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. 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.
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
28 .\" @(#)mmap.2 8.4 (Berkeley) 5/11/95
29 .\" $FreeBSD: src/lib/libc/sys/mmap.2,v 1.22.2.12 2002/02/27 03:40:13 dd Exp $
30 .\" $DragonFly: src/lib/libc/sys/mmap.2,v 1.9 2007/05/17 08:19:00 swildner Exp $
37 .Nd allocate memory, or map files or devices into memory
44 .Fn mmap "void *addr" "size_t len" "int prot" "int flags" "int fd" "off_t offset"
48 function causes the pages starting at
50 and continuing for at most
52 bytes to be mapped from the object described by
54 starting at byte offset
58 is not a multiple of the pagesize, the mapped region may extend past the
60 Any such extension beyond the end of the mapped object will be zero-filled.
64 is non-zero, it is used as a hint to the system.
65 (As a convenience to the system, the actual address of the region may differ
66 from the address supplied.)
69 is zero, an address will be selected by the system.
70 The actual starting address of the region is returned.
73 deletes any previous mapping in the allocated address range.
75 The protections (region accessibility) are specified in the
81 .Bl -tag -width PROT_WRITE -compact
83 Pages may not be accessed.
89 Pages may be executed.
94 parameter specifies the type of the mapped object, mapping options and
95 whether modifications made to the mapped copy of the page are private
96 to the process or are to be shared with other references.
97 Sharing, mapping type and options are specified in the
101 the following values:
102 .Bl -tag -width MAP_HASSEMAPHORE
104 Map anonymous memory not associated with any specific file.
105 The file descriptor used for creating
110 parameter is ignored.
112 .\"Mapped from a regular file or character-special device memory.
114 Do not permit the system to select a different address than the one
116 If the specified address contains other mappings those mappings will
118 If the specified address cannot otherwise be used,
125 must be a multiple of the pagesize.
127 Try to do a fixed mapping but fail if another mapping already exists in
128 the space instead of overwriting the mapping.
130 When used with MAP_STACK this flag allows one MAP_STACK mapping to be
131 made within another (typically the master user stack), as long as
132 no pages have been faulted in the area requested.
133 .It Dv MAP_HASSEMAPHORE
134 Notify the kernel that the region may contain semaphores and that special
135 handling may be necessary.
137 Region is not included in a core file.
139 Causes data dirtied via this VM map to be flushed to physical media
140 only when necessary (usually by the pager) rather than gratuitously.
141 Typically this prevents the update daemons from flushing pages dirtied
142 through such maps and thus allows efficient sharing of memory across
143 unassociated processes using a file-backed shared memory map. Without
144 this option any VM pages you dirty may be flushed to disk every so often
145 (every 30-60 seconds usually) which can create performance problems if you
146 do not need that to occur (such as when you are using shared file-backed
147 mmap regions for IPC purposes). Note that VM/filesystem coherency is
148 maintained whether you use
150 or not. This option is not portable
153 platforms (yet), though some may implement the same behavior
157 Extending a file with
159 thus creating a big hole, and then filling the hole by modifying a shared
161 can lead to severe file fragmentation.
162 In order to avoid such fragmentation you should always pre-allocate the
163 file's backing store by
165 zero's into the newly extended area prior to modifying the area via your
167 The fragmentation problem is especially sensitive to
169 pages, because pages may be flushed to disk in a totally random order.
171 The same applies when using
173 to implement a file-based shared memory store.
174 It is recommended that you create the backing store by
176 zero's to the backing file rather than
179 You can test file fragmentation by observing the KB/t (kilobytes per
180 transfer) results from an
182 while reading a large file sequentially, e.g. using
183 .Dq Li dd if=filename of=/dev/null bs=32k .
187 function will flush all dirty data and metadata associated with a file,
188 including dirty NOSYNC VM data, to physical media. The
192 system call generally do not flush dirty NOSYNC VM data.
195 system call is obsolete since
197 implements a coherent filesystem buffer cache. However, it may be
198 used to associate dirty VM pages with filesystem buffers and thus cause
199 them to be flushed to physical media sooner rather than later.
201 Modifications are private.
203 Modifications are shared.
205 Map the area as a stack.
213 should include at least
218 a memory region that grows to at most
220 bytes in size, starting from the stack top and growing down. The
221 stack top is the starting address returned by the call, plus
224 The bottom of the stack at maximum growth is the starting
225 address returned by the call.
227 The entire area is reserved from the point of view of other
229 calls, even if not faulted in yet.
231 WARNING. We currently allow
233 mappings to provide a hint that points within an existing
235 mapping's space, and this will succeed as long as no page have been
236 faulted in the area specified, but this behavior is no longer supported
237 unless you also specify the
245 is used, you cannot count on the returned address matching the hint
247 .It Dv MAP_VPAGETABLE
248 Memory accessed via this map is not linearly mapped and will be governed
249 by a virtual page table. The base address of the virtual page table may
254 Virtual page tables work with anonymous memory but there
255 is no way to populate the page table so for all intents and purposes
257 can only be used when mapping file descriptors. Since the kernel will
258 update the VPTE_M bit in the virtual page table, the mapping must R+W
259 even though actual access to the memory will be properly governed by
260 the virtual page table.
262 Addressable backing store is limited by the range supported in the virtual
263 page table entries. The kernel may implement a page table abstraction capable
264 of addressing a larger range within the backing store then could otherwise
265 be mapped into memory.
270 function does not unmap pages, see
272 for further information.
274 The current design does not allow a process to specify the location of
276 In the future we may define an additional mapping type,
279 the file descriptor argument specifies a file or device to which swapping
282 Upon successful completion,
284 returns a pointer to the mapped region.
285 Otherwise, a value of
289 is set to indicate the error.
297 was specified as part of the
301 was not open for reading.
306 were specified as part of the
312 was not open for writing.
315 is not a valid open file descriptor.
318 was specified and the
320 parameter was not page aligned, or part of the desired address space
321 resides out of the valid address space for a user process.
327 was specified and the
329 parameter was not -1.
332 has not been specified and
334 did not reference a regular or character special file.
337 was not page-aligned.
343 was specified and the
345 parameter wasn't available.
347 was specified and insufficient memory was available.
348 The system has reached the per-process mmap limit specified in the
363 is limited to 2GB. Mmapping slightly more than 2GB doesn't work, but
364 it is possible to map a window of size (filesize % 2GB) for file sizes
365 of slightly less than 2G, 4GB, 6GB and 8GB.
367 The limit is imposed for a variety of reasons.
368 Most of them have to do
371 not wanting to use 64 bit offsets in the VM system due to
372 the extreme performance penalty.
375 uses 32bit page indexes and
378 a maximum of 8TB filesizes.
379 It's actually bugs in
380 the filesystem code that causes the limit to be further restricted to
381 1TB (loss of precision when doing blockno calculations).
383 Another reason for the 2GB limit is that filesystem metadata can
384 reside at negative offsets.