xref: /freebsd/share/man/man4/bpf.4 (revision 1b6c76a2fe091c74f08427e6c870851025a9cf67)
1.\" Copyright (c) 1990 The Regents of the University of California.
2.\" All rights reserved.
3.\"
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.
19.\"
20.\" This document is derived in part from the enet man page (enet.4)
21.\" distributed with 4.3BSD Unix.
22.\"
23.\" $FreeBSD$
24.\"
25.Dd January 16, 1996
26.Dt BPF 4
27.Os BSD 4.4
28.Sh NAME
29.Nm bpf
30.Nd Berkeley Packet Filter
31.Sh SYNOPSIS
32.Cd device bpf
33.Sh DESCRIPTION
34The Berkeley Packet Filter
35provides a raw interface to data link layers in a protocol
36independent fashion.
37All packets on the network, even those destined for other hosts,
38are accessible through this mechanism.
39.Pp
40The packet filter appears as a character special device,
41.Pa /dev/bpf0 ,
42.Pa /dev/bpf1 ,
43etc.
44After opening the device, the file descriptor must be bound to a
45specific network interface with the
46.Dv BIOCSETIF
47ioctl.
48A given interface can be shared be multiple listeners, and the filter
49underlying each descriptor will see an identical packet stream.
50The total number of open
51files is limited to the value given in the kernel configuration; the
52example given in the
53.Sx SYNOPSIS
54above sets the limit to 16.
55.Pp
56A separate device file is required for each minor device.
57If a file is in use, the open will fail and
58.Va errno
59will be set to
60.Er EBUSY .
61.Pp
62Associated with each open instance of a
63.Nm
64file is a user-settable packet filter.
65Whenever a packet is received by an interface,
66all file descriptors listening on that interface apply their filter.
67Each descriptor that accepts the packet receives its own copy.
68.Pp
69Reads from these files return the next group of packets
70that have matched the filter.
71To improve performance, the buffer passed to read must be
72the same size as the buffers used internally by
73.Nm .
74This size is returned by the
75.Dv BIOCGBLEN
76ioctl (see below), and
77can be set with
78.Dv BIOCSBLEN .
79Note that an individual packet larger than this size is necessarily
80truncated.
81.Pp
82The packet filter will support any link level protocol that has fixed length
83headers.  Currently, only Ethernet,
84.Tn SLIP ,
85and
86.Tn PPP
87drivers have been modified to interact with
88.Nm .
89.Pp
90Since packet data is in network byte order, applications should use the
91.Xr byteorder 3
92macros to extract multi-byte values.
93.Pp
94A packet can be sent out on the network by writing to a
95.Nm
96file descriptor.  The writes are unbuffered, meaning only one
97packet can be processed per write.
98Currently, only writes to Ethernets and
99.Tn SLIP
100links are supported.
101.Sh IOCTLS
102The
103.Xr ioctl 2
104command codes below are defined in
105.Aq Pa net/bpf.h .
106All commands require
107these includes:
108.Bd -literal
109	#include <sys/types.h>
110	#include <sys/time.h>
111	#include <sys/ioctl.h>
112	#include <net/bpf.h>
113.Ed
114.Pp
115Additionally,
116.Dv BIOCGETIF
117and
118.Dv BIOCSETIF
119require
120.Aq Pa sys/socket.h
121and
122.Aq Pa net/if.h .
123.Pp
124In addition to
125.Dv FIONREAD
126and
127.Dv SIOCGIFADDR ,
128the following commands may be applied to any open
129.Nm
130file.
131The (third) argument to
132.Xr ioctl 2
133should be a pointer to the type indicated.
134.Bl -tag -width BIOCGRTIMEOUT
135.It Dv BIOCGBLEN
136.Pq Li u_int
137Returns the required buffer length for reads on
138.Nm
139files.
140.It Dv BIOCSBLEN
141.Pq Li u_int
142Sets the buffer length for reads on
143.Nm
144files.  The buffer must be set before the file is attached to an interface
145with
146.Dv BIOCSETIF .
147If the requested buffer size cannot be accommodated, the closest
148allowable size will be set and returned in the argument.
149A read call will result in
150.Er EIO
151if it is passed a buffer that is not this size.
