1 .\" Copyright (c) 1991, 1993
2 .\" The Regents of the University of California. All rights reserved.
4 .\" Redistribution and use in source and binary forms, with or without
5 .\" modification, are permitted provided that the following conditions
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. All advertising materials mentioning features or use of this software
13 .\" must display the following acknowledgement:
14 .\" This product includes software developed by the University of
15 .\" California, Berkeley and its contributors.
16 .\" 4. Neither the name of the University nor the names of its contributors
17 .\" may be used to endorse or promote products derived from this software
18 .\" without specific prior written permission.
20 .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 .\" @(#)mmap.2 8.4 (Berkeley) 5/11/95
40 .Nd allocate memory, or map files or devices into memory
46 .Fn mmap "void *addr" "size_t len" "int prot" "int flags" "int fd" "off_t offset"
50 system call causes the pages starting at
52 and continuing for at most
54 bytes to be mapped from the object described by
56 starting at byte offset
60 is not a multiple of the pagesize, the mapped region may extend past the
62 Any such extension beyond the end of the mapped object will be zero-filled.
66 is non-zero, it is used as a hint to the system.
67 (As a convenience to the system, the actual address of the region may differ
68 from the address supplied.)
71 is zero, an address will be selected by the system.
72 The actual starting address of the region is returned.
75 deletes any previous mapping in the allocated address range.
77 The protections (region accessibility) are specified in the
83 .Bl -tag -width PROT_WRITE -compact
85 Pages may not be accessed.
91 Pages may be executed.
96 argument specifies the type of the mapped object, mapping options and
97 whether modifications made to the mapped copy of the page are private
98 to the process or are to be shared with other references.
99 Sharing, mapping type and options are specified in the
103 the following values:
104 .Bl -tag -width MAP_HASSEMAPHORE
106 Map anonymous memory not associated with any specific file.
107 The file descriptor used for creating
114 .\"Mapped from a regular file or character-special device memory.
116 Do not permit the system to select a different address than the one
118 If the specified address cannot be used,
125 must be a multiple of the pagesize.
126 If a MAP_FIXED request is successful, the mapping established by
128 replaces any previous mappings for the process' pages in the range from
133 Use of this option is discouraged.
134 .It Dv MAP_HASSEMAPHORE
135 Notify the kernel that the region may contain semaphores and that special
136 handling may be necessary.
138 This flag never operated as advertised and is no longer supported.
141 for further information.
143 Region is not included in a core file.
145 Causes data dirtied via this VM map to be flushed to physical media
146 only when necessary (usually by the pager) rather than gratuitously.
147 Typically this prevents the update daemons from flushing pages dirtied
148 through such maps and thus allows efficient sharing of memory across
149 unassociated processes using a file-backed shared memory map.
151 this option any VM pages you dirty may be flushed to disk every so often
152 (every 30-60 seconds usually) which can create performance problems if you
153 do not need that to occur (such as when you are using shared file-backed
154 mmap regions for IPC purposes).
155 Note that VM/file system coherency is
156 maintained whether you use
159 This option is not portable
162 platforms (yet), though some may implement the same behavior
166 Extending a file with
168 thus creating a big hole, and then filling the hole by modifying a shared
170 can lead to severe file fragmentation.
171 In order to avoid such fragmentation you should always pre-allocate the
172 file's backing store by
174 zero's into the newly extended area prior to modifying the area via your
176 The fragmentation problem is especially sensitive to
178 pages, because pages may be flushed to disk in a totally random order.
180 The same applies when using
182 to implement a file-based shared memory store.
183 It is recommended that you create the backing store by
185 zero's to the backing file rather than
188 You can test file fragmentation by observing the KB/t (kilobytes per
189 transfer) results from an
191 while reading a large file sequentially, e.g.\& using
192 .Dq Li dd if=filename of=/dev/null bs=32k .
196 system call will flush all dirty data and metadata associated with a file,
197 including dirty NOSYNC VM data, to physical media.
202 system call generally do not flush dirty NOSYNC VM data.
205 system call is obsolete since
207 implements a coherent file system buffer cache.
209 used to associate dirty VM pages with file system buffers and thus cause
210 them to be flushed to physical media sooner rather than later.
212 Modifications are private.
214 Modifications are shared.
227 must include at least
232 a memory region that grows to at most
234 bytes in size, starting from the stack top and growing down.
236 stack top is the starting address returned by the call, plus
239 The bottom of the stack at maximum growth is the starting
240 address returned by the call.
245 system call does not unmap pages, see
247 for further information.
249 The current design does not allow a process to specify the location of
251 In the future we may define an additional mapping type,
254 the file descriptor argument specifies a file or device to which swapping
257 Upon successful completion,
259 returns a pointer to the mapped region.
260 Otherwise, a value of
264 is set to indicate the error.
274 was specified as part of the
278 was not open for reading.
283 were specified as part of the
289 was not open for writing.
294 is not a valid open file descriptor.
297 was specified and the
299 argument was not page aligned, or part of the desired address space
300 resides out of the valid address space for a user process.
308 was specified and the
313 has not been specified and
315 did not reference a regular or character special file.
320 was not page-aligned.
326 was specified and the
328 argument was not available.
330 was specified and insufficient memory was available.
331 The system has reached the per-process mmap limit specified in the
351 Mmapping slightly more than 2GB does not work, but
352 it is possible to map a window of size (filesize % 2GB) for file sizes
353 of slightly less than 2G, 4GB, 6GB and 8GB.
355 The limit is imposed for a variety of reasons.
356 Most of them have to do
359 not wanting to use 64 bit offsets in the VM system due to
360 the extreme performance penalty.
363 uses 32bit page indexes and
366 a maximum of 8TB filesizes.
367 It is actually bugs in
368 the file system code that causes the limit to be further restricted to
369 1TB (loss of precision when doing blockno calculations).
371 Another reason for the 2GB limit is that file system metadata can
372 reside at negative offsets.
374 Note that an attempt to
376 zero bytes has no effect and succeeds, while an attempt to
378 zero bytes will return