MMAP(2) Linux Programmer's Manual MMAP(2)
NAME
mmap, munmap - map or unmap files or devices into memory
SYNOPSIS
#include <sys/mman.h>
void * mmap(void *start, size_t length, int prot , int flags, int fd,
off_t offset);
int munmap(void *start, size_t length);
DESCRIPTION
The mmap function asks to map length bytes starting at offset offset
from the file(1,n) (or other object) specified by the file(1,n) descriptor fd
into memory, preferably at address start. This latter address is a
hint only, and is usually specified as 0. The actual place where the
object is mapped is returned by mmap, and is never 0.
The prot argument describes the desired memory protection (and must not
conflict with the open(2,3,n) mode of the file(1,n)). It is either PROT_NONE or is
the bitwise OR of one or more of the other PROT_* flags.
PROT_EXEC Pages may be executed.
PROT_READ Pages may be read.
PROT_WRITE Pages may be written.
PROT_NONE Pages may not be accessed.
The flags parameter specifies the type of the mapped object, mapping
options and whether modifications made to the mapped copy of the page
are private to the process or are to be shared with other references.
It has bits
MAP_FIXED Do not select(2,7,2 select_tut) a different address than the one specified.
If the specified address cannot be used, mmap will fail. If
MAP_FIXED is specified, start must be a multiple of the
pagesize. Use of this option is discouraged.
MAP_SHARED Share this mapping with all other processes that map this
object. Storing to the region is equivalent to writing to
the file. The file(1,n) may not actually be updated until
msync(2) or munmap(2) are called.
MAP_PRIVATE
Create a private copy-on-write mapping. Stores to the
region do not affect the original file. It is unspecified
whether changes made to the file(1,n) after the mmap call are
visible in(1,8) the mapped region.
You must specify exactly one of MAP_SHARED and MAP_PRIVATE.
The above three flags are described in(1,8) POSIX.1b (formerly POSIX.4) and
SUSv2. Linux also knows about the following non-standard flags:
MAP_DENYWRITE
This flag is ignored. (Long ago, it signalled that attempts to
write(1,2) to the underlying file(1,n) should fail with ETXTBUSY. But this
was a source of denial-of-service attacks.)
MAP_EXECUTABLE
This flag is ignored.
MAP_NORESERVE
(Used together with MAP_PRIVATE.) Do not reserve swap space
pages for this mapping. When swap space is reserved, one has the
guarantee that it is possible to modify this private copy-on-
write(1,2) region. When it is not reserved one might get SIGSEGV
upon a write(1,2) when no memory is available.
MAP_LOCKED (since Linux 2.5.37)
Lock the pages of the mapped region into memory in(1,8) the manner of
mlock(). This flag is ignored in(1,8) older kernels.
MAP_GROWSDOWN
Used for stacks. Indicates to the kernel VM system that the map-
ping should extend downwards in(1,8) memory.
MAP_ANONYMOUS
The mapping is not backed by any file(1,n); the fd and offset argu-
ments are ignored. This flag in(1,8) conjunction with MAP_SHARED is
implemented since Linux 2.4.
MAP_ANON
Alias for MAP_ANONYMOUS. Deprecated.
MAP_FILE
Compatibility flag. Ignored.
MAP_32BIT
Put the mapping into the first 2GB of the process address space.
Ignored when MAP_FIXED is set. This flag is currently only sup-
ported on x86-64 for 64bit programs.
MAP_POPULATE (since Linux 2.5.46)
Populate (prefault) pagetables.
MAP_NONBLOCK (since Linux 2.5.46)
Do not block on IO.
Some systems document the additional flags MAP_AUTOGROW, MAP_AUTORESRV,
MAP_COPY, and MAP_LOCAL.
fd should be a valid file(1,n) descriptor, unless MAP_ANONYMOUS is set(7,n,1 builtins), in(1,8)
which case the argument is ignored.
offset should be a multiple of the page size as returned by getpage-
size(2).
Memory mapped by mmap is preserved across fork(2), with the same
attributes.
A file(1,n) is mapped in(1,8) multiples of the page size. For a file(1,n) that is not
a multiple of the page size, the remaining memory is zeroed when
mapped, and writes to that region are not written out to the file. The
effect of changing the size of the underlying file(1,n) of a mapping on the
pages that correspond to added or removed regions of the file(1,n) is
unspecified.
