xref: /freebsd-13.1/lib/libc/sys/mmap.2 (revision a1d78367)
1.\" Copyright (c) 1991, 1993
2.\"	The Regents of the University of California.  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.\" 3. Neither the name of the University nor the names of its contributors
13.\"    may be used to endorse or promote products derived from this software
14.\"    without specific prior written permission.
15.\"
16.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
17.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
20.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26.\" SUCH DAMAGE.
27.\"
28.\"	@(#)mmap.2	8.4 (Berkeley) 5/11/95
29.\" $FreeBSD$
30.\"
31.Dd December 24, 2020
32.Dt MMAP 2
33.Os
34.Sh NAME
35.Nm mmap
36.Nd allocate memory, or map files or devices into memory
37.Sh LIBRARY
38.Lb libc
39.Sh SYNOPSIS
40.In sys/mman.h
41.Ft void *
42.Fn mmap "void *addr" "size_t len" "int prot" "int flags" "int fd" "off_t offset"
43.Sh DESCRIPTION
44The
45.Fn mmap
46system call causes the pages starting at
47.Fa addr
48and continuing for at most
49.Fa len
50bytes to be mapped from the object described by
51.Fa fd ,
52starting at byte offset
53.Fa offset .
54If
55.Fa len
56is not a multiple of the page size, the mapped region may extend past the
57specified range.
58Any such extension beyond the end of the mapped object will be zero-filled.
59.Pp
60If
61.Fa fd
62references a regular file or a shared memory object, the range of
63bytes starting at
64.Fa offset
65and continuing for
66.Fa len
67bytes must be legitimate for the possible (not necessarily
68current) offsets in the object.
69In particular, the
70.Fa offset
71value cannot be negative.
72If the object is truncated and the process later accesses a page that
73is wholly within the truncated region, the access is aborted and a
74.Dv SIGBUS
75signal is delivered to the process.
76.Pp
77If
78.Fa fd
79references a device file, the interpretation of the
80.Fa offset
81value is device specific and defined by the device driver.
82The virtual memory subsystem does not impose any restrictions on the
83.Fa offset
84value in this case, passing it unchanged to the driver.
85.Pp
86If
87.Fa addr
88is non-zero, it is used as a hint to the system.
89(As a convenience to the system, the actual address of the region may differ
90from the address supplied.)
91If
92.Fa addr
93is zero, an address will be selected by the system.
94The actual starting address of the region is returned.
95A successful
96.Fa mmap
97deletes any previous mapping in the allocated address range.
98.Pp
99The protections (region accessibility) are specified in the
100.Fa prot
101argument by
102.Em or Ns 'ing
103the following values:
104.Pp
105.Bl -tag -width PROT_WRITE -compact
106.It Dv PROT_NONE
107Pages may not be accessed.
108.It Dv PROT_READ
109Pages may be read.
110.It Dv PROT_WRITE
111Pages may be written.
112.It Dv PROT_EXEC
113Pages may be executed.
114.El
115.Pp
116In addition to these protection flags,
117.Fx
118provides the ability to set the maximum protection of a region allocated by
119.Nm
120and later altered by
121.Xr mprotect 2 .
122This is accomplished by
123.Em or Ns 'ing
124one or more
125.Dv PROT_
126values wrapped in the
127.Dv PROT_MAX()
128macro into the
129.Fa prot
130argument.
131.Pp
132The
133.Fa flags
134argument specifies the type of the mapped object, mapping options and
135whether modifications made to the mapped copy of the page are private
136to the process or are to be shared with other references.
137Sharing, mapping type and options are specified in the
138.Fa flags
139argument by
140.Em or Ns 'ing
141the following values:
142.Bl -tag -width MAP_PREFAULT_READ
143.It Dv MAP_32BIT
144Request a region in the first 2GB of the current process's address space.
145If a suitable region cannot be found,
146.Fn mmap
147will fail.
148This flag is only available on 64-bit platforms.
149.It Dv MAP_ALIGNED Ns Pq Fa n
150Align the region on a requested boundary.
151If a suitable region cannot be found,
152.Fn mmap
153will fail.
154The
155.Fa n
156argument specifies the binary logarithm of the desired alignment.
