Initial import from FreeBSD RELENG_4:
[dragonfly.git] / lib / libc / sys / mmap.2
CommitLineData
984263bc
MD
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. 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.
19.\"
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
30.\" SUCH DAMAGE.
31.\"
32.\" @(#)mmap.2 8.4 (Berkeley) 5/11/95
33.\" $FreeBSD: src/lib/libc/sys/mmap.2,v 1.22.2.12 2002/02/27 03:40:13 dd Exp $
34.\"
35.Dd November 17, 2001
36.Dt MMAP 2
37.Os
38.Sh NAME
39.Nm mmap
40.Nd allocate memory, or map files or devices into memory
41.Sh LIBRARY
42.Lb libc
43.Sh SYNOPSIS
44.In sys/types.h
45.In sys/mman.h
46.Ft void *
47.Fn mmap "void *addr" "size_t len" "int prot" "int flags" "int fd" "off_t offset"
48.Sh DESCRIPTION
49The
50.Fn mmap
51function causes the pages starting at
52.Fa addr
53and continuing for at most
54.Fa len
55bytes to be mapped from the object described by
56.Fa fd ,
57starting at byte offset
58.Fa offset .
59If
60.Fa len
61is not a multiple of the pagesize, the mapped region may extend past the
62specified range.
63Any such extension beyond the end of the mapped object will be zero-filled.
64.Pp
65If
66.Fa addr
67is non-zero, it is used as a hint to the system.
68(As a convenience to the system, the actual address of the region may differ
69from the address supplied.)
70If
71.Fa addr
72is zero, an address will be selected by the system.
73The actual starting address of the region is returned.
74A successful
75.Fa mmap
76deletes any previous mapping in the allocated address range.
77.Pp
78The protections (region accessibility) are specified in the
79.Fa prot
80argument by
81.Em or Ns 'ing
82the following values:
83.Pp
84.Bl -tag -width PROT_WRITE -compact
85.It Dv PROT_NONE
86Pages may not be accessed.
87.It Dv PROT_READ
88Pages may be read.
89.It Dv PROT_WRITE
90Pages may be written.
91.It Dv PROT_EXEC
92Pages may be executed.
93.El
94.Pp
95The
96.Fa flags
97parameter specifies the type of the mapped object, mapping options and
98whether modifications made to the mapped copy of the page are private
99to the process or are to be shared with other references.
100Sharing, mapping type and options are specified in the
101.Fa flags
102argument by
103.Em or Ns 'ing
104the following values:
105.Bl -tag -width MAP_HASSEMAPHORE
106.It Dv MAP_ANON
107Map anonymous memory not associated with any specific file.
108The file descriptor used for creating
109.Dv MAP_ANON
110must be \-1.
111The
112.Fa offset
113parameter is ignored.
114.\".It Dv MAP_FILE
115.\"Mapped from a regular file or character-special device memory.
116.It Dv MAP_FIXED
117Do not permit the system to select a different address than the one
118specified.
119If the specified address cannot be used,
120.Fn mmap
121will fail.
122If
123.Dv MAP_FIXED
124is specified,
125.Fa addr
126must be a multiple of the pagesize.
127Use of this option is discouraged.
128.It Dv MAP_HASSEMAPHORE
129Notify the kernel that the region may contain semaphores and that special
130handling may be necessary.
131.It Dv MAP_NOCORE
132Region is not included in a core file.
133.It Dv MAP_NOSYNC
134Causes data dirtied via this VM map to be flushed to physical media
135only when necessary (usually by the pager) rather then gratuitously.
136Typically this prevents the update daemons from flushing pages dirtied
137through such maps and thus allows efficient sharing of memory across
138unassociated processes using a file-backed shared memory map. Without
139this option any VM pages you dirty may be flushed to disk every so often
140(every 30-60 seconds usually) which can create performance problems if you
141do not need that to occur (such as when you are using shared file-backed
142mmap regions for IPC purposes). Note that VM/filesystem coherency is
143maintained whether you use
144.Dv MAP_NOSYNC
145or not. This option is not portable
146across
147.Ux
148platforms (yet), though some may implement the same behavior
149by default.
150.Pp
151.Em WARNING !
152Extending a file with
153.Xr ftruncate 2 ,
154thus creating a big hole, and then filling the hole by modifying a shared
155.Fn mmap
156can lead to severe file fragmentation.
157In order to avoid such fragmentation you should always pre-allocate the
158file's backing store by
159.Fn write Ns ing
160zero's into the newly extended area prior to modifying the area via your
161.Fn mmap .
162The fragmentation problem is especially sensitive to
163.Dv MAP_NOSYNC
164pages, because pages may be flushed to disk in a totally random order.
165.Pp
166The same applies when using
167.Dv MAP_NOSYNC
168to implement a file-based shared memory store.
169It is recommended that you create the backing store by
170.Fn write Ns ing
171zero's to the backing file rather then
172.Fn ftruncate Ns ing
173it.
