Open and possibly create a file
#include <sys/types.h> #include <sys/stat.h> #include <fcntl.h> int open(const char *pathname, int flags); int open(const char *pathname, int flags, mode_t mode); int creat(const char *pathname, mode_t mode); int openat(int dirfd, const char *pathname, int flags); int openat(int dirfd, const char *pathname, int flags, mode_t mode);
Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
openat():
Since glibc 2.10:
_XOPEN_SOURCE >= 700 || _POSIX_C_SOURCE >= 200809L
Before glibc 2.10:
_ATFILE_SOURCE
Given a pathname for a file, open() returns a file descriptor, a small, nonnegative integer for use in subsequent system calls (read(2), write(2), lseek(2), fcntl(2), etc.). The file descriptor returned by a successful call will be the lowest-numbered file descriptor not currently open for the process.
By default, the new file descriptor is set to remain open across an execve(2) (i.e., the FD_CLOEXEC file descriptor flag described in fcntl(2) is initially disabled); the O_CLOEXEC flag, described below, can be used to change this default. The file offset is set to the beginning of the file (see lseek(2)).
A call to open() creates a new open file description, an entry in the system-wide table of open files. The open file description records the file offset and the file status flags (see below). A file descriptor is a reference to an open file description; this reference is unaffected if pathname is subsequently removed or modified to refer to a different file. For further details on open file descriptions, see NOTES.
The argument flags must include one of the following access modes: O_RDONLY, O_WRONLY, or O_RDWR. These request opening the file read-only, write-only, or read/write, respectively.
In addition, zero or more file creation flags and file status flags can be bitwise-or'd in flags. The file creation flags are O_CLOEXEC, O_CREAT, O_DIRECTORY, O_EXCL, O_NOCTTY, O_NOFOLLOW, O_TMPFILE, O_TRUNC, and O_TTY_INIT. The file status flags are all of the remaining flags listed below. The distinction between these two groups of flags is that the file status flags can be retrieved and (in some cases) modified; see fcntl(2) for details.
The full list of file creation flags and file status flags is as follows:
O_APPEND
The file is opened in append mode. Before each write(2), the file offset is positioned at the end of the file, as if with lseek(2). O_APPEND may lead to corrupted files on NFS filesystems if more than one process appends data to a file at once. This is because NFS does not support appending to a file, so the client kernel has to simulate it, which can't be done without a race condition.
O_ASYNC
Enable signal-driven I/O: generate a signal (SIGIO by default, but this can be changed via fcntl(2)) when input or output becomes possible on this file descriptor. This feature is available only for terminals, pseudoterminals, sockets, and (since Linux 2.6) pipes and FIFOs. See fcntl(2) for further details. See also BUGS, below.
O_CLOEXEC (since Linux 2.6.23)
Enable the close-on-exec flag for the new file descriptor. Specifying this flag permits a program to avoid additional fcntl(2) F_SETFD operations to set the FD_CLOEXEC flag.
Note that the use of this flag is essential in some multithreaded programs, because using a separate fcntl(2) F_SETFD operation to set the FD_CLOEXEC flag does not suffice to avoid race conditions where one thread opens a file descriptor and attempts to set its close-on-exec flag using fcntl(2) at the same time as another thread does a fork(2) plus execve(2). Depending on the order of execution, the race may lead to the file descriptor returned by open() being unintentionally leaked to the program executed by the child process created by fork(2). (This kind of race is in principle possible for any system call that creates a file descriptor whose close-on-exec flag should be set, and various other Linux system calls provide an equivalent of the O_CLOEXEC flag to deal with this problem.)
O_CREAT
If the file does not exist, it will be created. The owner (user ID) of the file is set to the effective user ID of the process. The group ownership (group ID) is set either to the effective group ID of the process or to the group ID of the parent directory (depending on filesystem type and mount options, and the mode of the parent directory; see the mount options bsdgroups and sysvgroups described in mount(8)).
mode specifies the permissions to use in case a new file is created. This argument must be supplied when O_CREAT or O_TMPFILE is specified in flags; if neither O_CREAT nor O_TMPFILE is specified, then mode is ignored. The effective permissions are modified by the process's umask in the usual way: The permissions of the created file are (mode & ~umask). Note that this mode applies only to future accesses of the newly created file; the open() call that creates a read-only file may well return a read/write file descriptor.