157.It Dv MAP_ALIGNED_SUPER
158Align the region to maximize the potential use of large
159.Pq Dq super
160pages.
161If a suitable region cannot be found,
162.Fn mmap
163will fail.
164The system will choose a suitable page size based on the size of
165mapping.
166The page size used as well as the alignment of the region may both be
167affected by properties of the file being mapped.
168In particular,
169the physical address of existing pages of a file may require a specific
170alignment.
171The region is not guaranteed to be aligned on any specific boundary.
172.It Dv MAP_ANON
173Map anonymous memory not associated with any specific file.
174The file descriptor used for creating
175.Dv MAP_ANON
176must be \-1.
177The
178.Fa offset
179argument must be 0.
180.\".It Dv MAP_FILE
181.\"Mapped from a regular file or character-special device memory.
182.It Dv MAP_ANONYMOUS
183This flag is identical to
184.Dv MAP_ANON
185and is provided for compatibility.
186.It Dv MAP_EXCL
187This flag can only be used in combination with
188.Dv MAP_FIXED .
189Please see the definition of
190.Dv MAP_FIXED
191for the description of its effect.
192.It Dv MAP_FIXED
193Do not permit the system to select a different address than the one
194specified.
195If the specified address cannot be used,
196.Fn mmap
197will fail.
198If
199.Dv MAP_FIXED
200is specified,
201.Fa addr
202must be a multiple of the page size.
203If
204.Dv MAP_EXCL
205is not specified, a successful
206.Dv MAP_FIXED
207request replaces any previous mappings for the process'
208pages in the range from
209.Fa addr
210to
211.Fa addr
212+
213.Fa len .
214In contrast, if
215.Dv MAP_EXCL
216is specified, the request will fail if a mapping
217already exists within the range.
218.It Dv MAP_GUARD
219Instead of a mapping, create a guard of the specified size.
220Guards allow a process to create reservations in its address space,
221which can later be replaced by actual mappings.
222.Pp
223.Fa mmap
224will not create mappings in the address range of a guard unless
225the request specifies
226.Dv MAP_FIXED .
227Guards can be destroyed with
228.Xr munmap 2 .
229Any memory access by a thread to the guarded range results
230in the delivery of a
231.Dv SIGSEGV
232signal to that thread.
233.It Dv MAP_NOCORE
234Region is not included in a core file.
235.It Dv MAP_NOSYNC
236Causes data dirtied via this VM map to be flushed to physical media
237only when necessary (usually by the pager) rather than gratuitously.
238Typically this prevents the update daemons from flushing pages dirtied
239through such maps and thus allows efficient sharing of memory across
240unassociated processes using a file-backed shared memory map.
241Without
242this option any VM pages you dirty may be flushed to disk every so often
243(every 30-60 seconds usually) which can create performance problems if you
244do not need that to occur (such as when you are using shared file-backed
245mmap regions for IPC purposes).
246Dirty data will be flushed automatically when all mappings of an object are
247removed and all descriptors referencing the object are closed.
248Note that VM/file system coherency is
249maintained whether you use
250.Dv MAP_NOSYNC
251or not.
252This option is not portable
253across
254.Ux
255platforms (yet), though some may implement the same behavior
256by default.
257.Pp
258.Em WARNING !
259Extending a file with
260.Xr ftruncate 2 ,
261thus creating a big hole, and then filling the hole by modifying a shared
262.Fn mmap
263can lead to severe file fragmentation.
264In order to avoid such fragmentation you should always pre-allocate the
265file's backing store by
266.Fn write Ns ing
267zero's into the newly extended area prior to modifying the area via your
268.Fn mmap .
269The fragmentation problem is especially sensitive to
270.Dv MAP_NOSYNC
271pages, because pages may be flushed to disk in a totally random order.
272.Pp
273The same applies when using
274.Dv MAP_NOSYNC
275to implement a file-based shared memory store.
276It is recommended that you create the backing store by
277.Fn write Ns ing
278zero's to the backing file rather than
279.Fn ftruncate Ns ing
280it.
281You can test file fragmentation by observing the KB/t (kilobytes per
282transfer) results from an
283.Dq Li iostat 1
284while reading a large file sequentially, e.g.,\& using
285.Dq Li dd if=filename of=/dev/null bs=32k .