152.It Dv BIOCGDLT
153.Pq Li u_int
154Returns the type of the data link layer underlying the attached interface.
155.Er EINVAL
156is returned if no interface has been specified.
157The device types, prefixed with
158.Dq Li DLT_ ,
159are defined in
160.Aq Pa net/bpf.h .
161.It Dv BIOCPROMISC
162Forces the interface into promiscuous mode.
163All packets, not just those destined for the local host, are processed.
164Since more than one file can be listening on a given interface,
165a listener that opened its interface non-promiscuously may receive
166packets promiscuously.  This problem can be remedied with an
167appropriate filter.
168.It Dv BIOCFLUSH
169Flushes the buffer of incoming packets,
170and resets the statistics that are returned by BIOCGSTATS.
171.It Dv BIOCGETIF
172.Pq Li "struct ifreq"
173Returns the name of the hardware interface that the file is listening on.
174The name is returned in the ifr_name field of
175the
176.Li ifreq
177structure.
178All other fields are undefined.
179.It Dv BIOCSETIF
180.Pq Li "struct ifreq"
181Sets the hardware interface associate with the file.  This
182command must be performed before any packets can be read.
183The device is indicated by name using the
184.Li ifr_name
185field of the
186.Li ifreq
187structure.
188Additionally, performs the actions of
189.Dv BIOCFLUSH .
190.It Dv BIOCSRTIMEOUT
191.It Dv BIOCGRTIMEOUT
192.Pq Li "struct timeval"
193Set or get the read timeout parameter.
194The argument
195specifies the length of time to wait before timing
196out on a read request.
197This parameter is initialized to zero by
198.Xr open 2 ,
199indicating no timeout.
200.It Dv BIOCGSTATS
201.Pq Li "struct bpf_stat"
202Returns the following structure of packet statistics:
203.Bd -literal
204struct bpf_stat {
205	u_int bs_recv;    /* number of packets received */
206	u_int bs_drop;    /* number of packets dropped */
207};
208.Ed
209.Pp
210The fields are:
211.Bl -hang -offset indent
212.It Li bs_recv
213the number of packets received by the descriptor since opened or reset
214(including any buffered since the last read call);
215and
216.It Li bs_drop
217the number of packets which were accepted by the filter but dropped by the
218kernel because of buffer overflows
219(i.e., the application's reads aren't keeping up with the packet traffic).
220.El
221.It Dv BIOCIMMEDIATE
222.Pq Li u_int
223Enable or disable
224.Dq immediate mode ,
225based on the truth value of the argument.
226When immediate mode is enabled, reads return immediately upon packet
227reception.  Otherwise, a read will block until either the kernel buffer
228becomes full or a timeout occurs.
229This is useful for programs like
230.Xr rarpd 8
231which must respond to messages in real time.
232The default for a new file is off.
233.It Dv BIOCSETF
234.Pq Li "struct bpf_program"
235Sets the filter program used by the kernel to discard uninteresting
236packets.  An array of instructions and its length is passed in using
237the following structure:
238.Bd -literal
239struct bpf_program {
240	int bf_len;
241	struct bpf_insn *bf_insns;
242};
243.Ed
244.Pp
245The filter program is pointed to by the
246.Li bf_insns
247field while its length in units of
248.Sq Li struct bpf_insn
249is given by the
250.Li bf_len
251field.
252Also, the actions of
253.Dv BIOCFLUSH
254are performed.
255See section
256.Sx "FILTER MACHINE"
257for an explanation of the filter language.
258.It Dv BIOCVERSION
259.Pq Li "struct bpf_version"
260Returns the major and minor version numbers of the filter language currently
261recognized by the kernel.  Before installing a filter, applications must check
262that the current version is compatible with the running kernel.  Version
263numbers are compatible if the major numbers match and the application minor
264is less than or equal to the kernel minor.  The kernel version number is
265returned in the following structure:
266.Bd -literal
267struct bpf_version {
268        u_short bv_major;
269        u_short bv_minor;
270};
271.Ed
272.Pp
273The current version numbers are given by
274.Dv BPF_MAJOR_VERSION
275and
276.Dv BPF_MINOR_VERSION
277from
278.Aq Pa net/bpf.h .
