bsd/man/man2/kqueue.2

   1 .\" Copyright (c) 2000 Jonathan Lemon
   2 .\" All rights reserved.
   3 .\"
   4 .\" Redistribution and use in source and binary forms, with or without
   5 .\" modification, are permitted provided that the following conditions
   6 .\" are met:
   7 .\" 1. Redistributions of source code must retain the above copyright
   8 .\"    notice, this list of conditions and the following disclaimer.
   9 .\" 2. Redistributions in binary form must reproduce the above copyright
  10 .\"    notice, this list of conditions and the following disclaimer in the
  11 .\"    documentation and/or other materials provided with the distribution.
  12 .\"
  13 .\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND
  14 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  15 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  16 .\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  17 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  18 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  19 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  20 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  21 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  22 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  23 .\" SUCH DAMAGE.
  24 .\"
  25 .\" $FreeBSD: src/lib/libc/sys/kqueue.2,v 1.32 2002/12/19 09:40:25 ru Exp $
  26 .\"
  27 .Dd April 14, 2000
  28 .Dt KQUEUE 2
  29 .Os
  30 .Sh NAME
  31 .Nm kqueue ,
  32 .Nm kevent
  33 .Nd kernel event notification mechanism
  34 .Sh LIBRARY
  35 .Lb libc
  36 .Sh SYNOPSIS
  37 .In sys/types.h
  38 .In sys/event.h
  39 .In sys/time.h
  40 .Ft int
  41 .Fn kqueue "void"
  42 .Ft int
  43 .Fn kevent "int kq" "const struct kevent *changelist" "int nchanges" "struct kevent *eventlist" "int nevents" "const struct timespec *timeout"
  44 .Fn EV_SET "&kev" ident filter flags fflags data udata
  45 .Sh DESCRIPTION
  46 The
  47 .Fn kqueue
  48 system call
  49 provides a generic method of notifying the user when an kernel
  50 event (kevent) happens or a condition holds, based on the results
  51 of small pieces of kernel code termed filters.
  52 A kevent is identified by an (ident, filter) pair and specifies
  53 the interesting conditions to be notified about for that pair.
  54 An (ident, filter) pair can only appear once is a given kqueue.
  55 Subsequent attempts to register the same pair for a given kqueue
  56 will result in the replacement of the conditions being watched,
  57 not an addition.
  58 .Pp
  59 The filter identified in a kevent is executed upon the initial
  60 registration of that event in order to detect whether a preexisting
  61 condition is present, and is also executed whenever an event is
  62 passed to the filter for evaluation.
  63 If the filter determines that the condition should be reported,
  64 then the kevent is placed on the kqueue for the user to retrieve.
  65 .Pp
  66 The filter is also run when the user attempts to retrieve the kevent
  67 from the kqueue.
  68 If the filter indicates that the condition that triggered
  69 the event no longer holds, the kevent is removed from the kqueue and
  70 is not returned.
  71 .Pp
  72 Multiple events which trigger the filter do not result in multiple
  73 kevents being placed on the kqueue; instead, the filter will aggregate
  74 the events into a single struct kevent.
  75 Calling
  76 .Fn close
  77 on a file descriptor will remove any kevents that reference the descriptor.
  78 .Pp
  79 The
  80 .Fn kqueue
  81 system call
  82 creates a new kernel event queue and returns a descriptor.
  83 The queue is not inherited by a child created with
  84 .Xr fork 2 .
  85 .Pp
  86 The
  87 .Fn kevent
  88 system call
  89 is used to register events with the queue, and return any pending
  90 events to the user.
  91 The
  92 .Fa changelist
  93 argument
  94 is a pointer to an array of
  95 .Va kevent
  96 structures, as defined in
  97 .Aq Pa sys/event.h .
  98 All changes contained in the
  99 .Fa changelist
 100 are applied before any pending events are read from the queue.
 101 The
 102 .Fa nchanges
 103 argument
 104 gives the size of
 105 .Fa changelist .
 106 The
 107 .Fa eventlist
 108 argument
 109 is a pointer to an array of kevent structures.
 110 The
 111 .Fa nevents
 112 argument
 113 determines the size of
 114 .Fa eventlist .
 115 If
 116 .Fa timeout
 117 is a non-NULL pointer, it specifies a maximum interval to wait
 118 for an event, which will be interpreted as a struct timespec.  If
 119 .Fa timeout
 120 is a NULL pointer,
 121 .Fn kevent
 122 waits indefinitely.  To effect a poll, the
 123 .Fa timeout
 124 argument should be non-NULL, pointing to a zero-valued
 125 .Va timespec
 126 structure.  The same array may be used for the
 127 .Fa changelist
 128 and
 129 .Fa eventlist .
