.\"
.\" Copyright 2000 Massachusetts Institute of Technology
.\"
.\" Permission to use, copy, modify, and distribute this software and
.\" its documentation for any purpose and without fee is hereby
.\" granted, provided that both the above copyright notice and this
.\" permission notice appear in all copies, that both the above
.\" copyright notice and this permission notice appear in all
.\" supporting documentation, and that the name of M.I.T. not be used
.\" in advertising or publicity pertaining to distribution of the
.\" software without specific, written prior permission.  M.I.T. makes
.\" no representations about the suitability of this software for any
.\" purpose.  It is provided "as is" without express or implied
.\" warranty.
.\"
.\" THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''.  M.I.T. DISCLAIMS
.\" ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
.\" INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
.\" SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
.\" SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
.\" USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
.\" ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
.\" OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
.\" OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd January 30, 2023
.Dt SHM_OPEN 2
.Os
.Sh NAME
.Nm memfd_create , shm_create_largepage , shm_open , shm_rename, shm_unlink
.Nd "shared memory object operations"
.Sh LIBRARY
.Lb libc
.Sh SYNOPSIS
.In sys/types.h
.In sys/mman.h
.In fcntl.h
.Ft int
.Fn memfd_create "const char *name" "unsigned int flags"
.Ft int
.Fo shm_create_largepage
.Fa "const char *path"
.Fa "int flags"
.Fa "int psind"
.Fa "int alloc_policy"
.Fa "mode_t mode"
.Fc
.Ft int
.Fn shm_open "const char *path" "int flags" "mode_t mode"
.Ft int
.Fn shm_rename "const char *path_from" "const char *path_to" "int flags"
.Ft int
.Fn shm_unlink "const char *path"
.Sh DESCRIPTION
The
.Fn shm_open
function opens (or optionally creates) a
POSIX
shared memory object named
.Fa path .
The
.Fa flags
argument contains a subset of the flags used by
.Xr open 2 .
An access mode of either
.Dv O_RDONLY
or
.Dv O_RDWR
must be included in
.Fa flags .
The optional flags
.Dv O_CREAT ,
.Dv O_EXCL ,
and
.Dv O_TRUNC
may also be specified.
.Pp
If
.Dv O_CREAT
is specified,
then a new shared memory object named
.Fa path
will be created if it does not exist.
In this case,
the shared memory object is created with mode
.Fa mode
subject to the process' umask value.
If both the
.Dv O_CREAT
and
.Dv O_EXCL
flags are specified and a shared memory object named
.Fa path
already exists,
then
.Fn shm_open
will fail with
.Er EEXIST .
.Pp
Newly created objects start off with a size of zero.
If an existing shared memory object is opened with
.Dv O_RDWR
and the
.Dv O_TRUNC
flag is specified,
then the shared memory object will be truncated to a size of zero.
The size of the object can be adjusted via
.Xr ftruncate 2
and queried via
.Xr fstat 2 .
.Pp
The new descriptor is set to close during
.Xr execve 2
system calls;
see
.Xr close 2
and
.Xr fcntl 2 .
.Pp
The constant
.Dv SHM_ANON
may be used for the
.Fa path
argument to
.Fn shm_open .
In this case, an anonymous, unnamed shared memory object is created.
Since the object has no name,
it cannot be removed via a subsequent call to
.Fn shm_unlink ,
or moved with a call to
.Fn shm_rename .
Instead,
the shared memory object will be garbage collected when the last reference to
the shared memory object is removed.
The shared memory object may be shared with other processes by sharing the
file descriptor via
.Xr fork 2
or
.Xr sendmsg 2 .
Attempting to open an anonymous shared memory object with
.Dv O_RDONLY
will fail with
.Er EINVAL .
All other flags are ignored.
.Pp
The
.Fn shm_create_largepage
function behaves similarly to
.Fn shm_open ,
except that the
.Dv O_CREAT
flag is implicitly specified, and the returned
.Dq largepage
object is always backed by aligned, physically contiguous chunks of memory.
