bsd/man/man4/bpf.4

   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: src/share/man/man4/bpf.4,v 1.21.2.10 2001/08/17 13:08:37 ru Exp $
  24 .\"
  25 .Dd January 16, 1996
  26 .Dt BPF 4
  27 .Os
  28 .Sh NAME
  29 .Nm bpf
  30 .Nd Berkeley Packet Filter
  31 .Sh SYNOPSIS
  32 .Cd pseudo-device bpf
  33 .Sh DESCRIPTION
  34 The Berkeley Packet Filter
  35 provides a raw interface to data link layers in a protocol
  36 independent fashion.
  37 All packets on the network, even those destined for other hosts,
  38 are accessible through this mechanism.
  39 .Pp
  40 The packet filter appears as a character special device,
  41 .Pa /dev/bpf0 ,
  42 .Pa /dev/bpf1 ,
  43 etc.
  44 After opening the device, the file descriptor must be bound to a
  45 specific network interface with the
  46 .Dv BIOCSETIF
  47 ioctl.
  48 A given interface can be shared be multiple listeners, and the filter
  49 underlying each descriptor will see an identical packet stream.
  50 .Pp
  51 A separate device file is required for each minor device.
  52 If a file is in use, the open will fail and
  53 .Va errno
  54 will be set to
  55 .Er EBUSY .
  56 .Pp
  57 Associated with each open instance of a
  58 .Nm
  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.
  63 .Pp
  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
  68 .Nm .
  69 This size is returned by the
  70 .Dv BIOCGBLEN
  71 ioctl (see below), and
  72 can be set with
  73 .Dv BIOCSBLEN .
  74 Note that an individual packet larger than this size is necessarily
  75 truncated.
  76 .Pp
  77 The packet filter will support any link level protocol that has fixed length
  78 headers.  Currently, only Ethernet,
  79 .Tn SLIP ,
  80 and
  81 .Tn PPP
  82 drivers have been modified to interact with
  83 .Nm .
  84 .Pp
  85 Since packet data is in network byte order, applications should use the
  86 .Xr byteorder 3
  87 macros to extract multi-byte values.
  88 .Pp
  89 A packet can be sent out on the network by writing to a
  90 .Nm
  91 file descriptor.  The writes are unbuffered, meaning only one
  92 packet can be processed per write.
  93 Currently, only writes to Ethernets and
  94 .Tn SLIP
  95 links are supported.
  96 .Sh IOCTLS
  97 The
  98 .Xr ioctl 2
  99 command codes below are defined in
 100 .Aq Pa net/bpf.h .
 101 All commands require
 102 these includes:
 103 .Bd -literal
 104         #include <sys/types.h>
 105         #include <sys/time.h>
 106         #include <sys/ioctl.h>
 107         #include <net/bpf.h>
 108 .Ed
 109 .Pp
 110 Additionally,
 111 .Dv BIOCGETIF
 112 and
 113 .Dv BIOCSETIF
 114 require
 115 .Aq Pa sys/socket.h
 116 and
 117 .Aq Pa net/if.h .
 118 .Pp
 119 In addition to
 120 .Dv FIONREAD
 121 the following commands may be applied to any open
 122 .Nm
 123 file.
 124 The (third) argument to
 125 .Xr ioctl 2
 126 should be a pointer to the type indicated.
 127 .Bl -tag -width BIOCGRTIMEOUT
 128 .It Dv BIOCGBLEN
 129 .Pq Li u_int
 130 Returns the required buffer length for reads on
 131 .Nm
 132 files.
 133 .It Dv BIOCSBLEN
 134 .Pq Li u_int
 135 Sets the buffer length for reads on
 136 .Nm
 137 files.  The buffer must be set before the file is attached to an interface
 138 with
 139 .Dv BIOCSETIF .
 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
 143 .Er EIO
 144 if it is passed a buffer that is not this size.
 145 .It Dv BIOCGDLT
 146 .Pq Li u_int
 147 Returns the type of the data link layer underlying the attached interface.
 148 .Er EINVAL
 149 is returned if no interface has been specified.