286.Pp
287The
288.Xr fsync 2
289system call will flush all dirty data and metadata associated with a file,
290including dirty NOSYNC VM data, to physical media.
291The
292.Xr sync 8
293command and
294.Xr sync 2
295system call generally do not flush dirty NOSYNC VM data.
296The
297.Xr msync 2
298system call is usually not needed since
299.Bx
300implements a coherent file system buffer cache.
301However, it may be
302used to associate dirty VM pages with file system buffers and thus cause
303them to be flushed to physical media sooner rather than later.
304.It Dv MAP_PREFAULT_READ
305Immediately update the calling process's lowest-level virtual address
306translation structures, such as its page table, so that every memory
307resident page within the region is mapped for read access.
308Ordinarily these structures are updated lazily.
309The effect of this option is to eliminate any soft faults that would
310otherwise occur on the initial read accesses to the region.
311Although this option does not preclude
312.Fa prot
313from including
314.Dv PROT_WRITE ,
315it does not eliminate soft faults on the initial write accesses to the
316region.
317.It Dv MAP_PRIVATE
318Modifications are private.
319.It Dv MAP_SHARED
320Modifications are shared.
321.It Dv MAP_STACK
322Creates both a mapped region that grows downward on demand and an
323adjoining guard that both reserves address space for the mapped region
324to grow into and limits the mapped region's growth.
325Together, the mapped region and the guard occupy
326.Fa len
327bytes of the address space.
328The guard starts at the returned address, and the mapped region ends at
329the returned address plus
330.Fa len
331bytes.
332Upon access to the guard, the mapped region automatically grows in size,
333and the guard shrinks by an equal amount.
334Essentially, the boundary between the guard and the mapped region moves
335downward so that the access falls within the enlarged mapped region.
336However, the guard will never shrink to less than the number of pages
337specified by the sysctl
338.Dv security.bsd.stack_guard_page ,
339thereby ensuring that a gap for detecting stack overflow always exists
340between the downward growing mapped region and the closest mapped region
341beneath it.
342.Pp
343.Dv MAP_STACK
344implies
345.Dv MAP_ANON
346and
347.Fa offset
348of 0.
349The
350.Fa fd
351argument
352must be -1 and
353.Fa prot
354must include at least
355.Dv PROT_READ
356and
357.Dv PROT_WRITE .
358The size of the guard, in pages, is specified by sysctl
359.Dv security.bsd.stack_guard_page .
360.El
361.Pp
362The
363.Xr close 2
364system call does not unmap pages, see
365.Xr munmap 2
366for further information.
367.Sh NOTES
368Although this implementation does not impose any alignment restrictions on
369the
370.Fa offset
371argument, a portable program must only use page-aligned values.
372.Pp
373Large page mappings require that the pages backing an object be
374aligned in matching blocks in both the virtual address space and RAM.
375The system will automatically attempt to use large page mappings when
376mapping an object that is already backed by large pages in RAM by
377aligning the mapping request in the virtual address space to match the
378alignment of the large physical pages.
379The system may also use large page mappings when mapping portions of an
380object that are not yet backed by pages in RAM.
381The
382.Dv MAP_ALIGNED_SUPER
383flag is an optimization that will align the mapping request to the
384size of a large page similar to
385.Dv MAP_ALIGNED ,
386except that the system will override this alignment if an object already
387uses large pages so that the mapping will be consistent with the existing
388large pages.
389This flag is mostly useful for maximizing the use of large pages on the
390first mapping of objects that do not yet have pages present in RAM.
391.Sh RETURN VALUES
392Upon successful completion,
393.Fn mmap
394returns a pointer to the mapped region.
395Otherwise, a value of
396.Dv MAP_FAILED
397is returned and
398.Va errno
399is set to indicate the error.
400.Sh ERRORS
401The
402.Fn mmap
403system call
404will fail if:
405.Bl -tag -width Er
406.It Bq Er EACCES
407The flag
408.Dv PROT_READ
409was specified as part of the
410.Fa prot
411argument and
412.Fa fd
413was not open for reading.