279An incompatible filter
280may result in undefined behavior (most likely, an error returned by
281.Fn ioctl
282or haphazard packet matching).
283.It Dv BIOCSHDRCMPLT
284.It Dv BIOCGHDRCMPLT
285.Pq Li u_int
286Set or get the status of the
287.Dq header complete
288flag.
289Set to zero if the link level source address should be filled in automatically
290by the interface output routine.  Set to one if the link level source
291address will be written, as provided, to the wire.  This flag is initialized
292to zero by default.
293.It Dv BIOCSSEESENT
294.It Dv BIOCGSEESENT
295.Pq Li u_int
296Set or get the flag determining whether locally generated packets on the
297interface should be returned by BPF.  Set to zero to see only incoming
298packets on the interface.  Set to one to see packets originating
299locally and remotely on the interface.  This flag is initialized to one by
300default.
301.El
302.Sh BPF HEADER
303The following structure is prepended to each packet returned by
304.Xr read 2 :
305.Bd -literal
306struct bpf_hdr {
307        struct timeval bh_tstamp;     /* time stamp */
308        u_long bh_caplen;             /* length of captured portion */
309        u_long bh_datalen;            /* original length of packet */
310        u_short bh_hdrlen;            /* length of bpf header (this struct
311					 plus alignment padding */
312};
313.Ed
314.Pp
315The fields, whose values are stored in host order, and are:
316.Pp
317.Bl -tag -compact -width bh_datalen
318.It Li bh_tstamp
319The time at which the packet was processed by the packet filter.
320.It Li bh_caplen
321The length of the captured portion of the packet.  This is the minimum of
322the truncation amount specified by the filter and the length of the packet.
323.It Li bh_datalen
324The length of the packet off the wire.
325This value is independent of the truncation amount specified by the filter.
326.It Li bh_hdrlen
327The length of the
328.Nm
329header, which may not be equal to
330.\" XXX - not really a function call
331.Fn sizeof "struct bpf_hdr" .
332.El
333.Pp
334The
335.Li bh_hdrlen
336field exists to account for
337padding between the header and the link level protocol.
338The purpose here is to guarantee proper alignment of the packet
339data structures, which is required on alignment sensitive
340architectures and improves performance on many other architectures.
341The packet filter insures that the
342.Li bpf_hdr
343and the network layer
344header will be word aligned.  Suitable precautions
345must be taken when accessing the link layer protocol fields on alignment
346restricted machines.  (This isn't a problem on an Ethernet, since
347the type field is a short falling on an even offset,
348and the addresses are probably accessed in a bytewise fashion).
349.Pp
350Additionally, individual packets are padded so that each starts
351on a word boundary.  This requires that an application
352has some knowledge of how to get from packet to packet.
353The macro
354.Dv BPF_WORDALIGN
355is defined in
356.Aq Pa net/bpf.h
357to facilitate
358this process.  It rounds up its argument
359to the nearest word aligned value (where a word is
360.Dv BPF_ALIGNMENT
361bytes wide).
362.Pp
363For example, if
364.Sq Li p
365points to the start of a packet, this expression
366will advance it to the next packet:
367.Dl p = (char *)p + BPF_WORDALIGN(p->bh_hdrlen + p->bh_caplen)
368.Pp
369For the alignment mechanisms to work properly, the
370buffer passed to
371.Xr read 2
372must itself be word aligned.
373The
374.Xr malloc 3
375function
376will always return an aligned buffer.
377.Sh FILTER MACHINE
378A filter program is an array of instructions, with all branches forwardly
379directed, terminated by a
380.Em return
381instruction.
382Each instruction performs some action on the pseudo-machine state,
383which consists of an accumulator, index register, scratch memory store,
384and implicit program counter.
385.Pp
386The following structure defines the instruction format:
387.Bd -literal
388struct bpf_insn {
389	u_short	code;
390	u_char 	jt;
391	u_char 	jf;
392	u_long k;
393};
394.Ed
395.Pp
396The
397.Li k
398field is used in different ways by different instructions,
399and the
400.Li jt
401and
402.Li jf
403fields are used as offsets
404by the branch instructions.
405The opcodes are encoded in a semi-hierarchical fashion.