 150 The device types, prefixed with
 151 .Dq Li DLT_ ,
 152 are defined in
 153 .Aq Pa net/bpf.h .
 154 .It Dv BIOCPROMISC
 155 Forces the interface into promiscuous mode.
 156 All packets, not just those destined for the local host, are processed.
 157 Since more than one file can be listening on a given interface,
 158 a listener that opened its interface non-promiscuously may receive
 159 packets promiscuously.  This problem can be remedied with an
 160 appropriate filter.
 161 .It Dv BIOCFLUSH
 162 Flushes the buffer of incoming packets,
 163 and resets the statistics that are returned by BIOCGSTATS.
 164 .It Dv BIOCGETIF
 165 .Pq Li "struct ifreq"
 166 Returns the name of the hardware interface that the file is listening on.
 167 The name is returned in the ifr_name field of
 168 the
 169 .Li ifreq
 170 structure.
 171 All other fields are undefined.
 172 .It Dv BIOCSETIF
 173 .Pq Li "struct ifreq"
 174 Sets the hardware interface associate with the file.  This
 175 command must be performed before any packets can be read.
 176 The device is indicated by name using the
 177 .Li ifr_name
 178 field of the
 179 .Li ifreq
 180 structure.
 181 Additionally, performs the actions of
 182 .Dv BIOCFLUSH .
 183 .It Dv BIOCSRTIMEOUT
 184 .It Dv BIOCGRTIMEOUT
 185 .Pq Li "struct timeval"
 186 Set or get the read timeout parameter.
 187 The argument
 188 specifies the length of time to wait before timing
 189 out on a read request.
 190 This parameter is initialized to zero by
 191 .Xr open 2 ,
 192 indicating no timeout.
 193 .It Dv BIOCGSTATS
 194 .Pq Li "struct bpf_stat"
 195 Returns the following structure of packet statistics:
 196 .Bd -literal
 197 struct bpf_stat {
 198         u_int bs_recv;    /* number of packets received */
 199         u_int bs_drop;    /* number of packets dropped */
 200 };
 201 .Ed
 202 .Pp
 203 The fields are:
 204 .Bl -hang -offset indent
 205 .It Li bs_recv
 206 the number of packets received by the descriptor since opened or reset
 207 (including any buffered since the last read call);
 208 and
 209 .It Li bs_drop
 210 the number of packets which were accepted by the filter but dropped by the
 211 kernel because of buffer overflows
 212 (i.e., the application's reads aren't keeping up with the packet traffic).
 213 .El
 214 .It Dv BIOCIMMEDIATE
 215 .Pq Li u_int
 216 Enable or disable
 217 .Dq immediate mode ,
 218 based on the truth value of the argument.
 219 When immediate mode is enabled, reads return immediately upon packet
 220 reception.  Otherwise, a read will block until either the kernel buffer
 221 becomes full or a timeout occurs.
 222 This is useful for programs like
 223 .Xr rarpd 8
 224 which must respond to messages in real time.
 225 The default for a new file is off.
 226 .It Dv BIOCSETF
 227 .Pq Li "struct bpf_program"
 228 Sets the filter program used by the kernel to discard uninteresting
 229 packets.  An array of instructions and its length is passed in using
 230 the following structure:
 231 .Bd -literal
 232 struct bpf_program {
 233         int bf_len;
 234         struct bpf_insn *bf_insns;
 235 };
 236 .Ed
 237 .Pp
 238 The filter program is pointed to by the
 239 .Li bf_insns
 240 field while its length in units of
 241 .Sq Li struct bpf_insn
 242 is given by the
 243 .Li bf_len
 244 field.
 245 Also, the actions of
 246 .Dv BIOCFLUSH
 247 are performed.
 248 See section
 249 .Sx "FILTER MACHINE"
 250 for an explanation of the filter language.
