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 .\" 4. 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
36 .Nd allocate memory, or map files or devices into memory
42 .Fn mmap "void *addr" "size_t len" "int prot" "int flags" "int fd" "off_t offset"
46 system call causes the pages starting at
48 and continuing for at most
50 bytes to be mapped from the object described by
52 starting at byte offset
56 is not a multiple of the pagesize, the mapped region may extend past the
58 Any such extension beyond the end of the mapped object will be zero-filled.
62 is non-zero, it is used as a hint to the system.
63 (As a convenience to the system, the actual address of the region may differ
64 from the address supplied.)
67 is zero, an address will be selected by the system.
68 The actual starting address of the region is returned.
71 deletes any previous mapping in the allocated address range.
73 The protections (region accessibility) are specified in the
79 .Bl -tag -width PROT_WRITE -compact
81 Pages may not be accessed.
87 Pages may be executed.
92 argument specifies the type of the mapped object, mapping options and
93 whether modifications made to the mapped copy of the page are private
94 to the process or are to be shared with other references.
95 Sharing, mapping type and options are specified in the
100 .Bl -tag -width MAP_PREFAULT_READ
102 Request a region in the first 2GB of the current process's address space.
103 If a suitable region cannot be found,
106 This flag is only available on 64-bit platforms.
107 .It Dv MAP_ALIGNED Ns Pq Fa n
108 Align the region on a requested boundary.
109 If a suitable region cannot be found,
114 argument specifies the binary logarithm of the desired alignment.
115 .It Dv MAP_ALIGNED_SUPER
116 Align the region to maximize the potential use of large
119 If a suitable region cannot be found,
122 The system will choose a suitable page size based on the size of
124 The page size used as well as the alignment of the region may both be
125 affected by properties of the file being mapped.
127 the physical address of existing pages of a file may require a specific
129 The region is not guaranteed to be aligned on any specific boundary.
131 Map anonymous memory not associated with any specific file.
132 The file descriptor used for creating
139 .\"Mapped from a regular file or character-special device memory.
141 This flag is identical to
143 and is provided for compatibility.
145 Do not permit the system to select a different address than the one
147 If the specified address cannot be used,
154 must be a multiple of the pagesize.
157 request is successful, the mapping established by
159 replaces any previous mappings for the process' pages in the range from
165 Use of this option is discouraged.
166 .It Dv MAP_HASSEMAPHORE
167 Notify the kernel that the region may contain semaphores and that special
168 handling may be necessary.
170 This flag never operated as advertised and is no longer supported.
173 for further information.
175 Region is not included in a core file.
177 Causes data dirtied via this VM map to be flushed to physical media
178 only when necessary (usually by the pager) rather than gratuitously.
179 Typically this prevents the update daemons from flushing pages dirtied
180 through such maps and thus allows efficient sharing of memory across
181 unassociated processes using a file-backed shared memory map.
183 this option any VM pages you dirty may be flushed to disk every so often
184 (every 30-60 seconds usually) which can create performance problems if you
185 do not need that to occur (such as when you are using shared file-backed
186 mmap regions for IPC purposes).
187 Note that VM/file system coherency is
188 maintained whether you use
191 This option is not portable
194 platforms (yet), though some may implement the same behavior
198 Extending a file with
200 thus creating a big hole, and then filling the hole by modifying a shared
202 can lead to severe file fragmentation.
203 In order to avoid such fragmentation you should always pre-allocate the
204 file's backing store by
206 zero's into the newly extended area prior to modifying the area via your
208 The fragmentation problem is especially sensitive to
210 pages, because pages may be flushed to disk in a totally random order.
212 The same applies when using
214 to implement a file-based shared memory store.
215 It is recommended that you create the backing store by
217 zero's to the backing file rather than
220 You can test file fragmentation by observing the KB/t (kilobytes per
221 transfer) results from an
223 while reading a large file sequentially, e.g.\& using
224 .Dq Li dd if=filename of=/dev/null bs=32k .
228 system call will flush all dirty data and metadata associated with a file,
229 including dirty NOSYNC VM data, to physical media.
234 system call generally do not flush dirty NOSYNC VM data.
237 system call is usually not needed since
239 implements a coherent file system buffer cache.
241 used to associate dirty VM pages with file system buffers and thus cause
242 them to be flushed to physical media sooner rather than later.
243 .It Dv MAP_PREFAULT_READ
244 Immediately update the calling process's lowest-level virtual address
245 translation structures, such as its page table, so that every memory
246 resident page within the region is mapped for read access.
247 Ordinarily these structures are updated lazily.
248 The effect of this option is to eliminate any soft faults that would
249 otherwise occur on the initial read accesses to the region.
250 Although this option does not preclude
254 it does not eliminate soft faults on the initial write accesses to the
257 Modifications are private.
259 Modifications are shared.
272 must include at least
277 a memory region that grows to at most
279 bytes in size, starting from the stack top and growing down.
281 stack top is the starting address returned by the call, plus
284 The bottom of the stack at maximum growth is the starting
285 address returned by the call.
290 system call does not unmap pages, see
292 for further information.
294 The current design does not allow a process to specify the location of
296 In the future we may define an additional mapping type,
299 the file descriptor argument specifies a file or device to which swapping
302 Although this implementation does not impose any alignment restrictions on
305 argument, a portable program must only use page-aligned values.
307 Large page mappings require that the pages backing an object be
308 aligned in matching blocks in both the virtual address space and RAM.
309 The system will automatically attempt to use large page mappings when
310 mapping an object that is already backed by large pages in RAM by
311 aligning the mapping request in the virtual address space to match the
312 alignment of the large physical pages.
313 The system may also use large page mappings when mapping portions of an
314 object that are not yet backed by pages in RAM.
316 .Dv MAP_ALIGNED_SUPER
317 flag is an optimization that will align the mapping request to the
318 size of a large page similar to
320 except that the system will override this alignment if an object already
321 uses large pages so that the mapping will be consistent with the existing
323 This flag is mostly useful for maximizing the use of large pages on the
324 first mapping of objects that do not yet have pages present in RAM.
326 Upon successful completion,
328 returns a pointer to the mapped region.
329 Otherwise, a value of
333 is set to indicate the error.
343 was specified as part of the
347 was not open for reading.
352 were specified as part of the
358 was not open for writing.
363 is not a valid open file descriptor.
366 was specified and the
368 argument was not page aligned, or part of the desired address space
369 resides out of the valid address space for a user process.
375 were specified and part of the desired address space resides outside
376 of the first 2GB of user address space.
384 was specified and the desired alignment was either larger than the
385 virtual address size of the machine or smaller than a page.
388 was specified and the
393 was specified and the
398 has not been specified and
400 did not reference a regular or character special file.
403 was specified and the
405 argument was not available.
407 was specified and insufficient memory was available.
424 is limited to the maximum file size or available userland address
426 Files may not be able to be made more than 1TB large on 32 bit systems
427 due to file systems restrictions and bugs, but address space is far more
429 Larger files may be possible on 64 bit systems.
431 The previous documented limit of 2GB was a documentation bug.
432 That limit has not existed since