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 1.21.2.11 2002/04/07 04:57:13 dd 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 by 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 Since packet data is in network byte order, applications should use the
87 macros to extract multi-byte values.
89 A packet can be sent out on the network by writing to a
91 file descriptor. The writes are unbuffered, meaning only one
92 packet can be processed per write.
93 Currently, only writes to Ethernets and
99 command codes below are defined in
104 #include <sys/types.h>
105 #include <sys/time.h>
106 #include <sys/ioctl.h>
123 the following commands may be applied to any open
126 The (third) argument to
128 should be a pointer to the type indicated.
129 .Bl -tag -width BIOCGRTIMEOUT
132 Returns the required buffer length for reads on
137 Sets the buffer length for reads on
139 files. The buffer must be set before the file is attached to an interface
142 If the requested buffer size cannot be accommodated, the closest
143 allowable size will be set and returned in the argument.
144 A read call will result in
146 if it is passed a buffer that is not this size.
149 Returns the type of the data link layer underlying the attached interface.
151 is returned if no interface has been specified.
152 The device types, prefixed with
157 Forces the interface into promiscuous mode.
158 All packets, not just those destined for the local host, are processed.
159 Since more than one file can be listening on a given interface,
160 a listener that opened its interface non-promiscuously may receive
161 packets promiscuously. This problem can be remedied with an
164 Flushes the buffer of incoming packets,
165 and resets the statistics that are returned by BIOCGSTATS.
167 .Pq Li "struct ifreq"
168 Returns the name of the hardware interface that the file is listening on.
169 The name is returned in the ifr_name field of
173 All other fields are undefined.
175 .Pq Li "struct ifreq"
176 Sets the hardware interface associate with the file. This
177 command must be performed before any packets can be read.
178 The device is indicated by name using the
183 Additionally, performs the actions of
187 .Pq Li "struct timeval"
188 Set or get the read timeout parameter.
190 specifies the length of time to wait before timing
191 out on a read request.
192 This parameter is initialized to zero by
194 indicating no timeout.
196 .Pq Li "struct bpf_stat"
197 Returns the following structure of packet statistics:
200 u_int bs_recv; /* number of packets received */
201 u_int bs_drop; /* number of packets dropped */
206 .Bl -hang -offset indent
208 the number of packets received by the descriptor since opened or reset
209 (including any buffered since the last read call);
212 the number of packets which were accepted by the filter but dropped by the
213 kernel because of buffer overflows
214 (i.e., the application's reads aren't keeping up with the packet traffic).
220 based on the truth value of the argument.
221 When immediate mode is enabled, reads return immediately upon packet
222 reception. Otherwise, a read will block until either the kernel buffer
223 becomes full or a timeout occurs.
224 This is useful for programs like
226 which must respond to messages in real time.
227 The default for a new file is off.
229 .Pq Li "struct bpf_program"
230 Sets the filter program used by the kernel to discard uninteresting
231 packets. An array of instructions and its length is passed in using
232 the following structure:
236 struct bpf_insn *bf_insns;
240 The filter program is pointed to by the
242 field while its length in units of
243 .Sq Li struct bpf_insn
252 for an explanation of the filter language.
254 .Pq Li "struct bpf_version"
255 Returns the major and minor version numbers of the filter language currently
256 recognized by the kernel. Before installing a filter, applications must check
257 that the current version is compatible with the running kernel. Version
258 numbers are compatible if the major numbers match and the application minor
259 is less than or equal to the kernel minor. The kernel version number is
260 returned in the following structure:
268 The current version numbers are given by
269 .Dv BPF_MAJOR_VERSION
271 .Dv BPF_MINOR_VERSION
274 An incompatible filter
275 may result in undefined behavior (most likely, an error returned by
277 or haphazard packet matching).