The following symbolic constants are provided for mode:
S_IRWXU
00700 user (file owner) has read, write and execute permission
S_IRUSR
00400 user has read permission
S_IWUSR
00200 user has write permission
S_IXUSR
00100 user has execute permission
S_IRWXG
00070 group has read, write and execute permission
S_IRGRP
00040 group has read permission
S_IWGRP
00020 group has write permission
S_IXGRP
00010 group has execute permission
S_IRWXO
00007 others have read, write and execute permission
S_IROTH
00004 others have read permission
S_IWOTH
00002 others have write permission
S_IXOTH
00001 others have execute permission
O_DIRECT (since Linux 2.4.10)
Try to minimize cache effects of the I/O to and from this file. In general this will degrade performance, but it is useful in special situations, such as when applications do their own caching. File I/O is done directly to/from user-space buffers. The O_DIRECT flag on its own makes an effort to transfer data synchronously, but does not give the guarantees of the O_SYNC flag that data and necessary metadata are transferred. To guarantee synchronous I/O, O_SYNC must be used in addition to O_DIRECT. See NOTES below for further discussion.
A semantically similar (but deprecated) interface for block devices is described in raw(8).
O_DIRECTORY
If pathname is not a directory, cause the open to fail. This flag was added in kernel version 2.1.126, to avoid denial-of-service problems if opendir(3) is called on a FIFO or tape device.
O_DSYNC
Write operations on the file will complete according to the requirements of synchronized I/O data integrity completion.
By the time write(2) (and similar) return, the output data has been transferred to the underlying hardware, along with any file metadata that would be required to retrieve that data (i.e., as though each write(2) was followed by a call to fdatasync(2)). See NOTES below.
O_EXCL
Ensure that this call creates the file: if this flag is specified in conjunction with O_CREAT, and pathname already exists, then open() will fail.
When these two flags are specified, symbolic links are not followed: if pathname is a symbolic link, then open() fails regardless of where the symbolic link points to.
In general, the behavior of O_EXCL is undefined if it is used without O_CREAT. There is one exception: on Linux 2.6 and later, O_EXCL can be used without O_CREAT if pathname refers to a block device. If the block device is in use by the system (e.g., mounted), open() fails with the error EBUSY.
On NFS, O_EXCL is supported only when using NFSv3 or later on kernel 2.6 or later. In NFS environments where O_EXCL support is not provided, programs that rely on it for performing locking tasks will contain a race condition. Portable programs that want to perform atomic file locking using a lockfile, and need to avoid reliance on NFS support for O_EXCL, can create a unique file on the same filesystem (e.g., incorporating hostname and PID), and use link(2) to make a link to the lockfile. If link(2) returns 0, the lock is successful. Otherwise, use stat(2) on the unique file to check if its link count has increased to 2, in which case the lock is also successful.
O_LARGEFILE
(LFS) Allow files whose sizes cannot be represented in an off_t (but can be represented in an off64_t) to be opened. The _LARGEFILE64_SOURCE macro must be defined (before including any header files) in order to obtain this definition. Setting the _FILE_OFFSET_BITS feature test macro to 64 (rather than using O_LARGEFILE) is the preferred method of accessing large files on 32-bit systems (see feature_test_macros(7)).
O_NOATIME (since Linux 2.6.8)
Do not update the file last access time (st_atime in the inode) when the file is read(2). This flag is intended for use by indexing or backup programs, where its use can significantly reduce the amount of disk activity. This flag may not be effective on all filesystems. One example is NFS, where the server maintains the access time.
O_NOCTTY
If pathname refers to a terminal device–see tty(4)–it will not become the process's controlling terminal even if the process does not have one.
O_NOFOLLOW
If pathname is a symbolic link, then the open fails. This is a FreeBSD extension, which was added to Linux in version 2.1.126. Symbolic links in earlier components of the pathname will still be followed. See also O_PATH below.
O_NONBLOCK or O_NDELAY
When possible, the file is opened in nonblocking mode. Neither the open() nor any subsequent operations on the file descriptor which is returned will cause the calling process to wait. For the handling of FIFOs (named pipes), see also fifo(7). For a discussion of the effect of O_NONBLOCK in conjunction with mandatory file locks and with file leases, see fcntl(2).
O_PATH (since Linux 2.6.39)
Obtain a file descriptor that can be used for two purposes: to indicate a location in the filesystem tree and to perform operations that act purely at the file descriptor level. The file itself is not opened, and other file operations (e.g., read(2), write(2), fchmod(2), fchown(2), fgetxattr(2), mmap(2)) fail with the error EBADF.