 251 .It Dv BIOCVERSION
 252 .Pq Li "struct bpf_version"
 253 Returns the major and minor version numbers of the filter language currently
 254 recognized by the kernel.  Before installing a filter, applications must check
 255 that the current version is compatible with the running kernel.  Version
 256 numbers are compatible if the major numbers match and the application minor
 257 is less than or equal to the kernel minor.  The kernel version number is
 258 returned in the following structure:
 259 .Bd -literal
 260 struct bpf_version {
 261         u_short bv_major;
 262         u_short bv_minor;
 263 };
 264 .Ed
 265 .Pp
 266 The current version numbers are given by
 267 .Dv BPF_MAJOR_VERSION
 268 and
 269 .Dv BPF_MINOR_VERSION
 270 from
 271 .Aq Pa net/bpf.h .
 272 An incompatible filter
 273 may result in undefined behavior (most likely, an error returned by
 274 .Fn ioctl
 275 or haphazard packet matching).
 276 .It Dv BIOCSHDRCMPLT
 277 .It Dv BIOCGHDRCMPLT
 278 .Pq Li u_int
 279 Set or get the status of the
 280 .Dq header complete
 281 flag.
 282 Set to zero if the link level source address should be filled in automatically
 283 by the interface output routine.  Set to one if the link level source
 284 address will be written, as provided, to the wire.  This flag is initialized
 285 to zero by default.
 286 .It Dv BIOCSSEESENT
 287 .It Dv BIOCGSEESENT
 288 .Pq Li u_int
 289 Set or get the flag determining whether locally generated packets on the
 290 interface should be returned by BPF.  Set to zero to see only incoming
 291 packets on the interface.  Set to one to see packets originating
 292 locally and remotely on the interface.  This flag is initialized to one by
 293 default.
 294 .El
 295 .Sh BPF HEADER
 296 The following structure is prepended to each packet returned by
 297 .Xr read 2 :
 298 .Bd -literal
 299 struct bpf_hdr {
 300         struct timeval bh_tstamp;     /* time stamp */
 301         u_long bh_caplen;             /* length of captured portion */
 302         u_long bh_datalen;            /* original length of packet */
 303         u_short bh_hdrlen;            /* length of bpf header (this struct
 304                                          plus alignment padding */
 305 };
 306 .Ed
 307 .Pp
 308 The fields, whose values are stored in host order, and are:
 309 .Pp
 310 .Bl -tag -compact -width bh_datalen
 311 .It Li bh_tstamp
 312 The time at which the packet was processed by the packet filter.
 313 .It Li bh_caplen
 314 The length of the captured portion of the packet.  This is the minimum of
 315 the truncation amount specified by the filter and the length of the packet.
 316 .It Li bh_datalen
 317 The length of the packet off the wire.
 318 This value is independent of the truncation amount specified by the filter.
 319 .It Li bh_hdrlen
 320 The length of the
 321 .Nm
 322 header, which may not be equal to
 323 .\" XXX - not really a function call
 324 .Fn sizeof "struct bpf_hdr" .
 325 .El
 326 .Pp
 327 The
 328 .Li bh_hdrlen
 329 field exists to account for
 330 padding between the header and the link level protocol.
 331 The purpose here is to guarantee proper alignment of the packet
 332 data structures, which is required on alignment sensitive
 333 architectures and improves performance on many other architectures.
 334 The packet filter insures that the
 335 .Li bpf_hdr
 336 and the network layer
 337 header will be word aligned.  Suitable precautions
 338 must be taken when accessing the link layer protocol fields on alignment
 339 restricted machines.  (This isn't a problem on an Ethernet, since
 340 the type field is a short falling on an even offset,
 341 and the addresses are probably accessed in a bytewise fashion).
 342 .Pp
 343 Additionally, individual packets are padded so that each starts
 344 on a word boundary.  This requires that an application
 345 has some knowledge of how to get from packet to packet.
 346 The macro
 347 .Dv BPF_WORDALIGN
 348 is defined in
 349 .Aq Pa net/bpf.h
 350 to facilitate
 351 this process.  It rounds up its argument
 352 to the nearest word aligned value (where a word is
 353 .Dv BPF_ALIGNMENT
 354 bytes wide).