281 Set or get the status of the
284 Set to zero if the link level source address should be filled in automatically
285 by the interface output routine. Set to one if the link level source
286 address will be written, as provided, to the wire. This flag is initialized
291 Set or get the flag determining whether locally generated packets on the
292 interface should be returned by BPF. Set to zero to see only incoming
293 packets on the interface. Set to one to see packets originating
294 locally and remotely on the interface. This flag is initialized to one by
298 The following structure is prepended to each packet returned by
302 struct timeval bh_tstamp; /* time stamp */
303 u_long bh_caplen; /* length of captured portion */
304 u_long bh_datalen; /* original length of packet */
305 u_short bh_hdrlen; /* length of bpf header (this struct
306 plus alignment padding */
310 The fields, whose values are stored in host order, and are:
312 .Bl -tag -compact -width bh_datalen
314 The time at which the packet was processed by the packet filter.
316 The length of the captured portion of the packet. This is the minimum of
317 the truncation amount specified by the filter and the length of the packet.
319 The length of the packet off the wire.
320 This value is independent of the truncation amount specified by the filter.
324 header, which may not be equal to
325 .\" XXX - not really a function call
326 .Fn sizeof "struct bpf_hdr" .
331 field exists to account for
332 padding between the header and the link level protocol.
333 The purpose here is to guarantee proper alignment of the packet
334 data structures, which is required on alignment sensitive
335 architectures and improves performance on many other architectures.
336 The packet filter insures that the
338 and the network layer
339 header will be word aligned. Suitable precautions
340 must be taken when accessing the link layer protocol fields on alignment
341 restricted machines. (This isn't a problem on an Ethernet, since
342 the type field is a short falling on an even offset,
343 and the addresses are probably accessed in a bytewise fashion).
345 Additionally, individual packets are padded so that each starts
346 on a word boundary. This requires that an application
347 has some knowledge of how to get from packet to packet.
353 this process. It rounds up its argument
354 to the nearest word aligned value (where a word is
360 points to the start of a packet, this expression
361 will advance it to the next packet:
362 .Dl p = (char *)p + BPF_WORDALIGN(p->bh_hdrlen + p->bh_caplen)
364 For the alignment mechanisms to work properly, the
367 must itself be word aligned.
371 will always return an aligned buffer.
373 A filter program is an array of instructions, with all branches forwardly
374 directed, terminated by a
377 Each instruction performs some action on the pseudo-machine state,
378 which consists of an accumulator, index register, scratch memory store,
379 and implicit program counter.
381 The following structure defines the instruction format:
393 field is used in different ways by different instructions,
398 fields are used as offsets
399 by the branch instructions.
400 The opcodes are encoded in a semi-hierarchical fashion.
401 There are eight classes of instructions:
411 Various other mode and
412 operator bits are or'd into the class to give the actual instructions.
413 The classes and modes are defined in
416 Below are the semantics for each defined
419 We use the convention that A is the accumulator, X is the index register,
420 P[] packet data, and M[] scratch memory store.
421 P[i:n] gives the data at byte offset
424 interpreted as a word (n=4),
425 unsigned halfword (n=2), or unsigned byte (n=1).
426 M[i] gives the i'th word in the scratch memory store, which is only
427 addressed in word units. The memory store is indexed from 0 to
434 are the corresponding fields in the
435 instruction definition.
437 refers to the length of the packet.
439 .Bl -tag -width BPF_STXx
441 These instructions copy a value into the accumulator. The type of the
442 source operand is specified by an
444 and can be a constant
446 packet data at a fixed offset
448 packet data at a variable offset
452 or a word in the scratch memory store
458 the data size must be specified as a word
464 The semantics of all the recognized
468 .Bl -tag -width "BPF_LD+BPF_W+BPF_IND" -compact
469 .It Li BPF_LD+BPF_W+BPF_ABS
471 .It Li BPF_LD+BPF_H+BPF_ABS
473 .It Li BPF_LD+BPF_B+BPF_ABS
475 .It Li BPF_LD+BPF_W+BPF_IND
477 .It Li BPF_LD+BPF_H+BPF_IND
479 .It Li BPF_LD+BPF_B+BPF_IND
481 .It Li BPF_LD+BPF_W+BPF_LEN
483 .It Li BPF_LD+BPF_IMM
485 .It Li BPF_LD+BPF_MEM
489 These instructions load a value into the index register. Note that
490 the addressing modes are more restrictive than those of the accumulator loads,
493 a hack for efficiently loading the IP header length.