The following operations can be performed on the resulting file descriptor:
close(2); fchdir(2) (since Linux 3.5); fstat(2) (since Linux 3.6).
Duplicating the file descriptor (dup(2), fcntl(2) F_DUPFD, etc.).
Getting and setting file descriptor flags (fcntl(2) F_GETFD and F_SETFD).
Retrieving open file status flags using the fcntl(2) F_GETFL operation: the returned flags will include the bit O_PATH.
Passing the file descriptor as the dirfd argument of openat(2) and the other "*at()" system calls. This includes linkat(2) with AT_EMPTY_PATH (or via procfs using AT_SYMLINK_FOLLOW) even if the file is not a directory.
Passing the file descriptor to another process via a UNIX domain socket (see SCM_RIGHTS in unix(7)).
When O_PATH is specified in flags, flag bits other than O_CLOEXEC, O_DIRECTORY, and O_NOFOLLOW are ignored.
If pathname is a symbolic link and the O_NOFOLLOW flag is also specified, then the call returns a file descriptor referring to the symbolic link. This file descriptor can be used as the dirfd argument in calls to fchownat(2), fstatat(2), linkat(2), and readlinkat(2) with an empty pathname to have the calls operate on the symbolic link.
O_SYNC
Write operations on the file will complete according to the requirements of synchronized I/O file integrity completion (by contrast with the synchronized I/O data integrity completion provided by O_DSYNC.)
By the time write(2) (and similar) return, the output data and associated file metadata have been transferred to the underlying hardware (i.e., as though each write(2) was followed by a call to fsync(2)). See NOTES below.
O_TMPFILE (since Linux 3.11)
Create an unnamed temporary file. The pathname argument specifies a directory; an unnamed inode will be created in that directory's filesystem. Anything written to the resulting file will be lost when the last file descriptor is closed, unless the file is given a name.
O_TMPFILE must be specified with one of O_RDWR or O_WRONLY and, optionally, O_EXCL. If O_EXCL is not specified, then linkat(2) can be used to link the temporary file into the filesystem, making it permanent, using code like the following:
char path[PATH_MAX];
fd = open("/path/to/dir", O_TMPFILE | O_RDWR,
S_IRUSR | S_IWUSR);
/* File I/O on 'fd'... */
snprintf(path, PATH_MAX, "/proc/self/fd/%d", fd);
linkat(AT_FDCWD, path, AT_FDCWD, "/path/for/file",
AT_SYMLINK_FOLLOW);
In this case, the open() mode argument determines the file permission mode, as with O_CREAT.
Specifying O_EXCL in conjunction with O_TMPFILE prevents a temporary file from being linked into the filesystem in the above manner. (Note that the meaning of O_EXCL in this case is different from the meaning of O_EXCL otherwise.)
There are two main use cases for O_TMPFILE:
Improved tmpfile(3) functionality: race-free creation of temporary files that (1) are automatically deleted when closed; (2) can never be reached via any pathname; (3) are not subject to symlink attacks; and (4) do not require the caller to devise unique names.
Creating a file that is initially invisible, which is then populated with data and adjusted to have appropriate filesystem attributes (chown(2), chmod(2), fsetxattr(2), etc.) before being atomically linked into the filesystem in a fully formed state (using linkat(2) as described above).
O_TMPFILE requires support by the underlying filesystem; only a subset of Linux filesystems provide that support. In the initial implementation, support was provided in the ext2, ext3, ext4, UDF, Minix, and shmem filesystems. XFS support was added in Linux 3.15.
O_TRUNC
If the file already exists and is a regular file and the access mode allows writing (i.e., is O_RDWR or O_WRONLY) it will be truncated to length 0. If the file is a FIFO or terminal device file, the O_TRUNC flag is ignored. Otherwise, the effect of O_TRUNC is unspecified.
creat() is equivalent to open() with flags equal to O_CREAT|O_WRONLY|O_TRUNC.
The openat() system call operates in exactly the same way as open(), except for the differences described here.
If the pathname given in pathname is relative, then it is interpreted relative to the directory referred to by the file descriptor dirfd (rather than relative to the current working directory of the calling process, as is done by open() for a relative pathname).
If pathname is relative and dirfd is the special value AT_FDCWD, then pathname is interpreted relative to the current working directory of the calling process (like open()).
If pathname is absolute, then dirfd is ignored.
open(), openat(), and creat() return the new file descriptor, or -1 if an error occurred (in which case, errno is set appropriately).
open(), openat(), and creat() can fail with the following errors:
EACCES
The requested access to the file is not allowed, or search permission is denied for one of the directories in the path prefix of pathname, or the file did not exist yet and write access to the parent directory is not allowed. (See also path_resolution(7).)