406There are eight classes of instructions:
407.Dv BPF_LD ,
408.Dv BPF_LDX ,
409.Dv BPF_ST ,
410.Dv BPF_STX ,
411.Dv BPF_ALU ,
412.Dv BPF_JMP ,
413.Dv BPF_RET ,
414and
415.Dv BPF_MISC .
416Various other mode and
417operator bits are or'd into the class to give the actual instructions.
418The classes and modes are defined in
419.Aq Pa net/bpf.h .
420.Pp
421Below are the semantics for each defined
422.Nm
423instruction.
424We use the convention that A is the accumulator, X is the index register,
425P[] packet data, and M[] scratch memory store.
426P[i:n] gives the data at byte offset
427.Dq i
428in the packet,
429interpreted as a word (n=4),
430unsigned halfword (n=2), or unsigned byte (n=1).
431M[i] gives the i'th word in the scratch memory store, which is only
432addressed in word units.  The memory store is indexed from 0 to
433.Dv BPF_MEMWORDS
434- 1.
435.Li k ,
436.Li jt ,
437and
438.Li jf
439are the corresponding fields in the
440instruction definition.
441.Dq len
442refers to the length of the packet.
443.Pp
444.Bl -tag -width BPF_STXx
445.It Dv BPF_LD
446These instructions copy a value into the accumulator.  The type of the
447source operand is specified by an
448.Dq addressing mode
449and can be a constant
450.Pq Dv BPF_IMM ,
451packet data at a fixed offset
452.Pq Dv BPF_ABS ,
453packet data at a variable offset
454.Pq Dv BPF_IND ,
455the packet length
456.Pq Dv BPF_LEN ,
457or a word in the scratch memory store
458.Pq Dv BPF_MEM .
459For
460.Dv BPF_IND
461and
462.Dv BPF_ABS ,
463the data size must be specified as a word
464.Pq Dv BPF_W ,
465halfword
466.Pq Dv BPF_H ,
467or byte
468.Pq Dv BPF_B .
469The semantics of all the recognized
470.Dv BPF_LD
471instructions follow.
472.Pp
473.Bl -tag -width "BPF_LD+BPF_W+BPF_IND" -compact
474.It Li BPF_LD+BPF_W+BPF_ABS
475A <- P[k:4]
476.It Li BPF_LD+BPF_H+BPF_ABS
477A <- P[k:2]
478.It Li BPF_LD+BPF_B+BPF_ABS
479A <- P[k:1]
480.It Li BPF_LD+BPF_W+BPF_IND
481A <- P[X+k:4]
482.It Li BPF_LD+BPF_H+BPF_IND
483A <- P[X+k:2]
484.It Li BPF_LD+BPF_B+BPF_IND
485A <- P[X+k:1]
486.It Li BPF_LD+BPF_W+BPF_LEN
487A <- len
488.It Li BPF_LD+BPF_IMM
489A <- k
490.It Li BPF_LD+BPF_MEM
491A <- M[k]
492.El
493.It Dv BPF_LDX
494These instructions load a value into the index register.  Note that
495the addressing modes are more restrictive than those of the accumulator loads,
496but they include
497.Dv BPF_MSH ,
498a hack for efficiently loading the IP header length.
499.Pp
500.Bl -tag -width "BPF_LDX+BPF_W+BPF_MEM" -compact
501.It Li BPF_LDX+BPF_W+BPF_IMM
502X <- k
503.It Li BPF_LDX+BPF_W+BPF_MEM
504X <- M[k]
505.It Li BPF_LDX+BPF_W+BPF_LEN
506X <- len
507.It Li BPF_LDX+BPF_B+BPF_MSH
508X <- 4*(P[k:1]&0xf)
509.El
510.It Dv BPF_ST
511This instruction stores the accumulator into the scratch memory.
512We do not need an addressing mode since there is only one possibility
513for the destination.
514.Pp
515.Bl -tag -width "BPF_ST" -compact
516.It Li BPF_ST
517M[k] <- A
518.El
519.It Dv BPF_STX
520This instruction stores the index register in the scratch memory store.
521.Pp
522.Bl -tag -width "BPF_STX" -compact
523.It Li BPF_STX
524M[k] <- X
525.El
526.It Dv BPF_ALU
527The alu instructions perform operations between the accumulator and
528index register or constant, and store the result back in the accumulator.
