1 .\" Copyright (c) 1990 The Regents of the University of California.
2 .\" All rights reserved.
4 .\" Redistribution and use in source and binary forms, with or without
5 .\" modification, are permitted provided that: (1) source code distributions
6 .\" retain the above copyright notice and this paragraph in its entirety, (2)
7 .\" distributions including binary code include the above copyright notice and
8 .\" this paragraph in its entirety in the documentation or other materials
9 .\" provided with the distribution, and (3) all advertising materials mentioning
10 .\" features or use of this software display the following acknowledgement:
11 .\" ``This product includes software developed by the University of California,
12 .\" Lawrence Berkeley Laboratory and its contributors.'' Neither the name of
13 .\" the University nor the names of its contributors may be used to endorse
14 .\" or promote products derived from this software without specific prior
15 .\" written permission.
16 .\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
17 .\" WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
18 .\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
20 .\" This document is derived in part from the enet man page (enet.4)
21 .\" distributed with 4.3BSD Unix.
23 .\" $FreeBSD: src/share/man/man4/bpf.4,v 1.21.2.10 2001/08/17 13:08:37 ru Exp $
30 .Nd Berkeley Packet Filter
34 The Berkeley Packet Filter
35 provides a raw interface to data link layers in a protocol
37 All packets on the network, even those destined for other hosts,
38 are accessible through this mechanism.
40 The packet filter appears as a character special device,
44 After opening the device, the file descriptor must be bound to a
45 specific network interface with the
48 A given interface can be shared be multiple listeners, and the filter
49 underlying each descriptor will see an identical packet stream.
51 A separate device file is required for each minor device.
52 If a file is in use, the open will fail and
57 Associated with each open instance of a
59 file is a user-settable packet filter.
60 Whenever a packet is received by an interface,
61 all file descriptors listening on that interface apply their filter.
62 Each descriptor that accepts the packet receives its own copy.
64 Reads from these files return the next group of packets
65 that have matched the filter.
66 To improve performance, the buffer passed to read must be
67 the same size as the buffers used internally by
69 This size is returned by the
71 ioctl (see below), and
74 Note that an individual packet larger than this size is necessarily
77 The packet filter will support any link level protocol that has fixed length
78 headers. Currently, only Ethernet,
82 drivers have been modified to interact with
85 A packet can be sent out on the network by writing to a
87 file descriptor. The writes are unbuffered, meaning only one
88 packet can be processed per write.
89 Currently, only writes to Ethernets and
93 When the last minor device is opened, an additional minor device is
94 created on demand. The maximum number of devices that can be created is
95 controlled by the sysctl debug.bpf_maxdevices.
99 command codes below are defined in
104 #include <sys/types.h>
105 #include <sys/time.h>
106 #include <sys/ioctl.h>
121 the following commands may be applied to any open
124 The (third) argument to
126 should be a pointer to the type indicated.
127 .Bl -tag -width BIOCGRTIMEOUT
130 Returns the required buffer length for reads on
135 Sets the buffer length for reads on
137 files. The buffer must be set before the file is attached to an interface
140 If the requested buffer size cannot be accommodated, the closest
141 allowable size will be set and returned in the argument.
142 A read call will result in
144 if it is passed a buffer that is not this size.
147 Returns the type of the data link layer underlying the attached interface.
149 is returned if no interface has been specified.
150 The device types, prefixed with
156 Used to specify the type of data link layer of the interface
157 attached to the bpf descriptor. If the current interface is
158 not of the given type then the descriptor will be reattached
159 to an interface of the given type. If the descriptor has
160 promiscuous mode set, the new interface will be moved to
163 is returned if no interface has been specified.
164 The device types, prefixed with
169 .Pq Li struct bpf_dltlist
170 Returns a list of data link types of the the given interface.
171 A user allocated buffer to hold the list and length of expected
172 list should be provided in struct bpf_dltlist, defined in
175 is returned if no interface has been specified.
176 The device types, prefixed with
181 Forces the interface into promiscuous mode.
182 All packets, not just those destined for the local host, are processed.
183 Since more than one file can be listening on a given interface,
184 a listener that opened its interface non-promiscuously may receive
185 packets promiscuously. This problem can be remedied with an
188 Flushes the buffer of incoming packets,
189 and resets the statistics that are returned by BIOCGSTATS.
191 .Pq Li "struct ifreq"
192 Returns the name of the hardware interface that the file is listening on.
193 The name is returned in the ifr_name field of
197 All other fields are undefined.
199 .Pq Li "struct ifreq"
200 Sets the hardware interface associate with the file. This
201 command must be performed before any packets can be read.