174You can test file fragmentation by observing the KB/t (kilobytes per
175transfer) results from an
176.Dq Li iostat 1
177while reading a large file sequentially, e.g. using
178.Dq Li dd if=filename of=/dev/null bs=32k .
179.Pp
180The
181.Xr fsync 2
182function will flush all dirty data and metadata associated with a file,
183including dirty NOSYNC VM data, to physical media. The
184.Xr sync 8
185command and
186.Xr sync 2
187system call generally do not flush dirty NOSYNC VM data.
188The
189.Xr msync 2
190system call is obsolete since
191.Bx
192implements a coherent filesystem buffer cache. However, it may be
193used to associate dirty VM pages with filesystem buffers and thus cause
194them to be flushed to physical media sooner rather then later.
195.It Dv MAP_PRIVATE
196Modifications are private.
197.It Dv MAP_SHARED
198Modifications are shared.
199.It Dv MAP_STACK
200This option is only available if your system has been compiled with
201.Dv VM_STACK
202defined when compiling the kernel.
203This is the default for
204i386 only.
205Consider adding
206.Li -DVM_STACK
207to
208.Va COPTFLAGS
209in your
210.Pa /etc/make.conf
211to enable this option for other architechures.
212.Dv MAP_STACK
213implies
214.Dv MAP_ANON ,
215and
216.Fa offset
217of 0.
218.Fa fd
219must be -1 and
220.Fa prot
221must include at least
222.Dv PROT_READ
223and
224.Dv PROT_WRITE .
225This option creates
226a memory region that grows to at most
227.Fa len
228bytes in size, starting from the stack top and growing down. The
229stack top is the starting address returned by the call, plus
230.Fa len
231bytes. The bottom of the stack at maximum growth is the starting
232address returned by the call.
233.El
234.Pp
235The
236.Xr close 2
237function does not unmap pages, see
238.Xr munmap 2
239for further information.
240.Pp
241The current design does not allow a process to specify the location of
242swap space.
243In the future we may define an additional mapping type,
244.Dv MAP_SWAP ,
245in which
246the file descriptor argument specifies a file or device to which swapping
247should be done.
248.Sh RETURN VALUES
249Upon successful completion,
250.Fn mmap
251returns a pointer to the mapped region.
252Otherwise, a value of
253.Dv MAP_FAILED
254is returned and
255.Va errno
256is set to indicate the error.
257.Sh ERRORS
258.Fn Mmap
259will fail if:
260.Bl -tag -width Er
261.It Bq Er EACCES
262The flag
263.Dv PROT_READ
264was specified as part of the
265.Fa prot
266parameter and
267.Fa fd
268was not open for reading.
269The flags
270.Dv MAP_SHARED
271and
272.Dv PROT_WRITE
273were specified as part of the
274.Fa flags
275and
276.Fa prot
277parameters and
278.Fa fd
279was not open for writing.
280.It Bq Er EBADF
281.Fa fd
282is not a valid open file descriptor.
283.It Bq Er EINVAL
284.Dv MAP_FIXED
285was specified and the
286.Fa addr
287parameter was not page aligned, or part of the desired address space
288resides out of the valid address space for a user process.
289.It Bq Er EINVAL
290.Fa Len
291was negative.
292.It Bq Er EINVAL
293.Dv MAP_ANON
294was specified and the
295.Fa fd
296parameter was not -1.
297.It Bq Er EINVAL
298.Dv MAP_ANON
299has not been specified and
300.Fa fd
301did not reference a regular or character special file.
302.It Bq Er EINVAL
303.Fa Offset
304was not page-aligned.
305(See
306.Sx BUGS
307below.)
308.It Bq Er ENOMEM
309.Dv MAP_FIXED
310was specified and the
311.Fa addr
312parameter wasn't available.
313.Dv MAP_ANON
314was specified and insufficient memory was available.
315The system has reached the per-process mmap limit specified in the
316.Va vm.max_proc_mmap
317sysctl.
318.El
319.Sh SEE ALSO
320.Xr madvise 2 ,
321.Xr mincore 2 ,
322.Xr mlock 2 ,
323.Xr mprotect 2 ,
324.Xr msync 2 ,
325.Xr munlock 2 ,
326.Xr munmap 2 ,
327.Xr getpagesize 3
328.Sh BUGS
329.Fa len
330is limited to 2GB. Mmapping slightly more than 2GB doesn't work, but
331it is possible to map a window of size (filesize % 2GB) for file sizes
332of slightly less than 2G, 4GB, 6GB and 8GB.
333.Pp
334The limit is imposed for a variety of reasons.
335Most of them have to do
336with
337.Fx
338not wanting to use 64 bit offsets in the VM system due to
339the extreme performance penalty.
340So
341.Fx
342uses 32bit page indexes and
343this gives
344.Fx
345a maximum of 8TB filesizes.
346It's actually bugs in
347the filesystem code that causes the limit to be further restricted to
3481TB (loss of precision when doing blockno calculations).
349.Pp
350Another reason for the 2GB limit is that filesystem metadata can
351reside at negative offsets.