The munmap system call deletes the mappings for the specified address
range, and causes further references to addresses within the range to
generate invalid memory references. The region is also automatically
unmapped when the process is terminated. On the other hand, closing
the file(1,n) descriptor does not unmap the region.
The address start must be a multiple of the page size. All pages con-
taining a part of the indicated range are unmapped, and subsequent ref-
erences to these pages will generate SIGSEGV. It is not an error(8,n) if(3,n) the
indicated range does not contain any mapped pages.
For file-backed mappings, the st_atime field for the mapped file(1,n) may be
updated at any time(1,2,n) between the mmap() and the corresponding unmapping;
the first reference to a mapped page will update(7,n) the field if(3,n) it has
not been already.
The st_ctime and st_mtime field for a file(1,n) mapped with PROT_WRITE and
MAP_SHARED will be updated after a write(1,2) to the mapped region, and
before a subsequent msync() with the MS_SYNC or MS_ASYNC flag, if(3,n) one
occurs.
RETURN VALUE
On success, mmap returns a pointer to the mapped area. On error(8,n), the
value MAP_FAILED (that is, (void *) -1) is returned, and errno is set(7,n,1 builtins)
appropriately. On success, munmap returns 0, on failure -1, and errno
is set(7,n,1 builtins) (probably to EINVAL).
NOTES
It is architecture dependent whether PROT_READ includes PROT_EXEC or
not. Portable programs should always set(7,n,1 builtins) PROT_EXEC if(3,n) they intend to
execute code in(1,8) the new mapping.
ERRORS
EACCES A file(1,n) descriptor refers to a non-regular file. Or MAP_PRIVATE
was requested, but fd is not open(2,3,n) for reading. Or MAP_SHARED
was requested and PROT_WRITE is set(7,n,1 builtins), but fd is not open(2,3,n) in(1,8)
read(2,n,1 builtins)/write(1,2) (O_RDWR) mode. Or PROT_WRITE is set(7,n,1 builtins), but the file(1,n) is
append-only.
EAGAIN The file(1,n) has been locked, or too much memory has been locked.
EBADF fd is not a valid file(1,n) descriptor (and MAP_ANONYMOUS was not
set(7,n,1 builtins)).
EINVAL We don't like start or length or offset. (E.g., they are too
large, or not aligned on a PAGESIZE boundary.)
ENFILE The system limit on the total number of open(2,3,n) files has been
reached.
ENODEV The underlying filesystem of the specified file(1,n) does not support
memory mapping.
ENOMEM No memory is available, or the process's maximum number of map-
pings would have been exceeded.
EPERM The prot argument asks for PROT_EXEC but the mapped area belongs
to a file(1,n) on a filesystem that was mounted no-exec.
ETXTBSY
MAP_DENYWRITE was set(7,n,1 builtins) but the object specified by fd is open(2,3,n) for
writing.
Use of a mapped region can result in(1,8) these signals:
SIGSEGV
Attempted write(1,2) into a region specified to mmap as read-only.
SIGBUS Attempted access(2,5) to a portion of the buffer that does not corre-
spond to the file(1,n) (for example, beyond the end of the file(1,n),
including the case where another process has truncated the
file(1,n)).
AVAILABILITY
On POSIX systems on which mmap, msync and munmap are available,
_POSIX_MAPPED_FILES is defined in(1,8) <unistd.h> to a value greater than 0.
(See also sysconf(3).)
CONFORMING TO
SVr4, POSIX.1b (formerly POSIX.4), 4.4BSD, SUSv2. SVr4 documents addi-
tional error(8,n) codes ENXIO and ENODEV. SUSv2 documents additional error(8,n)
codes EMFILE and EOVERFLOW.
BUGS
On Linux there are no guarantees like those suggested above under
MAP_NORESERVE. By default, any process can be killed at any moment when
the system runs out of memory.
SEE ALSO
getpagesize(2), mlock(2), mmap2(2), mremap(2), msync(2), shm_open(2)
B.O. Gallmeister, POSIX.4, O'Reilly, pp. 128-129 and 389-391.
Linux 2.6.7 2004-09-11 MMAP(2)