202 The device is indicated by name using the
207 Additionally, performs the actions of
211 .Pq Li "struct timeval"
212 Set or get the read timeout parameter.
214 specifies the length of time to wait before timing
215 out on a read request.
216 This parameter is initialized to zero by
218 indicating no timeout.
220 .Pq Li "struct bpf_stat"
221 Returns the following structure of packet statistics:
224 u_int bs_recv; /* number of packets received */
225 u_int bs_drop; /* number of packets dropped */
230 .Bl -hang -offset indent
232 the number of packets received by the descriptor since opened or reset
233 (including any buffered since the last read call);
236 the number of packets which were accepted by the filter but dropped by the
237 kernel because of buffer overflows
238 (i.e., the application's reads aren't keeping up with the packet traffic).
244 based on the truth value of the argument.
245 When immediate mode is enabled, reads return immediately upon packet
246 reception. Otherwise, a read will block until either the kernel buffer
247 becomes full or a timeout occurs.
248 This is useful for programs like
250 which must respond to messages in real time.
251 The default for a new file is off.
253 .Pq Li "struct bpf_program"
254 Sets the filter program used by the kernel to discard uninteresting
255 packets. An array of instructions and its length is passed in using
256 the following structure:
260 struct bpf_insn *bf_insns;
264 The filter program is pointed to by the
266 field while its length in units of
267 .Sq Li struct bpf_insn
276 for an explanation of the filter language.
278 .Pq Li "struct bpf_version"
279 Returns the major and minor version numbers of the filter language currently
280 recognized by the kernel. Before installing a filter, applications must check
281 that the current version is compatible with the running kernel. Version
282 numbers are compatible if the major numbers match and the application minor
283 is less than or equal to the kernel minor. The kernel version number is
284 returned in the following structure:
292 The current version numbers are given by
293 .Dv BPF_MAJOR_VERSION
295 .Dv BPF_MINOR_VERSION
298 An incompatible filter
299 may result in undefined behavior (most likely, an error returned by
301 or haphazard packet matching).
305 Set or get the status of the
308 Set to zero if the link level source address should be filled in automatically
309 by the interface output routine. Set to one if the link level source
310 address will be written, as provided, to the wire. This flag is initialized
315 Set or get the flag determining whether locally generated packets on the
316 interface should be returned by BPF. Set to zero to see only incoming
317 packets on the interface. Set to one to see packets originating
318 locally and remotely on the interface. This flag is initialized to one by
322 Returns the signal that will be sent to a process waiting on the bpf descriptor upon packet reception. The default is SIGIO.
325 Sets the signal that should be sent to a process waiting on bpf descriptor upon packet reception. The default is SIGIO.
328 The following structure is prepended to each packet returned by
332 struct timeval bh_tstamp; /* time stamp */
333 u_long bh_caplen; /* length of captured portion */
334 u_long bh_datalen; /* original length of packet */
335 u_short bh_hdrlen; /* length of bpf header (this struct
336 plus alignment padding */
340 The fields, whose values are stored in host order, and are:
342 .Bl -tag -compact -width bh_datalen
344 The time at which the packet was processed by the packet filter.
346 The length of the captured portion of the packet. This is the minimum of
347 the truncation amount specified by the filter and the length of the packet.
349 The length of the packet off the wire.
350 This value is independent of the truncation amount specified by the filter.
354 header, which may not be equal to
355 .\" XXX - not really a function call
356 .Fn sizeof "struct bpf_hdr" .
361 field exists to account for
362 padding between the header and the link level protocol.
363 The purpose here is to guarantee proper alignment of the packet
364 data structures, which is required on alignment sensitive
365 architectures and improves performance on many other architectures.
366 The packet filter insures that the
368 and the network layer
369 header will be word aligned. Suitable precautions
370 must be taken when accessing the link layer protocol fields on alignment
371 restricted machines. (This isn't a problem on an Ethernet, since
372 the type field is a short falling on an even offset,
373 and the addresses are probably accessed in a bytewise fashion).
375 Additionally, individual packets are padded so that each starts
376 on a word boundary. This requires that an application
377 has some knowledge of how to get from packet to packet.
383 this process. It rounds up its argument
384 to the nearest word aligned value (where a word is
390 points to the start of a packet, this expression
391 will advance it to the next packet:
392 .Dl p = (char *)p + BPF_WORDALIGN(p->bh_hdrlen + p->bh_caplen)
394 For the alignment mechanisms to work properly, the
397 must itself be word aligned.
401 will always return an aligned buffer.
403 A filter program is an array of instructions, with all branches forwardly
404 directed, terminated by a
407 Each instruction performs some action on the pseudo-machine state,
408 which consists of an accumulator, index register, scratch memory store,
409 and implicit program counter.