414The flags
415.Dv MAP_SHARED
416and
417.Dv PROT_WRITE
418were specified as part of the
419.Fa flags
420and
421.Fa prot
422argument and
423.Fa fd
424was not open for writing.
425.It Bq Er EBADF
426The
427.Fa fd
428argument
429is not a valid open file descriptor.
430.It Bq Er EINVAL
431An invalid (negative) value was passed in the
432.Fa offset
433argument, when
434.Fa fd
435referenced a regular file or shared memory.
436.It Bq Er EINVAL
437An invalid value was passed in the
438.Fa prot
439argument.
440.It Bq Er EINVAL
441An undefined option was set in the
442.Fa flags
443argument.
444.It Bq Er EINVAL
445Both
446.Dv MAP_PRIVATE
447and
448.Dv MAP_SHARED
449were specified.
450.It Bq Er EINVAL
451None of
452.Dv MAP_ANON ,
453.Dv MAP_GUARD ,
454.Dv MAP_PRIVATE ,
455.Dv MAP_SHARED ,
456or
457.Dv MAP_STACK
458was specified.
459At least one of these flags must be included.
460.It Bq Er EINVAL
461.Dv MAP_STACK
462was specified and
463.Va len
464is less than or equal to the guard size.
465.It Bq Er EINVAL
466.Dv MAP_FIXED
467was specified and the
468.Fa addr
469argument was not page aligned, or part of the desired address space
470resides out of the valid address space for a user process.
471.It Bq Er EINVAL
472Both
473.Dv MAP_FIXED
474and
475.Dv MAP_32BIT
476were specified and part of the desired address space resides outside
477of the first 2GB of user address space.
478.It Bq Er EINVAL
479The
480.Fa len
481argument
482was equal to zero.
483.It Bq Er EINVAL
484.Dv MAP_ALIGNED
485was specified and the desired alignment was either larger than the
486virtual address size of the machine or smaller than a page.
487.It Bq Er EINVAL
488.Dv MAP_ANON
489was specified and the
490.Fa fd
491argument was not -1.
492.It Bq Er EINVAL
493.Dv MAP_ANON
494was specified and the
495.Fa offset
496argument was not 0.
497.It Bq Er EINVAL
498Both
499.Dv MAP_FIXED
500and
501.Dv MAP_EXCL
502were specified, but the requested region is already used by a mapping.
503.It Bq Er EINVAL
504.Dv MAP_EXCL
505was specified, but
506.Dv MAP_FIXED
507was not.
508.It Bq Er EINVAL
509.Dv MAP_GUARD
510was specified, but the
511.Fa offset
512argument was not zero, the
513.Fa fd
514argument was not -1, or the
515.Fa prot
516argument was not
517.Dv PROT_NONE .
518.It Bq Er EINVAL
519.Dv MAP_GUARD
520was specified together with one of the flags
521.Dv MAP_ANON ,
522.Dv MAP_PREFAULT ,
523.Dv MAP_PREFAULT_READ ,
524.Dv MAP_PRIVATE ,
525.Dv MAP_SHARED ,
526.Dv MAP_STACK .
527.It Bq Er ENODEV
528.Dv MAP_ANON
529has not been specified and
530.Fa fd
531did not reference a regular or character special file.
532.It Bq Er ENOMEM
533.Dv MAP_FIXED
534was specified and the
535.Fa addr
536argument was not available.
537.Dv MAP_ANON
538was specified and insufficient memory was available.
539.It Bq Er ENOTSUP
540The
541.Fa prot
542argument contains protections which are not a subset of the specified
543maximum protections.
544.El
545.Sh SEE ALSO
546.Xr madvise 2 ,
547.Xr mincore 2 ,
548.Xr minherit 2 ,
549.Xr mlock 2 ,
550.Xr mprotect 2 ,
551.Xr msync 2 ,
552.Xr munlock 2 ,
553.Xr munmap 2 ,
554.Xr getpagesize 3 ,
555.Xr getpagesizes 3
556.Sh HISTORY
557The
558.Nm
559system call was first documented in
560.Bx 4.2
561and implemented in
562.Bx 4.4 .
563.\" XXX: lots of missing history of FreeBSD additions.
564.Pp
565The
566.Dv PROT_MAX
567functionality was introduced in
568.Fx 13 .
569