EDQUOT
Where O_CREAT is specified, the file does not exist, and the user's quota of disk blocks or inodes on the filesystem has been exhausted.
EEXIST
pathname already exists and O_CREAT and O_EXCL were used.
EFAULT
pathname points outside your accessible address space.
EFBIG
See EOVERFLOW.
EINTR
While blocked waiting to complete an open of a slow device (e.g., a FIFO; see fifo(7)), the call was interrupted by a signal handler; see signal(7).
EINVAL
The filesystem does not support the O_DIRECT flag. See NOTES for more information.
EINVAL
Invalid value in flags.
EINVAL
O_TMPFILE was specified in flags, but neither O_WRONLY nor O_RDWR was specified.
EISDIR
pathname refers to a directory and the access requested involved writing (that is, O_WRONLY or O_RDWR is set).
EISDIR
pathname refers to an existing directory, O_TMPFILE and one of O_WRONLY or O_RDWR were specified in flags, but this kernel version does not provide the O_TMPFILE functionality.
ELOOP
Too many symbolic links were encountered in resolving pathname.
ELOOP
pathname was a symbolic link, and flags specified O_NOFOLLOW but not O_PATH.
EMFILE
The process already has the maximum number of files open.
ENAMETOOLONG
pathname was too long.
ENFILE
The system limit on the total number of open files has been reached.
ENODEV
pathname refers to a device special file and no corresponding device exists. (This is a Linux kernel bug; in this situation ENXIO must be returned.)
ENOENT
O_CREAT is not set and the named file does not exist. Or, a directory component in pathname does not exist or is a dangling symbolic link.
ENOENT
pathname refers to a nonexistent directory, O_TMPFILE and one of O_WRONLY or O_RDWR were specified in flags, but this kernel version does not provide the O_TMPFILE functionality.
ENOMEM
Insufficient kernel memory was available.
ENOSPC
pathname was to be created but the device containing pathname has no room for the new file.
ENOTDIR
A component used as a directory in pathname is not, in fact, a directory, or O_DIRECTORY was specified and pathname was not a directory.
ENXIO
O_NONBLOCK | O_WRONLY is set, the named file is a FIFO, and no process has the FIFO open for reading. Or, the file is a device special file and no corresponding device exists.
EOPNOTSUPP
The filesystem containing pathname does not support O_TMPFILE.
EOVERFLOW
pathname refers to a regular file that is too large to be opened. The usual scenario here is that an application compiled on a 32-bit platform without -D_FILE_OFFSET_BITS=64 tried to open a file whose size exceeds (1<<31)-1 bytes; see also O_LARGEFILE above. This is the error specified by POSIX.1-2001; in kernels before 2.6.24, Linux gave the error EFBIG for this case.
EPERM
The O_NOATIME flag was specified, but the effective user ID of the caller did not match the owner of the file and the caller was not privileged (CAP_FOWNER).
EROFS
pathname refers to a file on a read-only filesystem and write access was requested.
ETXTBSY
pathname refers to an executable image which is currently being executed and write access was requested.
EWOULDBLOCK
The O_NONBLOCK flag was specified, and an incompatible lease was held on the file (see fcntl(2)).
The following additional errors can occur for openat():
EBADF
dirfd is not a valid file descriptor.
ENOTDIR
pathname is a relative pathname and dirfd is a file descriptor referring to a file other than a directory.
openat() was added to Linux in kernel 2.6.16; library support was added to glibc in version 2.4.
open(), creat() SVr4, 4.3BSD, POSIX.1-2001, POSIX.1-2008.
openat(): POSIX.1-2008.
The O_DIRECT, O_NOATIME, O_PATH, and O_TMPFILE flags are Linux-specific. One must define _GNU_SOURCE to obtain their definitions.
The O_CLOEXEC, O_DIRECTORY, and O_NOFOLLOW flags are not specified in POSIX.1-2001, but are specified in POSIX.1-2008. Since glibc 2.12, one can obtain their definitions by defining either _POSIX_C_SOURCE with a value greater than or equal to 200809L or _XOPEN_SOURCE with a value greater than or equal to 700. In glibc 2.11 and earlier, one obtains the definitions by defining _GNU_SOURCE.
As noted in feature_test_macros(7), feature test macros such as _POSIX_C_SOURCE, _XOPEN_SOURCE, and _GNU_SOURCE must be defined before including any header files.
