1MMAP(2) Linux Programmer's Manual MMAP(2)
2
6 mmap, munmap - map or unmap files or devices into memory
7
9 #include <sys/mman.h>
10 void *mmap(void *addr, size_t length, int prot, int flags,
12 int fd, off_t offset);
13 int munmap(void *addr, size_t length);
14 See NOTES for information on feature test macro requirements.
16
18 mmap() creates a new mapping in the virtual address space of the call‐
19 ing process. The starting address for the new mapping is specified in
20 addr. The length argument specifies the length of the mapping (which
21 must be greater than 0).
22 If addr is NULL, then the kernel chooses the (page-aligned) address at
24 which to create the mapping; this is the most portable method of creat‐
25 ing a new mapping. If addr is not NULL, then the kernel takes it as a
26 hint about where to place the mapping; on Linux, the kernel will pick a
27 nearby page boundary (but always above or equal to the value specified
28 by /proc/sys/vm/mmap_min_addr) and attempt to create the mapping there.
29 If another mapping already exists there, the kernel picks a new address
30 that may or may not depend on the hint. The address of the new mapping
31 is returned as the result of the call.
32 The contents of a file mapping (as opposed to an anonymous mapping; see
34 MAP_ANONYMOUS below), are initialized using length bytes starting at
35 offset offset in the file (or other object) referred to by the file de‐
36 scriptor fd. offset must be a multiple of the page size as returned by
37 sysconf(_SC_PAGE_SIZE).
38 After the mmap() call has returned, the file descriptor, fd, can be
40 closed immediately without invalidating the mapping.
41 The prot argument describes the desired memory protection of the map‐
43 ping (and must not conflict with the open mode of the file). It is ei‐
44 ther PROT_NONE or the bitwise OR of one or more of the following flags:
45 PROT_EXEC Pages may be executed.
47 PROT_READ Pages may be read.
49 PROT_WRITE Pages may be written.
51 PROT_NONE Pages may not be accessed.
53 The flags argument
55 The flags argument determines whether updates to the mapping are visi‐
56 ble to other processes mapping the same region, and whether updates are
57 carried through to the underlying file. This behavior is determined by
58 including exactly one of the following values in flags:
59 MAP_SHARED
61 Share this mapping. Updates to the mapping are visible to other
62 processes mapping the same region, and (in the case of file-
63 backed mappings) are carried through to the underlying file.
64 (To precisely control when updates are carried through to the
65 underlying file requires the use of msync(2).)
66 MAP_SHARED_VALIDATE (since Linux 4.15)
68 This flag provides the same behavior as MAP_SHARED except that
69 MAP_SHARED mappings ignore unknown flags in flags. By contrast,
70 when creating a mapping using MAP_SHARED_VALIDATE, the kernel
71 verifies all passed flags are known and fails the mapping with
72 the error EOPNOTSUPP for unknown flags. This mapping type is
73 also required to be able to use some mapping flags (e.g.,
74 MAP_SYNC).
75 MAP_PRIVATE
77 Create a private copy-on-write mapping. Updates to the mapping
78 are not visible to other processes mapping the same file, and
79 are not carried through to the underlying file. It is unspeci‐
80 fied whether changes made to the file after the mmap() call are
81 visible in the mapped region.
82 Both MAP_SHARED and MAP_PRIVATE are described in POSIX.1-2001 and
84 POSIX.1-2008. MAP_SHARED_VALIDATE is a Linux extension.
85 In addition, zero or more of the following values can be ORed in flags:
87 MAP_32BIT (since Linux 2.4.20, 2.6)
89 Put the mapping into the first 2 Gigabytes of the process ad‐
90 dress space. This flag is supported only on x86-64, for 64-bit
91 programs. It was added to allow thread stacks to be allocated
92 somewhere in the first 2 GB of memory, so as to improve context-
93 switch performance on some early 64-bit processors. Modern
94 x86-64 processors no longer have this performance problem, so
95 use of this flag is not required on those systems. The
96 MAP_32BIT flag is ignored when MAP_FIXED is set.
97 MAP_ANON
99 Synonym for MAP_ANONYMOUS; provided for compatibility with other
100 implementations.