495 .Bl -tag -width "BPF_LDX+BPF_W+BPF_MEM" -compact
496 .It Li BPF_LDX+BPF_W+BPF_IMM
498 .It Li BPF_LDX+BPF_W+BPF_MEM
500 .It Li BPF_LDX+BPF_W+BPF_LEN
502 .It Li BPF_LDX+BPF_B+BPF_MSH
506 This instruction stores the accumulator into the scratch memory.
507 We do not need an addressing mode since there is only one possibility
510 .Bl -tag -width "BPF_ST" -compact
515 This instruction stores the index register in the scratch memory store.
517 .Bl -tag -width "BPF_STX" -compact
522 The alu instructions perform operations between the accumulator and
523 index register or constant, and store the result back in the accumulator.
524 For binary operations, a source mode is required
529 .Bl -tag -width "BPF_ALU+BPF_MUL+BPF_K" -compact
530 .It Li BPF_ALU+BPF_ADD+BPF_K
532 .It Li BPF_ALU+BPF_SUB+BPF_K
534 .It Li BPF_ALU+BPF_MUL+BPF_K
536 .It Li BPF_ALU+BPF_DIV+BPF_K
538 .It Li BPF_ALU+BPF_AND+BPF_K
540 .It Li BPF_ALU+BPF_OR+BPF_K
542 .It Li BPF_ALU+BPF_LSH+BPF_K
544 .It Li BPF_ALU+BPF_RSH+BPF_K
546 .It Li BPF_ALU+BPF_ADD+BPF_X
548 .It Li BPF_ALU+BPF_SUB+BPF_X
550 .It Li BPF_ALU+BPF_MUL+BPF_X
552 .It Li BPF_ALU+BPF_DIV+BPF_X
554 .It Li BPF_ALU+BPF_AND+BPF_X
556 .It Li BPF_ALU+BPF_OR+BPF_X
558 .It Li BPF_ALU+BPF_LSH+BPF_X
560 .It Li BPF_ALU+BPF_RSH+BPF_X
562 .It Li BPF_ALU+BPF_NEG
566 The jump instructions alter flow of control. Conditional jumps
567 compare the accumulator against a constant
569 or the index register
571 If the result is true (or non-zero),
572 the true branch is taken, otherwise the false branch is taken.
573 Jump offsets are encoded in 8 bits so the longest jump is 256 instructions.
574 However, the jump always
576 opcode uses the 32 bit
578 field as the offset, allowing arbitrarily distant destinations.
579 All conditionals use unsigned comparison conventions.
581 .Bl -tag -width "BPF_JMP+BPF_KSET+BPF_X" -compact
582 .It Li BPF_JMP+BPF_JA
584 .It Li BPF_JMP+BPF_JGT+BPF_K
585 pc += (A > k) ? jt : jf
586 .It Li BPF_JMP+BPF_JGE+BPF_K
587 pc += (A >= k) ? jt : jf
588 .It Li BPF_JMP+BPF_JEQ+BPF_K
589 pc += (A == k) ? jt : jf
590 .It Li BPF_JMP+BPF_JSET+BPF_K
591 pc += (A & k) ? jt : jf
592 .It Li BPF_JMP+BPF_JGT+BPF_X
593 pc += (A > X) ? jt : jf
594 .It Li BPF_JMP+BPF_JGE+BPF_X
595 pc += (A >= X) ? jt : jf
596 .It Li BPF_JMP+BPF_JEQ+BPF_X
597 pc += (A == X) ? jt : jf
598 .It Li BPF_JMP+BPF_JSET+BPF_X
599 pc += (A & X) ? jt : jf
602 The return instructions terminate the filter program and specify the amount
603 of packet to accept (i.e., they return the truncation amount). A return
604 value of zero indicates that the packet should be ignored.
