1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
|
.\" SPDX-License-Identifier: Linux-man-pages-1-para
.\"
.\" This man page is Copyright (C) 1998 Pawel Krawczyk.
.\"
.\" $Id: sendfile.2,v 1.5 1999/05/18 11:54:11 freitag Exp $
.\" 2000-11-19 bert hubert <ahu@ds9a.nl>: in_fd cannot be socket
.\"
.\" 2004-12-17, mtk
.\" updated description of in_fd and out_fd for 2.6
.\" Various wording and formatting changes
.\"
.\" 2005-03-31 Martin Pool <mbp@sourcefrog.net> mmap() improvements
.\"
.TH sendfile 2 2023-07-15 "Linux man-pages 6.05.01"
.SH NAME
sendfile \- transfer data between file descriptors
.SH LIBRARY
Standard C library
.RI ( libc ", " \-lc )
.SH SYNOPSIS
.nf
.B #include <sys/sendfile.h>
.PP
.BI "ssize_t sendfile(int" " out_fd" ", int" " in_fd" ", \
off_t *_Nullable " offset ,
.BI " size_t" " count" );
.\" The below is too ugly. Comments about glibc versions belong
.\" in the notes, not in the header.
.\"
.\" .B #include <features.h>
.\" .B #if (__GLIBC__==2 && __GLIBC_MINOR__>=1) || __GLIBC__>2
.\" .B #include <sys/sendfile.h>
.\" #else
.\" .B #include <sys/types.h>
.\" .B /* No system prototype before glibc 2.1. */
.\" .BI "ssize_t sendfile(int" " out_fd" ", int" " in_fd" ", off_t *" \
.\" offset ", size_t" " count" )
.\" .B #endif
.\"
.fi
.SH DESCRIPTION
.BR sendfile ()
copies data between one file descriptor and another.
Because this copying is done within the kernel,
.BR sendfile ()
is more efficient than the combination of
.BR read (2)
and
.BR write (2),
which would require transferring data to and from user space.
.PP
.I in_fd
should be a file descriptor opened for reading and
.I out_fd
should be a descriptor opened for writing.
.PP
If
.I offset
is not NULL, then it points
to a variable holding the file offset from which
.BR sendfile ()
will start reading data from
.IR in_fd .
When
.BR sendfile ()
returns, this variable
will be set to the offset of the byte following the last byte that was read.
If
.I offset
is not NULL, then
.BR sendfile ()
does not modify the file offset of
.IR in_fd ;
otherwise the file offset is adjusted to reflect
the number of bytes read from
.IR in_fd .
.PP
If
.I offset
is NULL, then data will be read from
.I in_fd
starting at the file offset,
and the file offset will be updated by the call.
.PP
.I count
is the number of bytes to copy between the file descriptors.
.PP
The
.I in_fd
argument must correspond to a file which supports
.BR mmap (2)-like
operations
(i.e., it cannot be a socket).
.PP
Before Linux 2.6.33,
.I out_fd
must refer to a socket.
Since Linux 2.6.33 it can be any file.
If it is a regular file, then
.BR sendfile ()
changes the file offset appropriately.
.SH RETURN VALUE
If the transfer was successful, the number of bytes written to
.I out_fd
is returned.
Note that a successful call to
.BR sendfile ()
may write fewer bytes than requested;
the caller should be prepared to retry the call if there were unsent bytes.
See also NOTES.
.PP
On error, \-1 is returned, and
.I errno
is set to indicate the error.
.SH ERRORS
.TP
.B EAGAIN
Nonblocking I/O has been selected using
.B O_NONBLOCK
and the write would block.
.TP
.B EBADF
The input file was not opened for reading or the output file
was not opened for writing.
.TP
.B EFAULT
Bad address.
.TP
.B EINVAL
Descriptor is not valid or locked, or an
.BR mmap (2)-like
operation is not available for
.IR in_fd ,
or
.I count
is negative.
.TP
.B EINVAL
.I out_fd
has the
.B O_APPEND
flag set.
This is not currently supported by
.BR sendfile ().
.TP
.B EIO
Unspecified error while reading from
.IR in_fd .
.TP
.B ENOMEM
Insufficient memory to read from
.IR in_fd .
.TP
.B EOVERFLOW
.I count
is too large, the operation would result in exceeding the maximum size of either
the input file or the output file.
.TP
.B ESPIPE
.I offset
is not NULL but the input file is not seekable.
.SH VERSIONS
Other UNIX systems implement
.BR sendfile ()
with different semantics and prototypes.
It should not be used in portable programs.
.SH STANDARDS
None.
.SH HISTORY
Linux 2.2,
glibc 2.1.
.PP
In Linux 2.4 and earlier,
.I out_fd
could also refer to a regular file;
this possibility went away in the Linux 2.6.x kernel series,
but was restored in Linux 2.6.33.
.PP
The original Linux
.BR sendfile ()
system call was not designed to handle large file offsets.
Consequently, Linux 2.4 added
.BR sendfile64 (),
with a wider type for the
.I offset
argument.
The glibc
.BR sendfile ()
wrapper function transparently deals with the kernel differences.
.SH NOTES
.BR sendfile ()
will transfer at most 0x7ffff000 (2,147,479,552) bytes,
returning the number of bytes actually transferred.
.\" commit e28cc71572da38a5a12c1cfe4d7032017adccf69
(This is true on both 32-bit and 64-bit systems.)
.PP
If you plan to use
.BR sendfile ()
for sending files to a TCP socket, but need
to send some header data in front of the file contents, you will find
it useful to employ the
.B TCP_CORK
option, described in
.BR tcp (7),
to minimize the number of packets and to tune performance.
.PP
Applications may wish to fall back to
.BR read (2)
and
.BR write (2)
in the case where
.BR sendfile ()
fails with
.B EINVAL
or
.BR ENOSYS .
.PP
If
.I out_fd
refers to a socket or pipe with zero-copy support, callers must ensure the
transferred portions of the file referred to by
.I in_fd
remain unmodified until the reader on the other end of
.I out_fd
has consumed the transferred data.
.PP
The Linux-specific
.BR splice (2)
call supports transferring data between arbitrary file descriptors
provided one (or both) of them is a pipe.
.SH SEE ALSO
.BR copy_file_range (2),
.BR mmap (2),
.BR open (2),
.BR socket (2),
.BR splice (2)
|