 355 .Pp
 356 For example, if
 357 .Sq Li p
 358 points to the start of a packet, this expression
 359 will advance it to the next packet:
 360 .Dl p = (char *)p + BPF_WORDALIGN(p->bh_hdrlen + p->bh_caplen)
 361 .Pp
 362 For the alignment mechanisms to work properly, the
 363 buffer passed to
 364 .Xr read 2
 365 must itself be word aligned.
 366 The
 367 .Xr malloc 3
 368 function
 369 will always return an aligned buffer.
 370 .Sh FILTER MACHINE
 371 A filter program is an array of instructions, with all branches forwardly
 372 directed, terminated by a
 373 .Em return
 374 instruction.
 375 Each instruction performs some action on the pseudo-machine state,
 376 which consists of an accumulator, index register, scratch memory store,
 377 and implicit program counter.
 378 .Pp
 379 The following structure defines the instruction format:
 380 .Bd -literal
 381 struct bpf_insn {
 382         u_short code;
 383         u_char  jt;
 384         u_char  jf;
 385         u_long k;
 386 };
 387 .Ed
 388 .Pp
 389 The
 390 .Li k
 391 field is used in different ways by different instructions,
 392 and the
 393 .Li jt
 394 and
 395 .Li jf
 396 fields are used as offsets
 397 by the branch instructions.
 398 The opcodes are encoded in a semi-hierarchical fashion.
 399 There are eight classes of instructions:
 400 .Dv BPF_LD ,
 401 .Dv BPF_LDX ,
 402 .Dv BPF_ST ,
 403 .Dv BPF_STX ,
 404 .Dv BPF_ALU ,
 405 .Dv BPF_JMP ,
 406 .Dv BPF_RET ,
 407 and
 408 .Dv BPF_MISC .
 409 Various other mode and
 410 operator bits are or'd into the class to give the actual instructions.
 411 The classes and modes are defined in
 412 .Aq Pa net/bpf.h .
 413 .Pp
 414 Below are the semantics for each defined
 415 .Nm
 416 instruction.
 417 We use the convention that A is the accumulator, X is the index register,
 418 P[] packet data, and M[] scratch memory store.
 419 P[i:n] gives the data at byte offset
 420 .Dq i
 421 in the packet,
 422 interpreted as a word (n=4),
 423 unsigned halfword (n=2), or unsigned byte (n=1).
 424 M[i] gives the i'th word in the scratch memory store, which is only
 425 addressed in word units.  The memory store is indexed from 0 to
 426 .Dv BPF_MEMWORDS
 427 - 1.
 428 .Li k ,
 429 .Li jt ,
 430 and
 431 .Li jf
 432 are the corresponding fields in the
 433 instruction definition.
 434 .Dq len
 435 refers to the length of the packet.
 436 .Pp
 437 .Bl -tag -width BPF_STXx
 438 .It Dv BPF_LD
 439 These instructions copy a value into the accumulator.  The type of the
 440 source operand is specified by an
 441 .Dq addressing mode
 442 and can be a constant
 443 .Pq Dv BPF_IMM ,
 444 packet data at a fixed offset
 445 .Pq Dv BPF_ABS ,
 446 packet data at a variable offset
 447 .Pq Dv BPF_IND ,
 448 the packet length
 449 .Pq Dv BPF_LEN ,
 450 or a word in the scratch memory store
 451 .Pq Dv BPF_MEM .
 452 For
 453 .Dv BPF_IND
 454 and
 455 .Dv BPF_ABS ,
 456 the data size must be specified as a word
 457 .Pq Dv BPF_W ,
 458 halfword
 459 .Pq Dv BPF_H ,
 460 or byte
 461 .Pq Dv BPF_B .
 462 The semantics of all the recognized
 463 .Dv BPF_LD
 464 instructions follow.