101 MAP_ANONYMOUS
103 The mapping is not backed by any file; its contents are initial‐
104 ized to zero. The fd argument is ignored; however, some imple‐
105 mentations require fd to be -1 if MAP_ANONYMOUS (or MAP_ANON) is
106 specified, and portable applications should ensure this. The
107 offset argument should be zero. The use of MAP_ANONYMOUS in
108 conjunction with MAP_SHARED is supported on Linux only since
109 kernel 2.4.
110 MAP_DENYWRITE
112 This flag is ignored. (Long ago—Linux 2.0 and earlier—it sig‐
113 naled that attempts to write to the underlying file should fail
114 with ETXTBSY. But this was a source of denial-of-service at‐
115 tacks.)
116 MAP_EXECUTABLE
118 This flag is ignored.
119 MAP_FILE
121 Compatibility flag. Ignored.
122 MAP_FIXED
124 Don't interpret addr as a hint: place the mapping at exactly
125 that address. addr must be suitably aligned: for most architec‐
126 tures a multiple of the page size is sufficient; however, some
127 architectures may impose additional restrictions. If the memory
128 region specified by addr and len overlaps pages of any existing
129 mapping(s), then the overlapped part of the existing mapping(s)
130 will be discarded. If the specified address cannot be used,
131 mmap() will fail.
132 Software that aspires to be portable should use the MAP_FIXED
134 flag with care, keeping in mind that the exact layout of a
135 process's memory mappings is allowed to change significantly be‐
136 tween kernel versions, C library versions, and operating system
137 releases. Carefully read the discussion of this flag in NOTES!
138 MAP_FIXED_NOREPLACE (since Linux 4.17)
140 This flag provides behavior that is similar to MAP_FIXED with
141 respect to the addr enforcement, but differs in that
142 MAP_FIXED_NOREPLACE never clobbers a preexisting mapped range.
143 If the requested range would collide with an existing mapping,
144 then this call fails with the error EEXIST. This flag can
145 therefore be used as a way to atomically (with respect to other
146 threads) attempt to map an address range: one thread will suc‐
147 ceed; all others will report failure.
148 Note that older kernels which do not recognize the
150 MAP_FIXED_NOREPLACE flag will typically (upon detecting a colli‐
151 sion with a preexisting mapping) fall back to a "non-MAP_FIXED"
152 type of behavior: they will return an address that is different
153 from the requested address. Therefore, backward-compatible
154 software should check the returned address against the requested
155 address.
156 MAP_GROWSDOWN
158 This flag is used for stacks. It indicates to the kernel vir‐
159 tual memory system that the mapping should extend downward in
160 memory. The return address is one page lower than the memory
161 area that is actually created in the process's virtual address
162 space. Touching an address in the "guard" page below the map‐
163 ping will cause the mapping to grow by a page. This growth can
164 be repeated until the mapping grows to within a page of the high
165 end of the next lower mapping, at which point touching the
166 "guard" page will result in a SIGSEGV signal.
167 MAP_HUGETLB (since Linux 2.6.32)
169 Allocate the mapping using "huge" pages. See the Linux kernel
170 source file Documentation/admin-guide/mm/hugetlbpage.rst for
171 further information, as well as NOTES, below.
172 MAP_HUGE_2MB, MAP_HUGE_1GB (since Linux 3.8)
174 Used in conjunction with MAP_HUGETLB to select alternative
175 hugetlb page sizes (respectively, 2 MB and 1 GB) on systems that
176 support multiple hugetlb page sizes.
177 More generally, the desired huge page size can be configured by
179 encoding the base-2 logarithm of the desired page size in the
180 six bits at the offset MAP_HUGE_SHIFT. (A value of zero in this
181 bit field provides the default huge page size; the default huge
182 page size can be discovered via the Hugepagesize field exposed
183 by /proc/meminfo.) Thus, the above two constants are defined
184 as:
185 #define MAP_HUGE_2MB (21 << MAP_HUGE_SHIFT)
187 #define MAP_HUGE_1GB (30 << MAP_HUGE_SHIFT)
188 The range of huge page sizes that are supported by the system
190 can be discovered by listing the subdirectories in /sys/ker‐
191 nel/mm/hugepages.
