MMAP(2) BSD System Calls Manual MMAP(2)NAMEmmap — allocate memory, or map files or devices into memory
LIBRARY
Standard C Library (libc, -lc)
SYNOPSIS
#include <sys/types.h>
#include <sys/mman.h>
void *
mmap(void *addr, size_t len, int prot, int flags, int fd, off_t offset);
DESCRIPTION
The mmap() function causes the pages starting at addr and continuing for
at most len bytes to be mapped from the object described by fd, starting
at byte offset offset. If len is not a multiple of the pagesize, the
mapped region may extend past the specified range. Any such extension
beyond the end of the mapped object will be zero-filled.
If addr is non-zero, it is used as a hint to the system. (As a conve‐
nience to the system, the actual address of the region may differ from
the address supplied.) If addr is zero, an address will be selected by
the system. The actual starting address of the region is returned. A
successful mmap deletes any previous mapping in the allocated address
range.
The protections (region accessibility) are specified in the prot argument
by or'ing the following values:
PROT_NONE Pages may not be accessed.
PROT_READ Pages may be read.
PROT_WRITE Pages may be written.
PROT_EXEC Pages may be executed.
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. Shar‐
ing, mapping type and options are specified in the flags argument by
or'ing the following values:
MAP_ANON Map anonymous memory not associated with any specific
file. The file descriptor used for creating MAP_ANON
must be -1. The offset parameter is ignored.
MAP_FIXED Do not permit the system to select a different address
than the one specified. If the specified address con‐
tains other mappings those mappings will be replaced.
If the specified address cannot otherwise be used,
mmap() will fail. If MAP_FIXED is specified, addr must
be a multiple of the pagesize.
MAP_TRYFIXED Try to do a fixed mapping but fail if another mapping
already exists in the space instead of overwriting the
mapping.
When used with MAP_STACK this flag allows one MAP_STACK
mapping to be made within another (typically the master
user stack), as long as no pages have been faulted in
the area requested.
MAP_HASSEMAPHORE Notify the kernel that the region may contain sema‐
phores and that special handling may be necessary.
MAP_NOCORE Region is not included in a core file.
MAP_NOSYNC Causes data dirtied via this VM map to be flushed to
physical media only when necessary (usually by the
pager) rather than gratuitously. Typically this pre‐
vents the update daemons from flushing pages dirtied
through such maps and thus allows efficient sharing of
memory across unassociated processes using a file-
backed shared memory map. Without this option any VM
pages you dirty may be flushed to disk every so often
(every 30-60 seconds usually) which can create perfor‐
mance problems if you do not need that to occur (such
as when you are using shared file-backed mmap regions
for IPC purposes). Note that VM/filesystem coherency
is maintained whether you use MAP_NOSYNC or not. This
option is not portable across UNIX platforms (yet),
though some may implement the same behavior by default.
WARNING! Extending a file with ftruncate(2), thus cre‐
ating a big hole, and then filling the hole by modify‐
ing a shared mmap() can lead to severe file fragmenta‐
tion. In order to avoid such fragmentation you should
always pre-allocate the file's backing store by
write()ing zero's into the newly extended area prior to
modifying the area via your mmap(). The fragmentation
problem is especially sensitive to MAP_NOSYNC pages,
because pages may be flushed to disk in a totally ran‐
dom order.
The same applies when using MAP_NOSYNC to implement a
file-based shared memory store. It is recommended that
you create the backing store by write()ing zero's to
the backing file rather than ftruncate()ing it. You
can test file fragmentation by observing the KB/t
(kilobytes per transfer) results from an “iostat 1”
while reading a large file sequentially, e.g. using “dd
if=filename of=/dev/null bs=32k”.
The fsync(2) function will flush all dirty data and
metadata associated with a file, including dirty NOSYNC
VM data, to physical media. The sync(8) command and
sync(2) system call generally do not flush dirty NOSYNC
VM data. The msync(2) system call is obsolete since
BSD implements a coherent filesystem buffer cache.