 465 .Pp
 466 .Bl -tag -width "BPF_LD+BPF_W+BPF_IND" -compact
 467 .It Li BPF_LD+BPF_W+BPF_ABS
 468 A <- P[k:4]
 469 .It Li BPF_LD+BPF_H+BPF_ABS
 470 A <- P[k:2]
 471 .It Li BPF_LD+BPF_B+BPF_ABS
 472 A <- P[k:1]
 473 .It Li BPF_LD+BPF_W+BPF_IND
 474 A <- P[X+k:4]
 475 .It Li BPF_LD+BPF_H+BPF_IND
 476 A <- P[X+k:2]
 477 .It Li BPF_LD+BPF_B+BPF_IND
 478 A <- P[X+k:1]
 479 .It Li BPF_LD+BPF_W+BPF_LEN
 480 A <- len
 481 .It Li BPF_LD+BPF_IMM
 482 A <- k
 483 .It Li BPF_LD+BPF_MEM
 484 A <- M[k]
 485 .El
 486 .It Dv BPF_LDX
 487 These instructions load a value into the index register.  Note that
 488 the addressing modes are more restrictive than those of the accumulator loads,
 489 but they include
 490 .Dv BPF_MSH ,
 491 a hack for efficiently loading the IP header length.
 492 .Pp
 493 .Bl -tag -width "BPF_LDX+BPF_W+BPF_MEM" -compact
 494 .It Li BPF_LDX+BPF_W+BPF_IMM
 495 X <- k
 496 .It Li BPF_LDX+BPF_W+BPF_MEM
 497 X <- M[k]
 498 .It Li BPF_LDX+BPF_W+BPF_LEN
 499 X <- len
 500 .It Li BPF_LDX+BPF_B+BPF_MSH
 501 X <- 4*(P[k:1]&0xf)
 502 .El
 503 .It Dv BPF_ST
 504 This instruction stores the accumulator into the scratch memory.
 505 We do not need an addressing mode since there is only one possibility
 506 for the destination.
 507 .Pp
 508 .Bl -tag -width "BPF_ST" -compact
 509 .It Li BPF_ST
 510 M[k] <- A
 511 .El
 512 .It Dv BPF_STX
 513 This instruction stores the index register in the scratch memory store.
 514 .Pp
 515 .Bl -tag -width "BPF_STX" -compact
 516 .It Li BPF_STX
 517 M[k] <- X
 518 .El
 519 .It Dv BPF_ALU
 520 The alu instructions perform operations between the accumulator and
 521 index register or constant, and store the result back in the accumulator.
 522 For binary operations, a source mode is required
 523 .Dv ( BPF_K
 524 or
 525 .Dv BPF_X ) .
 526 .Pp
 527 .Bl -tag -width "BPF_ALU+BPF_MUL+BPF_K" -compact
 528 .It Li BPF_ALU+BPF_ADD+BPF_K
 529 A <- A + k
 530 .It Li BPF_ALU+BPF_SUB+BPF_K
 531 A <- A - k
 532 .It Li BPF_ALU+BPF_MUL+BPF_K
 533 A <- A * k
 534 .It Li BPF_ALU+BPF_DIV+BPF_K
 535 A <- A / k
 536 .It Li BPF_ALU+BPF_AND+BPF_K
 537 A <- A & k
 538 .It Li BPF_ALU+BPF_OR+BPF_K
 539 A <- A | k
 540 .It Li BPF_ALU+BPF_LSH+BPF_K
 541 A <- A << k
 542 .It Li BPF_ALU+BPF_RSH+BPF_K
 543 A <- A >> k
 544 .It Li BPF_ALU+BPF_ADD+BPF_X
 545 A <- A + X
 546 .It Li BPF_ALU+BPF_SUB+BPF_X
 547 A <- A - X
 548 .It Li BPF_ALU+BPF_MUL+BPF_X
 549 A <- A * X
 550 .It Li BPF_ALU+BPF_DIV+BPF_X
 551 A <- A / X
 552 .It Li BPF_ALU+BPF_AND+BPF_X
 553 A <- A & X
 554 .It Li BPF_ALU+BPF_OR+BPF_X
 555 A <- A | X
 556 .It Li BPF_ALU+BPF_LSH+BPF_X
 557 A <- A << X
 558 .It Li BPF_ALU+BPF_RSH+BPF_X
 559 A <- A >> X
 560 .It Li BPF_ALU+BPF_NEG
 561 A <- -A
 562 .El
 563 .It Dv BPF_JMP
 564 The jump instructions alter flow of control.  Conditional jumps
 565 compare the accumulator against a constant
 566 .Pq Dv BPF_K
 567 or the index register
 568 .Pq Dv BPF_X .