192 MAP_LOCKED (since Linux 2.5.37)
194 Mark the mapped region to be locked in the same way as mlock(2).
195 This implementation will try to populate (prefault) the whole
196 range but the mmap() call doesn't fail with ENOMEM if this
197 fails. Therefore major faults might happen later on. So the
198 semantic is not as strong as mlock(2). One should use mmap()
199 plus mlock(2) when major faults are not acceptable after the
200 initialization of the mapping. The MAP_LOCKED flag is ignored
201 in older kernels.
202 MAP_NONBLOCK (since Linux 2.5.46)
204 This flag is meaningful only in conjunction with MAP_POPULATE.
205 Don't perform read-ahead: create page tables entries only for
206 pages that are already present in RAM. Since Linux 2.6.23, this
207 flag causes MAP_POPULATE to do nothing. One day, the combina‐
208 tion of MAP_POPULATE and MAP_NONBLOCK may be reimplemented.
209 MAP_NORESERVE
211 Do not reserve swap space for this mapping. When swap space is
212 reserved, one has the guarantee that it is possible to modify
213 the mapping. When swap space is not reserved one might get
214 SIGSEGV upon a write if no physical memory is available. See
215 also the discussion of the file /proc/sys/vm/overcommit_memory
216 in proc(5). In kernels before 2.6, this flag had effect only
217 for private writable mappings.
218 MAP_POPULATE (since Linux 2.5.46)
220 Populate (prefault) page tables for a mapping. For a file map‐
221 ping, this causes read-ahead on the file. This will help to re‐
222 duce blocking on page faults later. The mmap() call doesn't
223 fail if the mapping cannot be populated (for example, due to
224 limitations on the number of mapped huge pages when using
225 MAP_HUGETLB). MAP_POPULATE is supported for private mappings
226 only since Linux 2.6.23.
227 MAP_STACK (since Linux 2.6.27)
229 Allocate the mapping at an address suitable for a process or
230 thread stack.
231 This flag is currently a no-op on Linux. However, by employing
233 this flag, applications can ensure that they transparently ob‐
234 tain support if the flag is implemented in the future. Thus, it
235 is used in the glibc threading implementation to allow for the
236 fact that some architectures may (later) require special treat‐
237 ment for stack allocations. A further reason to employ this
238 flag is portability: MAP_STACK exists (and has an effect) on
239 some other systems (e.g., some of the BSDs).
240 MAP_SYNC (since Linux 4.15)
242 This flag is available only with the MAP_SHARED_VALIDATE mapping
243 type; mappings of type MAP_SHARED will silently ignore this
244 flag. This flag is supported only for files supporting DAX (di‐
245 rect mapping of persistent memory). For other files, creating a
246 mapping with this flag results in an EOPNOTSUPP error.
247 Shared file mappings with this flag provide the guarantee that
249 while some memory is mapped writable in the address space of the
250 process, it will be visible in the same file at the same offset
251 even after the system crashes or is rebooted. In conjunction
252 with the use of appropriate CPU instructions, this provides
253 users of such mappings with a more efficient way of making data
254 modifications persistent.
255 MAP_UNINITIALIZED (since Linux 2.6.33)
257 Don't clear anonymous pages. This flag is intended to improve
258 performance on embedded devices. This flag is honored only if
259 the kernel was configured with the CONFIG_MMAP_ALLOW_UNINITIAL‐
260 IZED option. Because of the security implications, that option
261 is normally enabled only on embedded devices (i.e., devices
262 where one has complete control of the contents of user memory).
263 Of the above flags, only MAP_FIXED is specified in POSIX.1-2001 and
265 POSIX.1-2008. However, most systems also support MAP_ANONYMOUS (or its
266 synonym MAP_ANON).
267 munmap()
269 The munmap() system call deletes the mappings for the specified address
270 range, and causes further references to addresses within the range to
271 generate invalid memory references. The region is also automatically
272 unmapped when the process is terminated. On the other hand, closing
273 the file descriptor does not unmap the region.
274 The address addr must be a multiple of the page size (but length need
276 not be). All pages containing a part of the indicated range are un‐
277 mapped, and subsequent references to these pages will generate SIGSEGV.
278 It is not an error if the indicated range does not contain any mapped
279 pages.
280
282 On success, mmap() returns a pointer to the mapped area. On error, the
283 value MAP_FAILED (that is, (void *) -1) is returned, and errno is set
284 to indicate the error.