However, it may be used to associate dirty VM pages
with filesystem buffers and thus cause them to be
flushed to physical media sooner rather than later.
MAP_PRIVATE Modifications are private.
MAP_SHARED Modifications are shared.
MAP_STACK Map the area as a stack. MAP_ANON is implied. Offset
should be 0, fd must be -1, and prot should include at
least PROT_READ and PROT_WRITE. This option creates a
memory region that grows to at most len bytes in size,
starting from the stack top and growing down. The
stack top is the starting address returned by the call,
plus len bytes. The bottom of the stack at maximum
growth is the starting address returned by the call.
The entire area is reserved from the point of view of
other mmap() calls, even if not faulted in yet.
WARNING. We currently allow MAP_STACK mappings to pro‐
vide a hint that points within an existing MAP_STACK
mapping's space, and this will succeed as long as no
page have been faulted in the area specified, but this
behavior is no longer supported unless you also specify
the MAP_TRYFIXED flag.
Note that unless MAP_FIXED or MAP_TRYFIXED is used, you
cannot count on the returned address matching the hint
you have provided.
MAP_VPAGETABLE Memory accessed via this map is not linearly mapped and
will be governed by a virtual page table. The base
address of the virtual page table may be set using
mcontrol(2) with MADV_SETMAP. Virtual page tables work
with anonymous memory but there is no way to populate
the page table so for all intents and purposes
MAP_VPAGETABLE can only be used when mapping file
descriptors. Since the kernel will update the VPTE_M
bit in the virtual page table, the mapping must R+W
even though actual access to the memory will be prop‐
erly governed by the virtual page table.
Addressable backing store is limited by the range sup‐
ported in the virtual page table entries. The kernel
may implement a page table abstraction capable of
addressing a larger range within the backing store then
could otherwise be mapped into memory.
The close(2) function does not unmap pages, see munmap(2) for further
information.
The current design does not allow a process to specify the location of
swap space. In the future we may define an additional mapping type,
MAP_SWAP, in which the file descriptor argument specifies a file or
device to which swapping should be done.
RETURN VALUES
Upon successful completion, mmap() returns a pointer to the mapped
region. Otherwise, a value of MAP_FAILED is returned and errno is set to
indicate the error.
ERRORSMmap() will fail if:
[EACCES] The flag PROT_READ was specified as part of the prot
parameter and fd was not open for reading. The flags
MAP_SHARED and PROT_WRITE were specified as part of
the flags and prot parameters and fd was not open for
writing.
[EBADF] fd is not a valid open file descriptor.
[EINVAL] MAP_FIXED was specified and the addr parameter was not
page aligned, or part of the desired address space
resides out of the valid address space for a user
process.
[EINVAL] Len was negative.
[EINVAL] MAP_ANON was specified and the fd parameter was not
-1.
[EINVAL] MAP_ANON has not been specified and fd did not refer‐
ence a regular or character special file.
[EINVAL] Offset was not page-aligned. (See BUGS below.)
[ENOMEM] MAP_FIXED was specified and the addr parameter wasn't
available. MAP_ANON was specified and insufficient
memory was available. The system has reached the per-
process mmap limit specified in the vm.max_proc_mmap
sysctl.
SEE ALSOmadvise(2), mincore(2), mlock(2), mprotect(2), msync(2), munlock(2),
munmap(2), getpagesize(3)BUGS
len is limited to 2GB. Mmapping slightly more than 2GB doesn't work, but
it is possible to map a window of size (filesize % 2GB) for file sizes of
slightly less than 2G, 4GB, 6GB and 8GB.
The limit is imposed for a variety of reasons. Most of them have to do
with DragonFly not wanting to use 64 bit offsets in the VM system due to
the extreme performance penalty. So DragonFly uses 32bit page indexes
and this gives DragonFly a maximum of 8TB filesizes. It's actually bugs
in the filesystem code that causes the limit to be further restricted to
1TB (loss of precision when doing blockno calculations).
Another reason for the 2GB limit is that filesystem metadata can reside
at negative offsets.
BSD December 11, 2006 BSD