 569 If the result is true (or non-zero),
 570 the true branch is taken, otherwise the false branch is taken.
 571 Jump offsets are encoded in 8 bits so the longest jump is 256 instructions.
 572 However, the jump always
 573 .Pq Dv BPF_JA
 574 opcode uses the 32 bit
 575 .Li k
 576 field as the offset, allowing arbitrarily distant destinations.
 577 All conditionals use unsigned comparison conventions.
 578 .Pp
 579 .Bl -tag -width "BPF_JMP+BPF_KSET+BPF_X" -compact
 580 .It Li BPF_JMP+BPF_JA
 581 pc += k
 582 .It Li BPF_JMP+BPF_JGT+BPF_K
 583 pc += (A > k) ? jt : jf
 584 .It Li BPF_JMP+BPF_JGE+BPF_K
 585 pc += (A >= k) ? jt : jf
 586 .It Li BPF_JMP+BPF_JEQ+BPF_K
 587 pc += (A == k) ? jt : jf
 588 .It Li BPF_JMP+BPF_JSET+BPF_K
 589 pc += (A & k) ? jt : jf
 590 .It Li BPF_JMP+BPF_JGT+BPF_X
 591 pc += (A > X) ? jt : jf
 592 .It Li BPF_JMP+BPF_JGE+BPF_X
 593 pc += (A >= X) ? jt : jf
 594 .It Li BPF_JMP+BPF_JEQ+BPF_X
 595 pc += (A == X) ? jt : jf
 596 .It Li BPF_JMP+BPF_JSET+BPF_X
 597 pc += (A & X) ? jt : jf
 598 .El
 599 .It Dv BPF_RET
 600 The return instructions terminate the filter program and specify the amount
 601 of packet to accept (i.e., they return the truncation amount).  A return
 602 value of zero indicates that the packet should be ignored.
 603 The return value is either a constant
 604 .Pq Dv BPF_K
 605 or the accumulator
 606 .Pq Dv BPF_A .
 607 .Pp
 608 .Bl -tag -width "BPF_RET+BPF_K" -compact
 609 .It Li BPF_RET+BPF_A
 610 accept A bytes
 611 .It Li BPF_RET+BPF_K
 612 accept k bytes
 613 .El
 614 .It Dv BPF_MISC
 615 The miscellaneous category was created for anything that doesn't
 616 fit into the above classes, and for any new instructions that might need to
 617 be added.  Currently, these are the register transfer instructions
 618 that copy the index register to the accumulator or vice versa.
 619 .Pp
 620 .Bl -tag -width "BPF_MISC+BPF_TAX" -compact
 621 .It Li BPF_MISC+BPF_TAX
 622 X <- A
 623 .It Li BPF_MISC+BPF_TXA
 624 A <- X
 625 .El
 626 .El
 627 .Pp
 628 The
 629 .Nm
 630 interface provides the following macros to facilitate
 631 array initializers:
 632 .Fn BPF_STMT opcode operand
 633 and
 634 .Fn BPF_JUMP opcode operand true_offset false_offset .
 635 .Sh EXAMPLES
 636 The following filter is taken from the Reverse ARP Daemon.  It accepts
 637 only Reverse ARP requests.
 638 .Bd -literal
 639 struct bpf_insn insns[] = {
 640         BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
 641         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_REVARP, 0, 3),
 642         BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
 643         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, REVARP_REQUEST, 0, 1),
 644         BPF_STMT(BPF_RET+BPF_K, sizeof(struct ether_arp) +
 645                  sizeof(struct ether_header)),
 646         BPF_STMT(BPF_RET+BPF_K, 0),
 647 };
 648 .Ed
 649 .Pp
 650 This filter accepts only IP packets between host 128.3.112.15 and
 651 128.3.112.35.