529For binary operations, a source mode is required
530.Po
531.Dv BPF_K
532or
533.Dv BPF_X
534.Pc .
535.Pp
536.Bl -tag -width "BPF_ALU+BPF_MUL+BPF_K" -compact
537.It Li BPF_ALU+BPF_ADD+BPF_K
538A <- A + k
539.It Li BPF_ALU+BPF_SUB+BPF_K
540A <- A - k
541.It Li BPF_ALU+BPF_MUL+BPF_K
542A <- A * k
543.It Li BPF_ALU+BPF_DIV+BPF_K
544A <- A / k
545.It Li BPF_ALU+BPF_AND+BPF_K
546A <- A & k
547.It Li BPF_ALU+BPF_OR+BPF_K
548A <- A | k
549.It Li BPF_ALU+BPF_LSH+BPF_K
550A <- A << k
551.It Li BPF_ALU+BPF_RSH+BPF_K
552A <- A >> k
553.It Li BPF_ALU+BPF_ADD+BPF_X
554A <- A + X
555.It Li BPF_ALU+BPF_SUB+BPF_X
556A <- A - X
557.It Li BPF_ALU+BPF_MUL+BPF_X
558A <- A * X
559.It Li BPF_ALU+BPF_DIV+BPF_X
560A <- A / X
561.It Li BPF_ALU+BPF_AND+BPF_X
562A <- A & X
563.It Li BPF_ALU+BPF_OR+BPF_X
564A <- A | X
565.It Li BPF_ALU+BPF_LSH+BPF_X
566A <- A << X
567.It Li BPF_ALU+BPF_RSH+BPF_X
568A <- A >> X
569.It Li BPF_ALU+BPF_NEG
570A <- -A
571.El
572.It Dv BPF_JMP
573The jump instructions alter flow of control.  Conditional jumps
574compare the accumulator against a constant
575.Pq Dv BPF_K
576or the index register
577.Pq Dv BPF_X .
578If the result is true (or non-zero),
579the true branch is taken, otherwise the false branch is taken.
580Jump offsets are encoded in 8 bits so the longest jump is 256 instructions.
581However, the jump always
582.Pq Dv BPF_JA
583opcode uses the 32 bit
584.Li k
585field as the offset, allowing arbitrarily distant destinations.
586All conditionals use unsigned comparison conventions.
587.Pp
588.Bl -tag -width "BPF_JMP+BPF_KSET+BPF_X" -compact
589.It Li BPF_JMP+BPF_JA
590pc += k
591.It Li BPF_JMP+BPF_JGT+BPF_K
592pc += (A > k) ? jt : jf
593.It Li BPF_JMP+BPF_JGE+BPF_K
594pc += (A >= k) ? jt : jf
595.It Li BPF_JMP+BPF_JEQ+BPF_K
596pc += (A == k) ? jt : jf
597.It Li BPF_JMP+BPF_JSET+BPF_K
598pc += (A & k) ? jt : jf
599.It Li BPF_JMP+BPF_JGT+BPF_X
600pc += (A > X) ? jt : jf
601.It Li BPF_JMP+BPF_JGE+BPF_X
602pc += (A >= X) ? jt : jf
603.It Li BPF_JMP+BPF_JEQ+BPF_X
604pc += (A == X) ? jt : jf
605.It Li BPF_JMP+BPF_JSET+BPF_X
606pc += (A & X) ? jt : jf
607.El
608.It Dv BPF_RET
609The return instructions terminate the filter program and specify the amount
610of packet to accept (i.e., they return the truncation amount).  A return
611value of zero indicates that the packet should be ignored.
612The return value is either a constant
613.Pq Dv BPF_K
614or the accumulator
615.Pq Dv BPF_A .
616.Pp
617.Bl -tag -width "BPF_RET+BPF_K" -compact
618.It Li BPF_RET+BPF_A
619accept A bytes
620.It Li BPF_RET+BPF_K
621accept k bytes
622.El
623.It Dv BPF_MISC
624The miscellaneous category was created for anything that doesn't
625fit into the above classes, and for any new instructions that might need to
626be added.  Currently, these are the register transfer instructions
627that copy the index register to the accumulator or vice versa.