285 On success, munmap() returns 0. On failure, it returns -1, and errno
287 is set to indicate the error (probably to EINVAL).
288
290 EACCES A file descriptor refers to a non-regular file. Or a file map‐
291 ping was requested, but fd is not open for reading. Or
292 MAP_SHARED was requested and PROT_WRITE is set, but fd is not
293 open in read/write (O_RDWR) mode. Or PROT_WRITE is set, but the
294 file is append-only.
295 EAGAIN The file has been locked, or too much memory has been locked
297 (see setrlimit(2)).
298 EBADF fd is not a valid file descriptor (and MAP_ANONYMOUS was not
300 set).
301 EEXIST MAP_FIXED_NOREPLACE was specified in flags, and the range cov‐
303 ered by addr and length clashes with an existing mapping.
304 EINVAL We don't like addr, length, or offset (e.g., they are too large,
306 or not aligned on a page boundary).
307 EINVAL (since Linux 2.6.12) length was 0.
309 EINVAL flags contained none of MAP_PRIVATE, MAP_SHARED, or
311 MAP_SHARED_VALIDATE.
312 ENFILE The system-wide limit on the total number of open files has been
314 reached.
315 ENODEV The underlying filesystem of the specified file does not support
317 memory mapping.
318 ENOMEM No memory is available.
320 ENOMEM The process's maximum number of mappings would have been ex‐
322 ceeded. This error can also occur for munmap(), when unmapping
323 a region in the middle of an existing mapping, since this re‐
324 sults in two smaller mappings on either side of the region being
325 unmapped.
326 ENOMEM (since Linux 4.7) The process's RLIMIT_DATA limit, described in
328 getrlimit(2), would have been exceeded.
329 EOVERFLOW
331 On 32-bit architecture together with the large file extension
332 (i.e., using 64-bit off_t): the number of pages used for length
333 plus number of pages used for offset would overflow unsigned
334 long (32 bits).
335 EPERM The prot argument asks for PROT_EXEC but the mapped area belongs
337 to a file on a filesystem that was mounted no-exec.
338 EPERM The operation was prevented by a file seal; see fcntl(2).
340 EPERM The MAP_HUGETLB flag was specified, but the caller was not priv‐
342 ileged (did not have the CAP_IPC_LOCK capability) and is not a
343 member of the sysctl_hugetlb_shm_group group; see the descrip‐
344 tion of /proc/sys/vm/sysctl_hugetlb_shm_group in
345 ETXTBSY
347 MAP_DENYWRITE was set but the object specified by fd is open for
348 writing.
349 Use of a mapped region can result in these signals:
351 SIGSEGV
353 Attempted write into a region mapped as read-only.
354 SIGBUS Attempted access to a page of the buffer that lies beyond the
356 end of the mapped file. For an explanation of the treatment of
357 the bytes in the page that corresponds to the end of a mapped
358 file that is not a multiple of the page size, see NOTES.
359
361 For an explanation of the terms used in this section, see at‐
362 tributes(7).
363 ┌────────────────────────────────────────────┬───────────────┬─────────┐
365 │Interface │ Attribute │ Value │
366 ├────────────────────────────────────────────┼───────────────┼─────────┤
367 │mmap(), munmap() │ Thread safety │ MT-Safe │
368 └────────────────────────────────────────────┴───────────────┴─────────┘
369
371 POSIX.1-2001, POSIX.1-2008, SVr4, 4.4BSD.
372 On POSIX systems on which mmap(), msync(2), and munmap() are available,
374 _POSIX_MAPPED_FILES is defined in <unistd.h> to a value greater than 0.
375 (See also sysconf(3).)
376
378 Memory mapped by mmap() is preserved across fork(2), with the same at‐
379 tributes.
380 A file is mapped in multiples of the page size. For a file that is not
382 a multiple of the page size, the remaining bytes in the partial page at
383 the end of the mapping are zeroed when mapped, and modifications to
384 that region are not written out to the file. The effect of changing
385 the size of the underlying file of a mapping on the pages that corre‐
386 spond to added or removed regions of the file is unspecified.
