reply to the same request saying host techpit has laserwriter
984 "techpit" registered on port 186.
985 ATP packet formatting is demonstrated by the following example:
987 jssmag.209.165 > helios.132: atp-req 12266<0-7> 0xae030001
988 helios.132 > jssmag.209.165: atp-resp 12266:0 (512) 0xae040000
989 helios.132 > jssmag.209.165: atp-resp 12266:1 (512) 0xae040000
990 helios.132 > jssmag.209.165: atp-resp 12266:2 (512) 0xae040000
991 helios.132 > jssmag.209.165: atp-resp 12266:3 (512) 0xae040000
992 helios.132 > jssmag.209.165: atp-resp 12266:4 (512) 0xae040000
993 helios.132 > jssmag.209.165: atp-resp 12266:5 (512) 0xae040000
994 helios.132 > jssmag.209.165: atp-resp 12266:6 (512) 0xae040000
995 helios.132 > jssmag.209.165: atp-resp*12266:7 (512) 0xae040000
996 jssmag.209.165 > helios.132: atp-req 12266<3,5> 0xae030001
997 helios.132 > jssmag.209.165: atp-resp 12266:3 (512) 0xae040000
998 helios.132 > jssmag.209.165: atp-resp 12266:5 (512) 0xae040000
999 jssmag.209.165 > helios.132: atp-rel 12266<0-7> 0xae030001
1000 jssmag.209.133 > helios.132: atp-req* 12267<0-7> 0xae030002
1001 Jssmag.209 initiates transaction id 12266 with host helios by request‐
1002 ing up to 8 packets (the `<0-7>'). The hex number at the end of the
1003 line is the value of the `userdata' field in the request.
1004 Helios responds with 8 512-byte packets. The `:digit' following the
1006 transaction id gives the packet sequence number in the transaction and
1007 the number in parens is the amount of data in the packet, excluding the
1008 atp header. The `*' on packet 7 indicates that the EOM bit was set.
1009 Jssmag.209 then requests that packets 3 & 5 be retransmitted. Helios
1011 resends them then jssmag.209 releases the transaction. Finally, jss‐
1012 mag.209 initiates the next request. The `*' on the request indicates
1013 that XO (`exactly once') was not set.
1014 IP Fragmentation
1017 Fragmented Internet datagrams are printed as
1019 (frag id:size@offset+)
1020 (frag id:size@offset)
1021 (The first form indicates there are more fragments. The second indi‐
1022 cates this is the last fragment.)
1023 Id is the fragment id. Size is the fragment size (in bytes) excluding
1025 the IP header. Offset is this fragment's offset (in bytes) in the
1026 original datagram.
1027 The fragment information is output for each fragment. The first frag‐
1029 ment contains the higher level protocol header and the frag info is
1030 printed after the protocol info. Fragments after the first contain no
1031 higher level protocol header and the frag info is printed after the
1032 source and destination addresses. For example, here is part of an ftp
1033 from arizona.edu to lbl-rtsg.arpa over a CSNET connection that doesn't
1034 appear to handle 576 byte datagrams:
1035 arizona.ftp-data > rtsg.1170: . 1024:1332(308) ack 1 win 4096 (frag 595a:328@0+)
1036 arizona > rtsg: (frag 595a:204@328)
1037 rtsg.1170 > arizona.ftp-data: . ack 1536 win 2560
1038 There are a couple of things to note here: First, addresses in the 2nd
1039 line don't include port numbers. This is because the TCP protocol
1040 information is all in the first fragment and we have no idea what the
1041 port or sequence numbers are when we print the later fragments. Sec‐
1042 ond, the tcp sequence information in the first line is printed as if
1043 there were 308 bytes of user data when, in fact, there are 512 bytes
1044 (308 in the first frag and 204 in the second). If you are looking for
1045 holes in the sequence space or trying to match up acks with packets,
1046 this can fool you.
1047 A packet with the IP don't fragment flag is marked with a trailing
1049 (DF).
1050 Timestamps
1052 By default, all output lines are preceded by a timestamp. The time‐
1054 stamp is the current clock time in the form
1055 hh:mm:ss.frac
1056 and is as accurate as the kernel's clock. The timestamp reflects the
1057 time the kernel first saw the packet. No attempt is made to account
1058 for the time lag between when the Ethernet interface removed the packet
1059 from the wire and when the kernel serviced the `new packet' interrupt.
1060
1062 stty(1), pcap(3PCAP), bpf(4), nit(4P), pcap-savefile(5), pcap-filter(7)
1063
1065 The original authors are:
1066 Van Jacobson, Craig Leres and Steven McCanne, all of the Lawrence
1068 Berkeley National Laboratory, University of California, Berkeley, CA.
1069 It is currently being maintained by tcpdump.org.
1071 The current version is available via http:
1073 http://www.tcpdump.org/
1075 The original distribution is available via anonymous ftp:
1077 ftp://ftp.ee.lbl.gov/tcpdump.tar.Z
1079 IPv6/IPsec support is added by WIDE/KAME project. This program uses
1081 Eric Young's SSLeay library, under specific configurations.
1082
1084 Please send problems, bugs, questions, desirable enhancements, patches
1085 etc. to:
1086 tcpdump-workers@lists.tcpdump.org
1088 NIT doesn't let you watch your own outbound traffic, BPF will. We rec‐
1090 ommend that you use the latter.
1091 On Linux systems with 2.0[.x] kernels:
1093 packets on the loopback device will be seen twice;
1095 packet filtering cannot be done in the kernel, so that all pack‐
1097 ets must be copied from the kernel in order to be filtered in
1098 user mode;
1099 all of a packet, not just the part that's within the snapshot
1101 length, will be copied from the kernel (the 2.0[.x] packet cap‐
1102 ture mechanism, if asked to copy only part of a packet to user‐
1103 land, will not report the true length of the packet; this would
1104 cause most IP packets to get an error from tcpdump);
1105 capturing on some PPP devices won't work correctly.
1107 We recommend that you upgrade to a 2.2 or later kernel.
1109 Some attempt should be made to reassemble IP fragments or, at least to
1111 compute the right length for the higher level protocol.
1112 Name server inverse queries are not dumped correctly: the (empty) ques‐
1114 tion section is printed rather than real query in the answer section.
1115 Some believe that inverse queries are themselves a bug and prefer to
1116 fix the program generating them rather than tcpdump.
1117 A packet trace that crosses a daylight savings time change will give
1119 skewed time stamps (the time change is ignored).
1120 Filter expressions on fields other than those in Token Ring headers
1122 will not correctly handle source-routed Token Ring packets.
1123 Filter expressions on fields other than those in 802.11 headers will
1125 not correctly handle 802.11 data packets with both To DS and From DS
1126 set.
1127 ip6 proto should chase header chain, but at this moment it does not.
1129 ip6 protochain is supplied for this behavior.
1130 Arithmetic expression against transport layer headers, like tcp[0],
1132 does not work against IPv6 packets. It only looks at IPv4 packets.
1133 05 March 2009 TCPDUMP(8)