fsync(2) - f33

1FSYNC(2)                   Linux Programmer's Manual                  FSYNC(2)
2
3
4

NAME

6       fsync,  fdatasync  -  synchronize  a  file's in-core state with storage
7       device
8

SYNOPSIS

10       #include <unistd.h>
11
12       int fsync(int fd);
13
14       int fdatasync(int fd);
15
16   Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
17
18       fsync():
19           Glibc 2.16 and later:
20               No feature test macros need be defined
21           Glibc up to and including 2.15:
22               _BSD_SOURCE || _XOPEN_SOURCE
23                   || /* since glibc 2.8: */ _POSIX_C_SOURCE >= 200112L
24       fdatasync():
25           _POSIX_C_SOURCE >= 199309L || _XOPEN_SOURCE >= 500
26

DESCRIPTION

28       fsync() transfers ("flushes") all modified in-core data of (i.e., modi‐
29       fied  buffer cache pages for) the file referred to by the file descrip‐
30       tor fd to the disk device (or other permanent storage device)  so  that
31       all  changed information can be retrieved even if the system crashes or
32       is rebooted.  This includes writing through or flushing a disk cache if
33       present.   The  call  blocks until the device reports that the transfer
34       has completed.
35
36       As well as flushing the file data, fsync() also  flushes  the  metadata
37       information associated with the file (see inode(7)).
38
39       Calling  fsync()  does  not  necessarily  ensure  that the entry in the
40       directory containing the file has  also  reached  disk.   For  that  an
41       explicit fsync() on a file descriptor for the directory is also needed.
42
43       fdatasync() is similar to fsync(), but does not flush modified metadata
44       unless that metadata is needed in order  to  allow  a  subsequent  data
45       retrieval to be correctly handled.  For example, changes to st_atime or
46       st_mtime (respectively, time of last access and time of last  modifica‐
47       tion; see inode(7)) do not require flushing because they are not neces‐
48       sary for a subsequent data read to be handled correctly.  On the  other
49       hand, a change to the file size (st_size, as made by say ftruncate(2)),
50       would require a metadata flush.
51
52       The aim of fdatasync() is to reduce disk activity for applications that
53       do not require all metadata to be synchronized with the disk.
54

RETURN VALUE

56       On  success, these system calls return zero.  On error, -1 is returned,
57       and errno is set appropriately.
58

ERRORS

60       EBADF  fd is not a valid open file descriptor.
61
62       EIO    An error occurred during synchronization.  This error may relate
63              to  data written to some other file descriptor on the same file.
64              Since Linux 4.13, errors from write-back will be reported to all
65              file  descriptors  that  might have written the data which trig‐
66              gered the error.  Some filesystems (e.g., NFS) keep close  track
67              of  which data came through which file descriptor, and give more
68              precise reporting.  Other filesystems (e.g., most local filesys‐
69              tems)  will report errors to all file descriptors that were open
70              on the file when the error was recorded.
71
72       ENOSPC Disk space was exhausted while synchronizing.
73
74       EROFS, EINVAL
75              fd is bound to a special file (e.g., a pipe,  FIFO,  or  socket)
76              which does not support synchronization.
77
78       ENOSPC, EDQUOT
79              fd  is  bound  to a file on NFS or another filesystem which does
80              not allocate space at the time of a write(2)  system  call,  and
81              some previous write failed due to insufficient storage space.
82

CONFORMING TO

84       POSIX.1-2001, POSIX.1-2008, 4.3BSD.
85
86       On  POSIX  systems  on  which fdatasync() is available, _POSIX_SYNCHRO‐
87       NIZED_IO is defined in <unistd.h> to a value greater than 0.  (See also
88       sysconf(3).)
89

NOTES

91       On  some  UNIX  systems  (but  not  Linux),  fd must be a writable file
92       descriptor.
93
94       In Linux 2.2 and earlier, fdatasync() is equivalent to fsync(), and  so
95       has no performance advantage.
96
97       The  fsync()  implementations in older kernels and lesser used filesys‐
98       tems do not know how to flush disk caches.  In these cases disk  caches
99       need  to  be  disabled  using  hdparm(8) or sdparm(8) to guarantee safe
100       operation.
101

COLOPHON

107       This  page  is  part of release 5.07 of the Linux man-pages project.  A
108       description of the project, information about reporting bugs,  and  the
109       latest     version     of     this    page,    can    be    found    at
110       https://www.kernel.org/doc/man-pages/.
111
112
113
114Linux                             2020-06-09                          FSYNC(2)