411 The following structure defines the instruction format:
423 field is used in different ways by different instructions,
428 fields are used as offsets
429 by the branch instructions.
430 The opcodes are encoded in a semi-hierarchical fashion.
431 There are eight classes of instructions:
441 Various other mode and
442 operator bits are or'd into the class to give the actual instructions.
443 The classes and modes are defined in
446 Below are the semantics for each defined
449 We use the convention that A is the accumulator, X is the index register,
450 P[] packet data, and M[] scratch memory store.
451 P[i:n] gives the data at byte offset
454 interpreted as a word (n=4),
455 unsigned halfword (n=2), or unsigned byte (n=1).
456 M[i] gives the i'th word in the scratch memory store, which is only
457 addressed in word units. The memory store is indexed from 0 to
464 are the corresponding fields in the
465 instruction definition.
467 refers to the length of the packet.
469 .Bl -tag -width BPF_STXx
471 These instructions copy a value into the accumulator. The type of the
472 source operand is specified by an
474 and can be a constant
476 packet data at a fixed offset
478 packet data at a variable offset
482 or a word in the scratch memory store
488 the data size must be specified as a word
494 The semantics of all the recognized
498 .Bl -tag -width "BPF_LD+BPF_W+BPF_IND" -compact
499 .It Li BPF_LD+BPF_W+BPF_ABS
501 .It Li BPF_LD+BPF_H+BPF_ABS
503 .It Li BPF_LD+BPF_B+BPF_ABS
505 .It Li BPF_LD+BPF_W+BPF_IND
507 .It Li BPF_LD+BPF_H+BPF_IND
509 .It Li BPF_LD+BPF_B+BPF_IND
511 .It Li BPF_LD+BPF_W+BPF_LEN
513 .It Li BPF_LD+BPF_IMM
515 .It Li BPF_LD+BPF_MEM
519 These instructions load a value into the index register. Note that
520 the addressing modes are more restrictive than those of the accumulator loads,
523 a hack for efficiently loading the IP header length.
525 .Bl -tag -width "BPF_LDX+BPF_W+BPF_MEM" -compact
526 .It Li BPF_LDX+BPF_W+BPF_IMM
528 .It Li BPF_LDX+BPF_W+BPF_MEM
530 .It Li BPF_LDX+BPF_W+BPF_LEN
532 .It Li BPF_LDX+BPF_B+BPF_MSH
536 This instruction stores the accumulator into the scratch memory.
537 We do not need an addressing mode since there is only one possibility
540 .Bl -tag -width "BPF_ST" -compact
545 This instruction stores the index register in the scratch memory store.
547 .Bl -tag -width "BPF_STX" -compact
552 The alu instructions perform operations between the accumulator and
553 index register or constant, and store the result back in the accumulator.
554 For binary operations, a source mode is required
559 .Bl -tag -width "BPF_ALU+BPF_MUL+BPF_K" -compact
560 .It Li BPF_ALU+BPF_ADD+BPF_K
562 .It Li BPF_ALU+BPF_SUB+BPF_K
564 .It Li BPF_ALU+BPF_MUL+BPF_K
566 .It Li BPF_ALU+BPF_DIV+BPF_K
568 .It Li BPF_ALU+BPF_AND+BPF_K
570 .It Li BPF_ALU+BPF_OR+BPF_K
572 .It Li BPF_ALU+BPF_LSH+BPF_K
574 .It Li BPF_ALU+BPF_RSH+BPF_K
576 .It Li BPF_ALU+BPF_ADD+BPF_X
578 .It Li BPF_ALU+BPF_SUB+BPF_X
580 .It Li BPF_ALU+BPF_MUL+BPF_X
582 .It Li BPF_ALU+BPF_DIV+BPF_X
584 .It Li BPF_ALU+BPF_AND+BPF_X
586 .It Li BPF_ALU+BPF_OR+BPF_X
588 .It Li BPF_ALU+BPF_LSH+BPF_X
590 .It Li BPF_ALU+BPF_RSH+BPF_X
592 .It Li BPF_ALU+BPF_NEG
596 The jump instructions alter flow of control. Conditional jumps
597 compare the accumulator against a constant
599 or the index register
601 If the result is true (or non-zero),
602 the true branch is taken, otherwise the false branch is taken.
603 Jump offsets are encoded in 8 bits so the longest jump is 256 instructions.
604 However, the jump always
606 opcode uses the 32 bit
608 field as the offset, allowing arbitrarily distant destinations.