605 The return value is either a constant
610 .Bl -tag -width "BPF_RET+BPF_K" -compact
617 The miscellaneous category was created for anything that doesn't
618 fit into the above classes, and for any new instructions that might need to
619 be added. Currently, these are the register transfer instructions
620 that copy the index register to the accumulator or vice versa.
622 .Bl -tag -width "BPF_MISC+BPF_TAX" -compact
623 .It Li BPF_MISC+BPF_TAX
625 .It Li BPF_MISC+BPF_TXA
632 interface provides the following macros to facilitate
634 .Fn BPF_STMT opcode operand
636 .Fn BPF_JUMP opcode operand true_offset false_offset .
638 The following filter is taken from the Reverse ARP Daemon. It accepts
639 only Reverse ARP requests.
641 struct bpf_insn insns[] = {
642 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
643 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_REVARP, 0, 3),
644 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
645 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, REVARP_REQUEST, 0, 1),
646 BPF_STMT(BPF_RET+BPF_K, sizeof(struct ether_arp) +
647 sizeof(struct ether_header)),
648 BPF_STMT(BPF_RET+BPF_K, 0),
652 This filter accepts only IP packets between host 128.3.112.15 and
655 struct bpf_insn insns[] = {
656 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
657 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 8),
658 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 26),
659 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 2),
660 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
661 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 3, 4),
662 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 0, 3),
663 BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
664 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 1),
665 BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
666 BPF_STMT(BPF_RET+BPF_K, 0),
670 Finally, this filter returns only TCP finger packets. We must parse
671 the IP header to reach the TCP header. The
674 checks that the IP fragment offset is 0 so we are sure
675 that we have a TCP header.
677 struct bpf_insn insns[] = {
678 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
679 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 10),
680 BPF_STMT(BPF_LD+BPF_B+BPF_ABS, 23),
681 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, IPPROTO_TCP, 0, 8),
682 BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
683 BPF_JUMP(BPF_JMP+BPF_JSET+BPF_K, 0x1fff, 6, 0),
684 BPF_STMT(BPF_LDX+BPF_B+BPF_MSH, 14),
685 BPF_STMT(BPF_LD+BPF_H+BPF_IND, 14),
686 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 2, 0),
687 BPF_STMT(BPF_LD+BPF_H+BPF_IND, 16),
688 BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 0, 1),
689 BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
690 BPF_STMT(BPF_RET+BPF_K, 0),
701 .%T "An efficient, extensible, and portable network monitor"
704 .Bl -tag -compact -width /dev/bpfXXX
705 .It Pa /dev/bpf Ns Sy n
706 the packet filter device
709 The read buffer must be of a fixed size (returned by the
713 A file that does not request promiscuous mode may receive promiscuously
714 received packets as a side effect of another file requesting this
715 mode on the same hardware interface. This could be fixed in the kernel
716 with additional processing overhead. However, we favor the model where
717 all files must assume that the interface is promiscuous, and if
718 so desired, must utilize a filter to reject foreign packets.
720 Data link protocols with variable length headers are not currently supported.
722 The Enet packet filter was created in 1980 by Mike Accetta and
723 Rick Rashid at Carnegie-Mellon University. Jeffrey Mogul, at
724 Stanford, ported the code to
726 and continued its development from
727 1983 on. Since then, it has evolved into the Ultrix Packet Filter
740 of Lawrence Berkeley Laboratory, implemented BPF in
741 Summer 1990. Much of the design is due to