628.Pp
629.Bl -tag -width "BPF_MISC+BPF_TAX" -compact
630.It Li BPF_MISC+BPF_TAX
631X <- A
632.It Li BPF_MISC+BPF_TXA
633A <- X
634.El
635.El
636.Pp
637The
638.Nm
639interface provides the following macros to facilitate
640array initializers:
641.Fn BPF_STMT opcode operand
642and
643.Fn BPF_JUMP opcode operand true_offset false_offset .
644.Pp
645.Sh EXAMPLES
646The following filter is taken from the Reverse ARP Daemon.  It accepts
647only Reverse ARP requests.
648.Bd -literal
649struct bpf_insn insns[] = {
650	BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
651	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_REVARP, 0, 3),
652	BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
653	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, REVARP_REQUEST, 0, 1),
654	BPF_STMT(BPF_RET+BPF_K, sizeof(struct ether_arp) +
655		 sizeof(struct ether_header)),
656	BPF_STMT(BPF_RET+BPF_K, 0),
657};
658.Ed
659.Pp
660This filter accepts only IP packets between host 128.3.112.15 and
661128.3.112.35.
662.Bd -literal
663struct bpf_insn insns[] = {
664	BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
665	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 8),
666	BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 26),
667	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 2),
668	BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
669	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 3, 4),
670	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 0, 3),
671	BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
672	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 1),
673	BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
674	BPF_STMT(BPF_RET+BPF_K, 0),
675};
676.Ed
677.Pp
678Finally, this filter returns only TCP finger packets.  We must parse
679the IP header to reach the TCP header.  The
680.Dv BPF_JSET
681instruction
682checks that the IP fragment offset is 0 so we are sure
683that we have a TCP header.
684.Bd -literal
685struct 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, 10),
688	BPF_STMT(BPF_LD+BPF_B+BPF_ABS, 23),
689	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, IPPROTO_TCP, 0, 8),
690	BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
691	BPF_JUMP(BPF_JMP+BPF_JSET+BPF_K, 0x1fff, 6, 0),
692	BPF_STMT(BPF_LDX+BPF_B+BPF_MSH, 14),
693	BPF_STMT(BPF_LD+BPF_H+BPF_IND, 14),
694	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 2, 0),
695	BPF_STMT(BPF_LD+BPF_H+BPF_IND, 16),
696	BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 0, 1),
697	BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
698	BPF_STMT(BPF_RET+BPF_K, 0),
699};
700.Ed
701.Sh SEE ALSO
702.Xr tcpdump 1 ,
703.Xr ioctl 2 ,
704.Xr byteorder 3 ,
705.Xr ng_bpf 4
706.Rs
707.%A McCanne, S.
708.%A Jacobson V.
709.%T "An efficient, extensible, and portable network monitor"
710.Re
711.Sh FILES
712.Bl -tag -compact -width /dev/bpfXXX
713.It Pa /dev/bpf Ns Sy n
714the packet filter device
715.El
716.Sh BUGS
717The read buffer must be of a fixed size (returned by the
718.Dv BIOCGBLEN
719ioctl).
720.Pp
721A file that does not request promiscuous mode may receive promiscuously
722received packets as a side effect of another file requesting this
723mode on the same hardware interface.  This could be fixed in the kernel
724with additional processing overhead.  However, we favor the model where
725all files must assume that the interface is promiscuous, and if
726so desired, must utilize a filter to reject foreign packets.
727.Pp
728Data link protocols with variable length headers are not currently supported.
729.Sh HISTORY
730.Pp
731The Enet packet filter was created in 1980 by Mike Accetta and
732Rick Rashid at Carnegie-Mellon University.  Jeffrey Mogul, at
733Stanford, ported the code to BSD and continued its development from
7341983 on.  Since then, it has evolved into the Ultrix Packet Filter
735at
736.Tn DEC ,
737a
738.Tn STREAMS
739.Tn NIT
740module under
741.Tn SunOS 4.1 ,
742and
743.Tn BPF .
744.Sh AUTHORS
745.An -nosplit
746.An Steven McCanne ,
747of Lawrence Berkeley Laboratory, implemented BPF in
748Summer 1990.  Much of the design is due to
749.An Van Jacobson .
750