fsync(2)

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

DESCRIPTION

17       fsync() transfers ("flushes") all modified in-core data of (i.e., modi‐
18       fied  buffer cache pages for) the file referred to by the file descrip‐
19       tor fd to the disk device (or other  permanent  storage  device)  where
20       that  file  resides.  The call blocks until the device reports that the
21       transfer has completed.  It also flushes  metadata information  associ‐
22       ated with the file (see stat(2)).
23
24       Calling  fsync()  does  not  necessarily  ensure  that the entry in the
25       directory containing the file has  also  reached  disk.   For  that  an
26       explicit fsync() on a file descriptor for the directory is also needed.
27
28       fdatasync() is similar to fsync(), but does not flush modified metadata
29       unless that metadata is needed in order  to  allow  a  subsequent  data
30       retrieval to be correctly handled.  For example, changes to st_atime or
31       st_mtime (respectively, time of last access and time of last  modifica‐
32       tion;  see stat(2)) do not require flushing because they are not neces‐
33       sary for a subsequent data read to be handled correctly.  On the  other
34       hand, a change to the file size (st_size, as made by say ftruncate(2)),
35       would require a metadata flush.
36
37       The aim of fdatasync(2) is to reduce  disk  activity  for  applications
38       that do not require all metadata to be synchronised with the disk.
39

RETURN VALUE

41       On  success,  zero is returned.  On error, -1 is returned, and errno is
42       set appropriately.
43

ERRORS

45       EBADF  fd is not a valid file descriptor open for writing.
46
47       EIO    An error occurred during synchronization.
48
49       EROFS, EINVAL
50              fd is bound to a special file which does  not  support  synchro‐
51              nization.
52

NOTES

54       If  the  underlying  hard disk has write caching enabled, then the data
55       may not really be on  permanent  storage  when  fsync()  /  fdatasync()
56       return.
57
58       When  an  ext2  file  system is mounted with the sync option, directory
59       entries are also implicitly synced by fsync().
60
61       On kernels before 2.4, fsync() on big files  can  be  inefficient.   An
62       alternative might be to use the O_SYNC flag to open(2).
63

CONFORMING TO

65       4.3BSD, POSIX.1-2001
66