 130 .Pp
 131 The
 132 .Fn EV_SET
 133 macro is provided for ease of initializing a
 134 kevent structure.
 135 .Pp
 136 The
 137 .Va kevent
 138 structure is defined as:
 139 .Bd -literal
 140 struct kevent {
 141         uintptr_t ident;        /* identifier for this event */
 142         short     filter;       /* filter for event */
 143         u_short   flags;        /* action flags for kqueue */
 144         u_int     fflags;       /* filter flag value */
 145         intptr_t  data;         /* filter data value */
 146         void      *udata;       /* opaque user data identifier */
 147 };
 148 .Ed
 149 .Pp
 150 The fields of
 151 .Fa struct kevent
 152 are:
 153 .Bl -tag -width XXXfilter
 154 .It ident
 155 Value used to identify this event.
 156 The exact interpretation is determined by the attached filter,
 157 but often is a file descriptor.
 158 .It filter
 159 Identifies the kernel filter used to process this event.  The pre-defined
 160 system filters are described below.
 161 .It flags
 162 Actions to perform on the event.
 163 .It fflags
 164 Filter-specific flags.
 165 .It data
 166 Filter-specific data value.
 167 .It udata
 168 Opaque user-defined value passed through the kernel unchanged.
 169 .El
 170 .Pp
 171 The
 172 .Va flags
 173 field can contain the following values:
 174 .Bl -tag -width XXXEV_ONESHOT
 175 .It EV_ADD
 176 Adds the event to the kqueue.  Re-adding an existing event
 177 will modify the parameters of the original event, and not result
 178 in a duplicate entry.  Adding an event automatically enables it,
 179 unless overridden by the EV_DISABLE flag.
 180 .It EV_ENABLE
 181 Permit
 182 .Fn kevent
 183 to return the event if it is triggered.
 184 .It EV_DISABLE
 185 Disable the event so
 186 .Fn kevent
 187 will not return it.  The filter itself is not disabled.
 188 .It EV_DELETE
 189 Removes the event from the kqueue.  Events which are attached to
 190 file descriptors are automatically deleted on the last close of
 191 the descriptor.
 192 .It EV_RECEIPT
 193 This flag is useful for making bulk changes to a kqueue without draining any
 194 pending events. When passed as input, it forces EV_ERROR to always be returned.
 195 When a filter is successfully added. The
 196 .Va data
 197 field will be zero.
 198 .It EV_ONESHOT
 199 Causes the event to return only the first occurrence of the filter
 200 being triggered.  After the user retrieves the event from the kqueue,
 201 it is deleted.
 202 .It EV_CLEAR
 203 After the event is retrieved by the user, its state is reset.
 204 This is useful for filters which report state transitions
 205 instead of the current state.  Note that some filters may automatically
 206 set this flag internally.
 207 .It EV_EOF
 208 Filters may set this flag to indicate filter-specific EOF condition.
 209 .It EV_ERROR
 210 See
 211 .Sx RETURN VALUES
 212 below.
 213 .El
 214 .Pp
 215 The predefined system filters are listed below.
 216 Arguments may be passed to and from the filter via the
 217 .Va fflags
 218 and
 219 .Va data
 220 fields in the kevent structure.
 221 .Bl -tag -width EVFILT_SIGNAL
 222 .It EVFILT_READ
 223 Takes a file descriptor as the identifier, and returns whenever
 224 there is data available to read.
 225 The behavior of the filter is slightly different depending
 226 on the descriptor type.
 227 .Pp
 228 .Bl -tag -width 2n
 229 .It Sockets
 230 Sockets which have previously been passed to
 231 .Fn listen
 232 return when there is an incoming connection pending.
 233 .Va data
 234 contains the size of the listen backlog.
 235 .Pp
 236 Other socket descriptors return when there is data to be read,
 237 subject to the
 238 .Dv SO_RCVLOWAT
 239 value of the socket buffer.
 240 This may be overridden with a per-filter low water mark at the
 241 time the filter is added by setting the
 242 NOTE_LOWAT
 243 flag in
 244 .Va fflags ,
 245 and specifying the new low water mark in
 246 .Va data .
 247 On return,
 248 .Va data
 249 contains the number of bytes of protocol data available to read.
 250 .Pp
 251 If the read direction of the socket has shutdown, then the filter
 252 also sets EV_EOF in
 253 .Va flags ,
 254 and returns the socket error (if any) in
 255 .Va fflags .