This ensures that the object can be mapped using so-called
.Dq superpages ,
which can improve application performance in some workloads by reducing the
number of translation lookaside buffer (TLB) entries required to access a
mapping of the object,
and by reducing the number of page faults performed when accessing a mapping.
This happens automatically for all largepage objects.
.Pp
An existing largepage object can be opened using the
.Fn shm_open
function.
Largepage shared memory objects behave slightly differently from non-largepage
objects:
.Bl -bullet -offset indent
.It
Memory for a largepage object is allocated when the object is
extended using the
.Xr ftruncate 2
system call, whereas memory for regular shared memory objects is allocated
lazily and may be paged out to a swap device when not in use.
.It
The size of a mapping of a largepage object must be a multiple of the
underlying large page size.
Most attributes of such a mapping can only be modified at the granularity
of the large page size.
For example, when using
.Xr munmap 2
to unmap a portion of a largepage object mapping, or when using
.Xr mprotect 2
to adjust protections of a mapping of a largepage object, the starting address
must be large page size-aligned, and the length of the operation must be a
multiple of the large page size.
If not, the corresponding system call will fail and set
.Va errno
to
.Er EINVAL .
.El
.Pp
The
.Fa psind
argument to
.Fn shm_create_largepage
specifies the size of large pages used to back the object.
This argument is an index into the page sizes array returned by
.Xr getpagesizes 3 .
In particular, all large pages backing a largepage object must be of the
same size.
For example, on a system with large page sizes of 2MB and 1GB, a 2GB largepage
object will consist of either 1024 2MB pages, or 2 1GB pages, depending on
the value specified for the
.Fa psind
argument.
The
.Fa alloc_policy
parameter specifies what happens when an attempt to use
.Xr ftruncate 2
to allocate memory for the object fails.
The following values are accepted:
.Bl -tag -offset indent -width SHM_
.It Dv SHM_LARGEPAGE_ALLOC_DEFAULT
If the (non-blocking) memory allocation fails because there is insufficient free
contiguous memory, the kernel will attempt to defragment physical memory and
try another allocation.
The subsequent allocation may or may not succeed.
If this subsequent allocation also fails,
.Xr ftruncate 2
will fail and set
.Va errno
to
.Er ENOMEM .
.It Dv SHM_LARGEPAGE_ALLOC_NOWAIT
If the memory allocation fails,
.Xr ftruncate 2
will fail and set
.Va errno
to
.Er ENOMEM .
.It Dv SHM_LARGEPAGE_ALLOC_HARD
The kernel will attempt defragmentation until the allocation succeeds,
or an unblocked signal is delivered to the thread.
However, it is possible for physical memory to be fragmented such that the
allocation will never succeed.
.El
.Pp
The
.Dv FIOSSHMLPGCNF
and
.Dv FIOGSHMLPGCNF
.Xr ioctl 2
commands can be used with a largepage shared memory object to get and set
largepage object parameters.
Both commands operate on the following structure:
.Bd -literal
struct shm_largepage_conf {
	int psind;
	int alloc_policy;
};

.Ed
The
.Dv FIOGSHMLPGCNF
command populates this structure with the current values of these parameters,
while the
.Dv FIOSSHMLPGCNF
command modifies the largepage object.
Currently only the
.Va alloc_policy
parameter may be modified.
Internally,
.Fn shm_create_largepage
works by creating a regular shared memory object using
.Fn shm_open ,
and then converting it into a largepage object using the
.Dv FIOSSHMLPGCNF
ioctl command.
.Pp
The
.Fn shm_rename
system call atomically removes a shared memory object named
.Fa path_from
and relinks it at
.Fa path_to .
If another object is already linked at
.Fa path_to ,
that object will be unlinked, unless one of the following flags are provided:
.Bl -tag -offset indent -width Er
.It Er SHM_RENAME_EXCHANGE
Atomically exchange the shms at
.Fa path_from
and
.Fa path_to .
.It Er SHM_RENAME_NOREPLACE
Return an error if an shm exists at
.Fa path_to ,
rather than unlinking it.
.El
.Pp
The
.Fn shm_unlink
system call removes a shared memory object named
.Fa path .
.Pp
The
.Fn memfd_create
function creates an anonymous shared memory object, identical to that created
by
.Fn shm_open
when
.Dv SHM_ANON
is specified.