387 On some hardware architectures (e.g., i386), PROT_WRITE implies
389 PROT_READ. It is architecture dependent whether PROT_READ implies
390 PROT_EXEC or not. Portable programs should always set PROT_EXEC if
391 they intend to execute code in the new mapping.
392 The portable way to create a mapping is to specify addr as 0 (NULL),
394 and omit MAP_FIXED from flags. In this case, the system chooses the
395 address for the mapping; the address is chosen so as not to conflict
396 with any existing mapping, and will not be 0. If the MAP_FIXED flag is
397 specified, and addr is 0 (NULL), then the mapped address will be 0
398 (NULL).
399 Certain flags constants are defined only if suitable feature test
401 macros are defined (possibly by default): _DEFAULT_SOURCE with glibc
402 2.19 or later; or _BSD_SOURCE or _SVID_SOURCE in glibc 2.19 and earli‐
403 er. (Employing _GNU_SOURCE also suffices, and requiring that macro
404 specifically would have been more logical, since these flags are all
405 Linux-specific.) The relevant flags are: MAP_32BIT, MAP_ANONYMOUS (and
406 the synonym MAP_ANON), MAP_DENYWRITE, MAP_EXECUTABLE, MAP_FILE,
407 MAP_GROWSDOWN, MAP_HUGETLB, MAP_LOCKED, MAP_NONBLOCK, MAP_NORESERVE,
408 MAP_POPULATE, and MAP_STACK.
409 An application can determine which pages of a mapping are currently
411 resident in the buffer/page cache using mincore(2).
412 Using MAP_FIXED safely
414 The only safe use for MAP_FIXED is where the address range specified by
415 addr and length was previously reserved using another mapping; other‐
416 wise, the use of MAP_FIXED is hazardous because it forcibly removes
417 preexisting mappings, making it easy for a multithreaded process to
418 corrupt its own address space.
419 For example, suppose that thread A looks through /proc/<pid>/maps in
421 order to locate an unused address range that it can map using
422 MAP_FIXED, while thread B simultaneously acquires part or all of that
423 same address range. When thread A subsequently employs
424 mmap(MAP_FIXED), it will effectively clobber the mapping that thread B
425 created. In this scenario, thread B need not create a mapping direct‐
426 ly; simply making a library call that, internally, uses dlopen(3) to
427 load some other shared library, will suffice. The dlopen(3) call will
428 map the library into the process's address space. Furthermore, almost
429 any library call may be implemented in a way that adds memory mappings
430 to the address space, either with this technique, or by simply allocat‐
431 ing memory. Examples include brk(2), malloc(3), pthread_create(3), and
432 the PAM libraries ⟨http://www.linux-pam.org⟩.
433 Since Linux 4.17, a multithreaded program can use the MAP_FIXED_NORE‐
435 PLACE flag to avoid the hazard described above when attempting to cre‐
436 ate a mapping at a fixed address that has not been reserved by a preex‐
437 isting mapping.
438 Timestamps changes for file-backed mappings
440 For file-backed mappings, the st_atime field for the mapped file may be
441 updated at any time between the mmap() and the corresponding unmapping;
442 the first reference to a mapped page will update the field if it has
443 not been already.
444 The st_ctime and st_mtime field for a file mapped with PROT_WRITE and
446 MAP_SHARED will be updated after a write to the mapped region, and be‐
447 fore a subsequent msync(2) with the MS_SYNC or MS_ASYNC flag, if one
448 occurs.
449 Huge page (Huge TLB) mappings
451 For mappings that employ huge pages, the requirements for the arguments
452 of mmap() and munmap() differ somewhat from the requirements for map‐
453 pings that use the native system page size.
454 For mmap(), offset must be a multiple of the underlying huge page size.
456 The system automatically aligns length to be a multiple of the underly‐
457 ing huge page size.
458 For munmap(), addr, and length must both be a multiple of the underly‐
460 ing huge page size.
461 C library/kernel differences
463 This page describes the interface provided by the glibc mmap() wrapper
464 function. Originally, this function invoked a system call of the same
465 name. Since kernel 2.4, that system call has been superseded by
466 mmap2(2), and nowadays the glibc mmap() wrapper function invokes
467 mmap2(2) with a suitably adjusted value for offset.
468
470 On Linux, there are no guarantees like those suggested above under
471 MAP_NORESERVE. By default, any process can be killed at any moment
472 when the system runs out of memory.