 256 It is possible for EOF to be returned (indicating the connection is gone)
 257 while there is still data pending in the socket buffer.
 258 .It Vnodes
 259 Returns when the file pointer is not at the end of file.
 260 .Va data
 261 contains the offset from current position to end of file,
 262 and may be negative.
 263 .It "Fifos, Pipes"
 264 Returns when the there is data to read;
 265 .Va data
 266 contains the number of bytes available.
 267 .Pp
 268 When the last writer disconnects, the filter will set EV_EOF in
 269 .Va flags .
 270 This may be cleared by passing in EV_CLEAR, at which point the
 271 filter will resume waiting for data to become available before
 272 returning.
 273 .El
 274 .It EVFILT_WRITE
 275 Takes a file descriptor as the identifier, and returns whenever
 276 it is possible to write to the descriptor.  For sockets, pipes
 277 and fifos,
 278 .Va data
 279 will contain the amount of space remaining in the write buffer.
 280 The filter will set EV_EOF when the reader disconnects, and for
 281 the fifo case, this may be cleared by use of EV_CLEAR.
 282 Note that this filter is not supported for vnodes.
 283 .Pp
 284 For sockets, the low water mark and socket error handling is
 285 identical to the EVFILT_READ case.
 286 .It EVFILT_AIO
 287 This filter is currently unsupported.
 288 .\"The sigevent portion of the AIO request is filled in, with
 289 .\".Va sigev_notify_kqueue
 290 .\"containing the descriptor of the kqueue that the event should
 291 .\"be attached to,
 292 .\".Va sigev_value
 293 .\"containing the udata value, and
 294 .\".Va sigev_notify
 295 .\"set to SIGEV_KEVENT.
 296 .\"When the
 297 .\".Fn aio_*
 298 .\"system call is made, the event will be registered
 299 .\"with the specified kqueue, and the
 300 .\".Va ident
 301 .\"argument set to the
 302 .\".Fa struct aiocb
 303 .\"returned by the
 304 .\".Fn aio_*
 305 .\"system call.
 306 .\"The filter returns under the same conditions as aio_error.
 307 .\".Pp
 308 .\"Alternatively, a kevent structure may be initialized, with
 309 .\".Va ident
 310 .\"containing the descriptor of the kqueue, and the
 311 .\"address of the kevent structure placed in the
 312 .\".Va aio_lio_opcode
 313 .\"field of the AIO request.  However, this approach will not work on
 314 .\"architectures with 64-bit pointers, and should be considered deprecated.
 315 .It EVFILT_VNODE
 316 Takes a file descriptor as the identifier and the events to watch for in
 317 .Va fflags ,
 318 and returns when one or more of the requested events occurs on the descriptor.
 319 The events to monitor are:
 320 .Bl -tag -width XXNOTE_RENAME
 321 .It NOTE_DELETE
 322 The
 323 .Fn unlink
 324 system call
 325 was called on the file referenced by the descriptor.
 326 .It NOTE_WRITE
 327 A write occurred on the file referenced by the descriptor.
 328 .It NOTE_EXTEND
 329 The file referenced by the descriptor was extended.
 330 .It NOTE_ATTRIB
 331 The file referenced by the descriptor had its attributes changed.
 332 .It NOTE_LINK
 333 The link count on the file changed.
 334 .It NOTE_RENAME
 335 The file referenced by the descriptor was renamed.
 336 .It NOTE_REVOKE
 337 Access to the file was revoked via
 338 .Xr revoke 2
 339 or the underlying fileystem was unmounted.
 340 .El
 341 .Pp
 342 On return,
 343 .Va fflags
 344 contains the events which triggered the filter.
 345 .It EVFILT_PROC
 346 Takes the process ID to monitor as the identifier and the events to watch for
 347 in
 348 .Va fflags ,
 349 and returns when the process performs one or more of the requested events.
 350 If a process can normally see another process, it can attach an event to it.
 351 The events to monitor are:
 352 .Bl -tag -width
 353 .It NOTE_EXIT
 354 The process has exited.
 355 .It NOTE_FORK
 356 The process created a child process via
 357 .Xr fork 2
 358 or similar call.
 359 .It NOTE_EXEC
 360 The process executed a new process via
 361 .Xr execve 2
 362 or similar call.
 363 .It NOTE_SIGNAL
 364 The process was sent a signal. Status can be checked via
 365 .Xr waitpid 2
 366 or similar call.
 367 .It NOTE_REAP
 368 The process was reaped by the parent via
 369 .Xr wait 2
 370 or similar call.