 652 .Bd -literal
 653 struct bpf_insn insns[] = {
 654         BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
 655         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 8),
 656         BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 26),
 657         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 2),
 658         BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
 659         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 3, 4),
 660         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x80037023, 0, 3),
 661         BPF_STMT(BPF_LD+BPF_W+BPF_ABS, 30),
 662         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0x8003700f, 0, 1),
 663         BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
 664         BPF_STMT(BPF_RET+BPF_K, 0),
 665 };
 666 .Ed
 667 .Pp
 668 Finally, this filter returns only TCP finger packets.  We must parse
 669 the IP header to reach the TCP header.  The
 670 .Dv BPF_JSET
 671 instruction
 672 checks that the IP fragment offset is 0 so we are sure
 673 that we have a TCP header.
 674 .Bd -literal
 675 struct bpf_insn insns[] = {
 676         BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 12),
 677         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ETHERTYPE_IP, 0, 10),
 678         BPF_STMT(BPF_LD+BPF_B+BPF_ABS, 23),
 679         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, IPPROTO_TCP, 0, 8),
 680         BPF_STMT(BPF_LD+BPF_H+BPF_ABS, 20),
 681         BPF_JUMP(BPF_JMP+BPF_JSET+BPF_K, 0x1fff, 6, 0),
 682         BPF_STMT(BPF_LDX+BPF_B+BPF_MSH, 14),
 683         BPF_STMT(BPF_LD+BPF_H+BPF_IND, 14),
 684         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 2, 0),
 685         BPF_STMT(BPF_LD+BPF_H+BPF_IND, 16),
 686         BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 79, 0, 1),
 687         BPF_STMT(BPF_RET+BPF_K, (u_int)-1),
 688         BPF_STMT(BPF_RET+BPF_K, 0),
 689 };
 690 .Ed
 691 .Sh SEE ALSO
 692 .Xr tcpdump 1 ,
 693 .Xr ioctl 2 ,
 694 .Xr byteorder 3 ,
 695 .Xr ng_bpf 4
 696 .Rs
 697 .%A McCanne, S.
 698 .%A Jacobson V.
 699 .%T "An efficient, extensible, and portable network monitor"
 700 .Re
 701 .Sh FILES
 702 .Bl -tag -compact -width /dev/bpfXXX
 703 .It Pa /dev/bpf Ns Sy n
 704 the packet filter device
 705 .El
 706 .Sh BUGS
 707 The read buffer must be of a fixed size (returned by the
 708 .Dv BIOCGBLEN
 709 ioctl).
 710 .Pp
 711 A file that does not request promiscuous mode may receive promiscuously
 712 received packets as a side effect of another file requesting this
 713 mode on the same hardware interface.  This could be fixed in the kernel
 714 with additional processing overhead.  However, we favor the model where
 715 all files must assume that the interface is promiscuous, and if
 716 so desired, must utilize a filter to reject foreign packets.
 717 .Pp
 718 Data link protocols with variable length headers are not currently supported.
 719 .Sh HISTORY
 720 The Enet packet filter was created in 1980 by Mike Accetta and
 721 Rick Rashid at Carnegie-Mellon University.  Jeffrey Mogul, at
 722 Stanford, ported the code to
 723 .Bx
 724 and continued its development from
 725 1983 on.  Since then, it has evolved into the Ultrix Packet Filter
 726 at
 727 .Tn DEC ,
 728 a
 729 .Tn STREAMS
 730 .Tn NIT
 731 module under
 732 .Tn SunOS 4.1 ,
 733 and
 734 .Tn BPF .
 735 .Sh AUTHORS
 736 .An -nosplit
 737 .An Steven McCanne ,
 738 of Lawrence Berkeley Laboratory, implemented BPF in
 739 Summer 1990.  Much of the design is due to
 740 .An Van Jacobson .