Under Linux, the O_NONBLOCK flag indicates that one wants to open but does not necessarily have the intention to read or write. This is typically used to open devices in order to get a file descriptor for use with ioctl(2).
The (undefined) effect of O_RDONLY | O_TRUNC varies among implementations. On many systems the file is actually truncated.
Note that open() can open device special files, but creat() cannot create them; use mknod(2) instead.
If the file is newly created, its st_atime, st_ctime, st_mtime fields (respectively, time of last access, time of last status change, and time of last modification; see stat(2)) are set to the current time, and so are the st_ctime and st_mtime fields of the parent directory. Otherwise, if the file is modified because of the O_TRUNC flag, its st_ctime and st_mtime fields are set to the current time.
The term open file description is the one used by POSIX to refer to the entries in the system-wide table of open files. In other contexts, this object is variously also called an "open file object", a "file handle", an "open file table entry", or–in kernel-developer parlance–a struct file.
When a file descriptor is duplicated (using dup(2) or similar), the duplicate refers to the same open file description as the original file descriptor, and the two file descriptors consequently share the file offset and file status flags. Such sharing can also occur between processes: a child process created via fork(2) inherits duplicates of its parent's file descriptors, and those duplicates refer to the same open file descriptions.
Each open(2) of a file creates a new open file description; thus, there may be multiple open file descriptions corresponding to a file inode.
The POSIX.1-2008 "synchronized I/O" option specifies different variants of synchronized I/O, and specifies the open() flags O_SYNC, O_DSYNC, and O_RSYNC for controlling the behavior. Regardless of whether an implementation supports this option, it must at least support the use of O_SYNC for regular files.
Linux implements O_SYNC and O_DSYNC, but not O_RSYNC. (Somewhat incorrectly, glibc defines O_RSYNC to have the same value as O_SYNC.)
O_SYNC provides synchronized I/O file integrity completion, meaning write operations will flush data and all associated metadata to the underlying hardware. O_DSYNC provides synchronized I/O data integrity completion, meaning write operations will flush data to the underlying hardware, but will only flush metadata updates that are required to allow a subsequent read operation to complete successfully. Data integrity completion can reduce the number of disk operations that are required for applications that don't need the guarantees of file integrity completion.
To understand the difference between the two types of completion, consider two pieces of file metadata: the file last modification timestamp (st_mtime) and the file length. All write operations will update the last file modification timestamp, but only writes that add data to the end of the file will change the file length. The last modification timestamp is not needed to ensure that a read completes successfully, but the file length is. Thus, O_DSYNC would only guarantee to flush updates to the file length metadata (whereas O_SYNC would also always flush the last modification timestamp metadata).
Before Linux 2.6.33, Linux implemented only the O_SYNC flag for open(). However, when that flag was specified, most filesystems actually provided the equivalent of synchronized I/O data integrity completion (i.e., O_SYNC was actually implemented as the equivalent of O_DSYNC).
Since Linux 2.6.33, proper O_SYNC support is provided. However, to ensure backward binary compatibility, O_DSYNC was defined with the same value as the historical O_SYNC, and O_SYNC was defined as a new (two-bit) flag value that includes the O_DSYNC flag value. This ensures that applications compiled against new headers get at least O_DSYNC semantics on pre-2.6.33 kernels.
There are many infelicities in the protocol underlying NFS, affecting amongst others O_SYNC and O_NDELAY.
On NFS filesystems with UID mapping enabled, open() may return a file descriptor but, for example, read(2) requests are denied with EACCES. This is because the client performs open() by checking the permissions, but UID mapping is performed by the server upon read and write requests.
Unlike the other values that can be specified in flags, the access mode values O_RDONLY, O_WRONLY, and O_RDWR do not specify individual bits. Rather, they define the low order two bits of flags, and are defined respectively as 0, 1, and 2. In other words, the combination O_RDONLY | O_WRONLY is a logical error, and certainly does not have the same meaning as O_RDWR.
Linux reserves the special, nonstandard access mode 3 (binary 11) in flags to mean: check for read and write permission on the file and return a descriptor that can't be used for reading or writing. This nonstandard access mode is used by some Linux drivers to return a descriptor that is to be used only for device-specific ioctl(2) operations.
openat() and the other system calls and library functions that take a directory file descriptor argument (i.e., faccessat(2), fanotify_mark(2), fchmodat(2), fchownat(2), fstatat(2), futimesat(2), linkat(2), mkdirat(2), mknodat(2), name_to_handle_at(2), readlinkat(2), renameat(2), symlinkat(2), unlinkat(2), utimensat(2) mkfifoat(3), and scandirat(3)) are supported for two reasons. Here, the explanation is in terms of the openat() call, but the rationale is analogous for the other interfaces.