609 All conditionals use unsigned comparison conventions.
611 .Bl -tag -width "BPF_JMP+BPF_KSET+BPF_X" -compact
612 .It Li BPF_JMP+BPF_JA
614 .It Li BPF_JMP+BPF_JGT+BPF_K
615 pc += (A > k) ? jt : jf
616 .It Li BPF_JMP+BPF_JGE+BPF_K
617 pc += (A >= k) ? jt : jf
618 .It Li BPF_JMP+BPF_JEQ+BPF_K
619 pc += (A == k) ? jt : jf
620 .It Li BPF_JMP+BPF_JSET+BPF_K
621 pc += (A & k) ? jt : jf
622 .It Li BPF_JMP+BPF_JGT+BPF_X
623 pc += (A > X) ? jt : jf
624 .It Li BPF_JMP+BPF_JGE+BPF_X
625 pc += (A >= X) ? jt : jf
626 .It Li BPF_JMP+BPF_JEQ+BPF_X
627 pc += (A == X) ? jt : jf
628 .It Li BPF_JMP+BPF_JSET+BPF_X
629 pc += (A & X) ? jt : jf
632 The return instructions terminate the filter program and specify the amount
633 of packet to accept (i.e., they return the truncation amount). A return
634 value of zero indicates that the packet should be ignored.
635 The return value is either a constant
640 .Bl -tag -width "BPF_RET+BPF_K" -compact
647 The miscellaneous category was created for anything that doesn't
648 fit into the above classes, and for any new instructions that might need to
649 be added. Currently, these are the register transfer instructions
650 that copy the index register to the accumulator or vice versa.
652 .Bl -tag -width "BPF_MISC+BPF_TAX" -compact
653 .It Li BPF_MISC+BPF_TAX
655 .It Li BPF_MISC+BPF_TXA
662 interface provides the following macros to facilitate
664 .Fn BPF_STMT opcode operand
666 .Fn BPF_JUMP opcode operand true_offset false_offset .
668 The following filter is taken from the Reverse ARP Daemon. It accepts
669 only Reverse ARP requests.
671 struct bpf_insn insns[] = {
672 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
673 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_REVARP, 0, 3),
674 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
675 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, REVARP_REQUEST, 0, 1),
676 BPF_STMT(BPF_RET+BPF_K, sizeof(struct ether_arp) +
677 sizeof(struct ether_header)),
678 BPF_STMT(BPF_RET+BPF_K, 0),
682 This filter accepts only IP packets between host 128.3.112.15 and
685 struct bpf_insn insns[] = {
686 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
687 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 8),
688 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 26),
689 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 2),
690 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
691 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 3, 4),
692 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 0, 3),
693 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
694 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 1),
695 BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
696 BPF_STMT(BPF_RET+BPF_K, 0),
700 Finally, this filter returns only TCP finger packets. We must parse
701 the IP header to reach the TCP header. The
704 checks that the IP fragment offset is 0 so we are sure
705 that we have a TCP header.
707 struct bpf_insn insns[] = {
708 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
709 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 10),
710 BPF_STMT(BPF_LD+BPF_B+BPF_ABS, 23),
711 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, IPPROTO_TCP, 0, 8),
712 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
713 BPF_JUMP(BPF_JMP+BPF_JSET+BPF_K, 0x1fff, 6, 0),
714 BPF_STMT(BPF_LDX+BPF_B+BPF_MSH, 14),
715 BPF_STMT(BPF_LD+BPF_H+BPF_IND, 14),
716 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 2, 0),
717 BPF_STMT(BPF_LD+BPF_H+BPF_IND, 16),
718 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 0, 1),
719 BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
720 BPF_STMT(BPF_RET+BPF_K, 0),
729 .%T "An efficient, extensible, and portable network monitor"
732 .Bl -tag -compact -width /dev/bpfXXX
733 .It Pa /dev/bpf Ns Sy n
734 the packet filter device
737 The read buffer must be of a fixed size (returned by the
741 A file that does not request promiscuous mode may receive promiscuously
742 received packets as a side effect of another file requesting this
743 mode on the same hardware interface. This could be fixed in the kernel
744 with additional processing overhead. However, we favor the model where
745 all files must assume that the interface is promiscuous, and if
746 so desired, must utilize a filter to reject foreign packets.
748 Data link protocols with variable length headers are not currently supported.
750 The Enet packet filter was created in 1980 by Mike Accetta and
751 Rick Rashid at Carnegie-Mellon University. Jeffrey Mogul, at
752 Stanford, ported the code to
754 and continued its development from
755 1983 on. Since then, it has evolved into the Ultrix Packet Filter
768 of Lawrence Berkeley Laboratory, implemented BPF in
769 Summer 1990. Much of the design is due to