473 In kernels before 2.6.7, the MAP_POPULATE flag has effect only if prot
475 is specified as PROT_NONE.
476 SUSv3 specifies that mmap() should fail if length is 0. However, in
478 kernels before 2.6.12, mmap() succeeded in this case: no mapping was
479 created and the call returned addr. Since kernel 2.6.12, mmap() fails
480 with the error EINVAL for this case.
481 POSIX specifies that the system shall always zero fill any partial page
483 at the end of the object and that system will never write any modifica‐
484 tion of the object beyond its end. On Linux, when you write data to
485 such partial page after the end of the object, the data stays in the
486 page cache even after the file is closed and unmapped and even though
487 the data is never written to the file itself, subsequent mappings may
488 see the modified content. In some cases, this could be fixed by call‐
489 ing msync(2) before the unmap takes place; however, this doesn't work
490 on tmpfs(5) (for example, when using the POSIX shared memory interface
491 documented in shm_overview(7)).
492
494 The following program prints part of the file specified in its first
495 command-line argument to standard output. The range of bytes to be
496 printed is specified via offset and length values in the second and
497 third command-line arguments. The program creates a memory mapping of
498 the required pages of the file and then uses write(2) to output the de‐
499 sired bytes.
500 Program source
502 #include <sys/mman.h>
503 #include <sys/stat.h>
504 #include <fcntl.h>
505 #include <stdio.h>
506 #include <stdlib.h>
507 #include <unistd.h>
508 #define handle_error(msg) \
510 do { perror(msg); exit(EXIT_FAILURE); } while (0)
511 int
513 main(int argc, char *argv[])
514 {
515 char *addr;
516 int fd;
517 struct stat sb;
518 off_t offset, pa_offset;
519 size_t length;
520 ssize_t s;
521 if (argc < 3 || argc > 4) {
523 fprintf(stderr, "%s file offset [length]\n", argv[0]);
524 exit(EXIT_FAILURE);
525 }
526 fd = open(argv[1], O_RDONLY);
528 if (fd == -1)
529 handle_error("open");
530 if (fstat(fd, &sb) == -1) /* To obtain file size */
532 handle_error("fstat");
533 offset = atoi(argv[2]);
535 pa_offset = offset & ~(sysconf(_SC_PAGE_SIZE) - 1);
536 /* offset for mmap() must be page aligned */
537 if (offset >= sb.st_size) {
539 fprintf(stderr, "offset is past end of file\n");
540 exit(EXIT_FAILURE);
541 }
542 if (argc == 4) {
544 length = atoi(argv[3]);
545 if (offset + length > sb.st_size)
546 length = sb.st_size - offset;
547 /* Can't display bytes past end of file */
548 } else { /* No length arg ==> display to end of file */
550 length = sb.st_size - offset;
551 }
552 addr = mmap(NULL, length + offset - pa_offset, PROT_READ,
554 MAP_PRIVATE, fd, pa_offset);
555 if (addr == MAP_FAILED)
556 handle_error("mmap");
557 s = write(STDOUT_FILENO, addr + offset - pa_offset, length);
559 if (s != length) {
560 if (s == -1)
561 handle_error("write");
562 fprintf(stderr, "partial write");
564 exit(EXIT_FAILURE);
565 }
566 munmap(addr, length + offset - pa_offset);
568 close(fd);
569 exit(EXIT_SUCCESS);
571 }
572
574 ftruncate(2), getpagesize(2), memfd_create(2), mincore(2), mlock(2),
575 mmap2(2), mprotect(2), mremap(2), msync(2), remap_file_pages(2), setr‐
576 limit(2), shmat(2), userfaultfd(2), shm_open(3), shm_overview(7)
577 The descriptions of the following files in proc(5): /proc/[pid]/maps,
579 /proc/[pid]/map_files, and /proc/[pid]/smaps.
580 B.O. Gallmeister, POSIX.4, O'Reilly, pp. 128–129 and 389–391.
582
584 This page is part of release 5.12 of the Linux man-pages project. A
585 description of the project, information about reporting bugs, and the
586 latest version of this page, can be found at
587 https://www.kernel.org/doc/man-pages/.
588Linux 2021-03-22 MMAP(2)