Newly created objects start off with a size of zero.
The size of the new object must be adjusted via
.Xr ftruncate 2 .
.Pp
The
.Fa name
argument must not be
.Dv NULL ,
but it may be an empty string.
The length of the
.Fa name
argument may not exceed
.Dv NAME_MAX
minus six characters for the prefix
.Dq memfd: ,
which will be prepended.
The
.Fa name
argument is intended solely for debugging purposes and will never be used by the
kernel to identify a memfd.
Names are therefore not required to be unique.
.Pp
The following
.Fa flags
may be specified to
.Fn memfd_create :
.Bl -tag -width MFD_ALLOW_SEALING
.It Dv MFD_CLOEXEC
Set
.Dv FD_CLOEXEC
on the resulting file descriptor.
.It Dv MFD_ALLOW_SEALING
Allow adding seals to the resulting file descriptor using the
.Dv F_ADD_SEALS
.Xr fcntl 2
command.
.It Dv MFD_HUGETLB
This flag is currently unsupported.
.El
.Sh RETURN VALUES
If successful,
.Fn memfd_create
and
.Fn shm_open
both return a non-negative integer,
and
.Fn shm_rename
and
.Fn shm_unlink
return zero.
All functions return -1 on failure, and set
.Va errno
to indicate the error.
.Sh COMPATIBILITY
The
.Fn shm_create_largepage
and
.Fn shm_rename
functions are
.Fx
extensions, as is support for the
.Dv SHM_ANON
value in
.Fn shm_open .
.Pp
The
.Fa path ,
.Fa path_from ,
and
.Fa path_to
arguments do not necessarily represent a pathname (although they do in
most other implementations).
Two processes opening the same
.Fa path
are guaranteed to access the same shared memory object if and only if
.Fa path
begins with a slash
.Pq Ql \&/
character.
.Pp
Only the
.Dv O_RDONLY ,
.Dv O_RDWR ,
.Dv O_CREAT ,
.Dv O_EXCL ,
and
.Dv O_TRUNC
flags may be used in portable programs.
.Pp
.Tn POSIX
specifications state that the result of using
.Xr open 2 ,
.Xr read 2 ,
or
.Xr write 2
on a shared memory object, or on the descriptor returned by
.Fn shm_open ,
is undefined.
However, the
.Fx
kernel implementation explicitly includes support for
.Xr read 2
and
.Xr write 2 .
.Pp
.Fx
also supports zero-copy transmission of data from shared memory
objects with
.Xr sendfile 2 .
.Pp
Neither shared memory objects nor their contents persist across reboots.
.Pp
Writes do not extend shared memory objects, so
.Xr ftruncate 2
must be called before any data can be written.
See
.Sx EXAMPLES .
.Sh EXAMPLES
This example fails without the call to
.Xr ftruncate 2 :
.Bd -literal -compact

        uint8_t buffer[getpagesize()];
        ssize_t len;
        int fd;

        fd = shm_open(SHM_ANON, O_RDWR | O_CREAT, 0600);
        if (fd < 0)
                err(EX_OSERR, "%s: shm_open", __func__);
        if (ftruncate(fd, getpagesize()) < 0)
                err(EX_IOERR, "%s: ftruncate", __func__);
        len = pwrite(fd, buffer, getpagesize(), 0);
        if (len < 0)
                err(EX_IOERR, "%s: pwrite", __func__);
        if (len != getpagesize())
                errx(EX_IOERR, "%s: pwrite length mismatch", __func__);
.Ed
.Sh ERRORS
.Fn memfd_create
fails with these error codes for these conditions:
.Bl -tag -width Er
.It Bq Er EBADF
The
.Fa name
argument was NULL.
.It Bq Er EINVAL
The
.Fa name
argument was too long.
.Pp
An invalid or unsupported flag was included in
.Fa flags .
.It Bq Er EMFILE
The process has already reached its limit for open file descriptors.
.It Bq Er ENFILE
The system file table is full.
.It Bq Er ENOSYS
In
.Fa memfd_create ,
.Dv MFD_HUGETLB
was specified in
.Fa flags ,
and this system does not support forced hugetlb mappings.