First, openat() allows an application to avoid race conditions that could occur when using open() to open files in directories other than the current working directory. These race conditions result from the fact that some component of the directory prefix given to open() could be changed in parallel with the call to open(). Such races can be avoided by opening a file descriptor for the target directory, and then specifying that file descriptor as the dirfd argument of openat().
Second, openat() allows the implementation of a per-thread "current working directory", via file descriptor(s) maintained by the application. (This functionality can also be obtained by tricks based on the use of /proc/self/fd/dirfd, but less efficiently.)
The O_DIRECT flag may impose alignment restrictions on the length and address of user-space buffers and the file offset of I/Os. In Linux alignment restrictions vary by filesystem and kernel version and might be absent entirely. However there is currently no filesystem-independent interface for an application to discover these restrictions for a given file or filesystem. Some filesystems provide their own interfaces for doing so, for example the XFS_IOC_DIOINFO operation in xfsctl(3).
Under Linux 2.4, transfer sizes, and the alignment of the user buffer and the file offset must all be multiples of the logical block size of the filesystem. Since Linux 2.6.0, alignment to the logical block size of the underlying storage (typically 512 bytes) suffices. The logical block size can be determined using the ioctl(2) BLKSSZGET operation or from the shell using the command:
blockdev --getss
O_DIRECT I/Os should never be run concurrently with the fork(2) system call, if the memory buffer is a private mapping (i.e., any mapping created with the mmap(2) MAP_PRIVATE flag; this includes memory allocated on the heap and statically allocated buffers). Any such I/Os, whether submitted via an asynchronous I/O interface or from another thread in the process, should be completed before fork(2) is called. Failure to do so can result in data corruption and undefined behavior in parent and child processes. This restriction does not apply when the memory buffer for the O_DIRECT I/Os was created using shmat(2) or mmap(2) with the MAP_SHARED flag. Nor does this restriction apply when the memory buffer has been advised as MADV_DONTFORK with madvise(2), ensuring that it will not be available to the child after fork(2).
The O_DIRECT flag was introduced in SGI IRIX, where it has alignment restrictions similar to those of Linux 2.4. IRIX has also a fcntl(2) call to query appropriate alignments, and sizes. FreeBSD 4.x introduced a flag of the same name, but without alignment restrictions.
O_DIRECT support was added under Linux in kernel version 2.4.10. Older Linux kernels simply ignore this flag. Some filesystems may not implement the flag and open() will fail with EINVAL if it is used.
Applications should avoid mixing O_DIRECT and normal I/O to the same file, and especially to overlapping byte regions in the same file. Even when the filesystem correctly handles the coherency issues in this situation, overall I/O throughput is likely to be slower than using either mode alone. Likewise, applications should avoid mixing mmap(2) of files with direct I/O to the same files.
The behavior of O_DIRECT with NFS will differ from local filesystems. Older kernels, or kernels configured in certain ways, may not support this combination. The NFS protocol does not support passing the flag to the server, so O_DIRECT I/O will bypass the page cache only on the client; the server may still cache the I/O. The client asks the server to make the I/O synchronous to preserve the synchronous semantics of O_DIRECT. Some servers will perform poorly under these circumstances, especially if the I/O size is small. Some servers may also be configured to lie to clients about the I/O having reached stable storage; this will avoid the performance penalty at some risk to data integrity in the event of server power failure. The Linux NFS client places no alignment restrictions on O_DIRECT I/O.
In summary, O_DIRECT is a potentially powerful tool that should be used with caution. It is recommended that applications treat use of O_DIRECT as a performance option which is disabled by default.
"The thing that has always disturbed me about O_DIRECT is that the whole interface is just stupid, and was probably designed by a deranged monkey on some serious mind-controlling substances."–Linus
Currently, it is not possible to enable signal-driven I/O by specifying O_ASYNC when calling open(); use fcntl(2) to enable this flag.
One must check for two different error codes, EISDIR and ENOENT, when trying to determine whether the kernel supports O_TMPFILE functionality.
This page is part of release 3.74 of the Linux man-pages project. A description of the project, information about reporting bugs, and the latest version of this page, can be found at http://www.kernel.org/doc/man-pages/.