1sync_file_range(2) System Calls Manual sync_file_range(2)
2
6 sync_file_range - sync a file segment with disk
7
9 Standard C library (libc, -lc)
10
12 #define _GNU_SOURCE /* See feature_test_macros(7) */
13 #define _FILE_OFFSET_BITS 64
14 #include <fcntl.h>
15 int sync_file_range(int fd, off_t offset, off_t nbytes,
17 unsigned int flags);
18
20 sync_file_range() permits fine control when synchronizing the open file
21 referred to by the file descriptor fd with disk.
22 offset is the starting byte of the file range to be synchronized.
24 nbytes specifies the length of the range to be synchronized, in bytes;
25 if nbytes is zero, then all bytes from offset through to the end of
26 file are synchronized. Synchronization is in units of the system page
27 size: offset is rounded down to a page boundary; (offset+nbytes-1) is
28 rounded up to a page boundary.
29 The flags bit-mask argument can include any of the following values:
31 SYNC_FILE_RANGE_WAIT_BEFORE
33 Wait upon write-out of all pages in the specified range that
34 have already been submitted to the device driver for write-out
35 before performing any write.
36 SYNC_FILE_RANGE_WRITE
38 Initiate write-out of all dirty pages in the specified range
39 which are not presently submitted write-out. Note that even
40 this may block if you attempt to write more than request queue
41 size.
42 SYNC_FILE_RANGE_WAIT_AFTER
44 Wait upon write-out of all pages in the range after performing
45 any write.
46 Specifying flags as 0 is permitted, as a no-op.
48 Warning
50 This system call is extremely dangerous and should not be used in por‐
51 table programs. None of these operations writes out the file's meta‐
52 data. Therefore, unless the application is strictly performing over‐
53 writes of already-instantiated disk blocks, there are no guarantees
54 that the data will be available after a crash. There is no user inter‐
55 face to know if a write is purely an overwrite. On filesystems using
56 copy-on-write semantics (e.g., btrfs) an overwrite of existing allo‐
57 cated blocks is impossible. When writing into preallocated space, many
58 filesystems also require calls into the block allocator, which this
59 system call does not sync out to disk. This system call does not flush
60 disk write caches and thus does not provide any data integrity on sys‐
61 tems with volatile disk write caches.
62 Some details
64 SYNC_FILE_RANGE_WAIT_BEFORE and SYNC_FILE_RANGE_WAIT_AFTER will detect
65 any I/O errors or ENOSPC conditions and will return these to the
66 caller.
67 Useful combinations of the flags bits are:
69 SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE
71 Ensures that all pages in the specified range which were dirty
72 when sync_file_range() was called are placed under write-out.
73 This is a start-write-for-data-integrity operation.
74 SYNC_FILE_RANGE_WRITE
76 Start write-out of all dirty pages in the specified range which
77 are not presently under write-out. This is an asynchronous
78 flush-to-disk operation. This is not suitable for data integ‐
79 rity operations.
80 SYNC_FILE_RANGE_WAIT_BEFORE (or SYNC_FILE_RANGE_WAIT_AFTER)
82 Wait for completion of write-out of all pages in the specified
83 range. This can be used after an earlier
84 SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE operation to
85 wait for completion of that operation, and obtain its result.
86 SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE |
88 SYNC_FILE_RANGE_WAIT_AFTER
89 This is a write-for-data-integrity operation that will ensure
90 that all pages in the specified range which were dirty when
91 sync_file_range() was called are committed to disk.
92
94 On success, sync_file_range() returns 0; on failure -1 is returned and
95 errno is set to indicate the error.
96
98 EBADF fd is not a valid file descriptor.
99 EINVAL flags specifies an invalid bit; or offset or nbytes is invalid.
101 EIO I/O error.
103 ENOMEM Out of memory.
105 ENOSPC Out of disk space.
107 ESPIPE fd refers to something other than a regular file, a block de‐
109 vice, or a directory.
110
112 sync_file_range2()
113 Some architectures (e.g., PowerPC, ARM) need 64-bit arguments to be
114 aligned in a suitable pair of registers. On such architectures, the
115 call signature of sync_file_range() shown in the SYNOPSIS would force a
116 register to be wasted as padding between the fd and offset arguments.
117 (See syscall(2) for details.) Therefore, these architectures define a
118 different system call that orders the arguments suitably:
119 int sync_file_range2(int fd, unsigned int flags,
121 off_t offset, off_t nbytes);
122 The behavior of this system call is otherwise exactly the same as
124 sync_file_range().
125
127 Linux.
128
130 Linux 2.6.17.
131 sync_file_range2()
133 A system call with this signature first appeared on the ARM architec‐
134 ture in Linux 2.6.20, with the name arm_sync_file_range(). It was re‐
135 named in Linux 2.6.22, when the analogous system call was added for
136 PowerPC. On architectures where glibc support is provided, glibc
137 transparently wraps sync_file_range2() under the name
138 sync_file_range().
139
141 _FILE_OFFSET_BITS should be defined to be 64 in code that takes the ad‐
142 dress of sync_file_range, if the code is intended to be portable to
143 traditional 32-bit x86 and ARM platforms where off_t's width defaults
144 to 32 bits.
145
147 fdatasync(2), fsync(2), msync(2), sync(2)
148Linux man-pages 6.05 2023-07-15 sync_file_range(2)