.El
.Pp
.Fn shm_open
fails with these error codes for these conditions:
.Bl -tag -width Er
.It Bq Er EINVAL
A flag other than
.Dv O_RDONLY ,
.Dv O_RDWR ,
.Dv O_CREAT ,
.Dv O_EXCL ,
or
.Dv O_TRUNC
was included in
.Fa flags .
.It Bq Er EMFILE
The process has already reached its limit for open file descriptors.
.It Bq Er ENFILE
The system file table is full.
.It Bq Er EINVAL
.Dv O_RDONLY
was specified while creating an anonymous shared memory object via
.Dv SHM_ANON .
.It Bq Er EFAULT
The
.Fa path
argument points outside the process' allocated address space.
.It Bq Er ENAMETOOLONG
The entire pathname exceeds 1023 characters.
.It Bq Er EINVAL
The
.Fa path
does not begin with a slash
.Pq Ql \&/
character.
.It Bq Er ENOENT
.Dv O_CREAT
is not specified and the named shared memory object does not exist.
.It Bq Er EEXIST
.Dv O_CREAT
and
.Dv O_EXCL
are specified and the named shared memory object does exist.
.It Bq Er EACCES
The required permissions (for reading or reading and writing) are denied.
.It Bq Er ECAPMODE
The process is running in capability mode (see
.Xr capsicum 4 )
and attempted to create a named shared memory object.
.El
.Pp
.Fn shm_create_largepage
can fail for the reasons listed above.
It also fails with these error codes for the following conditions:
.Bl -tag -width Er
.It Bq Er ENOTTY
The kernel does not support large pages on the current platform.
.El
.Pp
The following errors are defined for
.Fn shm_rename :
.Bl -tag -width Er
.It Bq Er EFAULT
The
.Fa path_from
or
.Fa path_to
argument points outside the process' allocated address space.
.It Bq Er ENAMETOOLONG
The entire pathname exceeds 1023 characters.
.It Bq Er ENOENT
The shared memory object at
.Fa path_from
does not exist.
.It Bq Er EACCES
The required permissions are denied.
.It Bq Er EEXIST
An shm exists at
.Fa path_to ,
and the
.Dv SHM_RENAME_NOREPLACE
flag was provided.
.El
.Pp
.Fn shm_unlink
fails with these error codes for these conditions:
.Bl -tag -width Er
.It Bq Er EFAULT
The
.Fa path
argument points outside the process' allocated address space.
.It Bq Er ENAMETOOLONG
The entire pathname exceeds 1023 characters.
.It Bq Er ENOENT
The named shared memory object does not exist.
.It Bq Er EACCES
The required permissions are denied.
.Fn shm_unlink
requires write permission to the shared memory object.
.El
.Sh SEE ALSO
.Xr close 2 ,
.Xr fstat 2 ,
.Xr ftruncate 2 ,
.Xr ioctl 2 ,
.Xr mmap 2 ,
.Xr munmap 2 ,
.Xr sendfile 2
.Sh STANDARDS
The
.Fn memfd_create
function is expected to be compatible with the Linux system call of the same
name.
.Pp
The
.Fn shm_open
and
.Fn shm_unlink
functions are believed to conform to
.St -p1003.1b-93 .
.Sh HISTORY
The
.Fn memfd_create
function appeared in
.Fx 13.0 .
.Pp
The
.Fn shm_open
and
.Fn shm_unlink
functions first appeared in
.Fx 4.3 .
The functions were reimplemented as system calls using shared memory objects
directly rather than files in
.Fx 8.0 .
.Pp
.Fn shm_rename
first appeared in
.Fx 13.0
as a
.Fx
extension.
.Sh AUTHORS
.An Garrett A. Wollman Aq Mt wollman@FreeBSD.org
(C library support and this manual page)
.Pp
.An Matthew Dillon Aq Mt dillon@FreeBSD.org
.Pq Dv MAP_NOSYNC
.Pp
.An Matthew Bryan Aq Mt matthew.bryan@isilon.com
.Pq Dv shm_rename implementation