 371 .El
 372 .Pp
 373 On return,
 374 .Va fflags
 375 contains the events which triggered the filter.
 376 .It EVFILT_SIGNAL
 377 Takes the signal number to monitor as the identifier and returns
 378 when the given signal is delivered to the process.
 379 This coexists with the
 380 .Fn signal
 381 and
 382 .Fn sigaction
 383 facilities, and has a lower precedence.  The filter will record
 384 all attempts to deliver a signal to a process, even if the signal has
 385 been marked as SIG_IGN.  Event notification happens after normal
 386 signal delivery processing.
 387 .Va data
 388 returns the number of times the signal has occurred since the last call to
 389 .Fn kevent .
 390 This filter automatically sets the EV_CLEAR flag internally.
 391 .It EVFILT_TIMER
 392 This filter is currently unsupported.
 393 .\"Establishes an arbitrary timer identified by
 394 .\".Va ident .
 395 .\"When adding a timer,
 396 .\".Va data
 397 .\"specifies the timeout period in milliseconds.
 398 .\"The timer will be periodic unless EV_ONESHOT is specified.
 399 .\"On return,
 400 .\".Va data
 401 .\"contains the number of times the timeout has expired since the last call to
 402 .\".Fn kevent .
 403 .\"This filter automatically sets the EV_CLEAR flag internally.
 404 .El
 405 .Sh RETURN VALUES
 406 The
 407 .Fn kqueue
 408 system call
 409 creates a new kernel event queue and returns a file descriptor.
 410 If there was an error creating the kernel event queue, a value of -1 is
 411 returned and errno set.
 412 .Pp
 413 The
 414 .Fn kevent
 415 system call
 416 returns the number of events placed in the
 417 .Fa eventlist ,
 418 up to the value given by
 419 .Fa nevents .
 420 If an error occurs while processing an element of the
 421 .Fa changelist
 422 and there is enough room in the
 423 .Fa eventlist ,
 424 then the event will be placed in the
 425 .Fa eventlist
 426 with
 427 .Dv EV_ERROR
 428 set in
 429 .Va flags
 430 and the system error in
 431 .Va data .
 432 Otherwise,
 433 .Dv -1
 434 will be returned, and
 435 .Dv errno
 436 will be set to indicate the error condition.
 437 If the time limit expires, then
 438 .Fn kevent
 439 returns 0.
 440 .Sh ERRORS
 441 The
 442 .Fn kqueue
 443 system call fails if:
 444 .Bl -tag -width Er
 445 .It Bq Er ENOMEM
 446 The kernel failed to allocate enough memory for the kernel queue.
 447 .It Bq Er EMFILE
 448 The per-process descriptor table is full.
 449 .It Bq Er ENFILE
 450 The system file table is full.
 451 .El
 452 .Pp
 453 The
 454 .Fn kevent
 455 system call fails if:
 456 .Bl -tag -width Er
 457 .It Bq Er EACCES
 458 The process does not have permission to register a filter.
 459 .It Bq Er EFAULT
 460 There was an error reading or writing the
 461 .Va kevent
 462 structure.
 463 .It Bq Er EBADF
 464 The specified descriptor is invalid.
 465 .It Bq Er EINTR
 466 A signal was delivered before the timeout expired and before any
 467 events were placed on the kqueue for return.
 468 .It Bq Er EINVAL
 469 The specified time limit or filter is invalid.
 470 .It Bq Er ENOENT
 471 The event could not be found to be modified or deleted.
 472 .It Bq Er ENOMEM
 473 No memory was available to register the event.
 474 .It Bq Er ESRCH
 475 The specified process to attach to does not exist.
 476 .El
 477 .Sh SEE ALSO
 478 .Xr aio_error 2 ,
 479 .Xr aio_read 2 ,
 480 .Xr aio_return 2 ,
 481 .Xr read 2 ,
 482 .Xr select 2 ,
 483 .Xr sigaction 2 ,
 484 .Xr write 2 ,
 485 .Xr signal 3
 486 .Sh HISTORY
 487 The
 488 .Fn kqueue
 489 and
 490 .Fn kevent
 491 system calls first appeared in
 492 .Fx 4.1 .
 493 .Sh AUTHORS
 494 The
 495 .Fn kqueue
 496 system and this manual page were written by
 497 .An Jonathan Lemon Aq jlemon@FreeBSD.org .
 498 .Sh BUGS
 499 Not all filesystem types support kqueue-style notifications.
 500 And even some that do, like some remote filesystems, may only
 501 support a subset of the notification semantics described
 502 here.