freebsd-nq/lib/libc/sys/sendfile.2
Yaroslav Tykhiy df19774d2f Note that, thanks to the work by Alan Cox et al, some arch'es
don't need sendfile() buffers any more.

The report on the work referenced can be found at
http://usenix.org/events/usenix05/tech/general/elmeleegy.html

MFC after:	1 week
2006-11-24 11:44:19 +00:00

296 lines
7.0 KiB
Groff

.\" Copyright (c) 2003, David G. Lawrence
.\" All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice unmodified, this list of conditions, and the following
.\" disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.\" $FreeBSD$
.\"
.Dd November 24, 2006
.Dt SENDFILE 2
.Os
.Sh NAME
.Nm sendfile
.Nd send a file to a socket
.Sh LIBRARY
.Lb libc
.Sh SYNOPSIS
.In sys/types.h
.In sys/socket.h
.In sys/uio.h
.Ft int
.Fo sendfile
.Fa "int fd" "int s" "off_t offset" "size_t nbytes"
.Fa "struct sf_hdtr *hdtr" "off_t *sbytes" "int flags"
.Fc
.Sh DESCRIPTION
The
.Fn sendfile
system call
sends a regular file specified by descriptor
.Fa fd
out a stream socket specified by descriptor
.Fa s .
.Pp
The
.Fa offset
argument specifies where to begin in the file.
Should
.Fa offset
fall beyond the end of file, the system will return
success and report 0 bytes sent as described below.
The
.Fa nbytes
argument specifies how many bytes of the file should be sent, with 0 having the special
meaning of send until the end of file has been reached.
.Pp
An optional header and/or trailer can be sent before and after the file data by specifying
a pointer to a
.Vt "struct sf_hdtr" ,
which has the following structure:
.Pp
.Bd -literal -offset indent -compact
struct sf_hdtr {
struct iovec *headers; /* pointer to header iovecs */
int hdr_cnt; /* number of header iovecs */
struct iovec *trailers; /* pointer to trailer iovecs */
int trl_cnt; /* number of trailer iovecs */
};
.Ed
.Pp
The
.Fa headers
and
.Fa trailers
pointers, if
.Pf non- Dv NULL ,
point to arrays of
.Vt "struct iovec"
structures.
See the
.Fn writev
system call for information on the iovec structure.
The number of iovecs in these
arrays is specified by
.Fa hdr_cnt
and
.Fa trl_cnt .
.Pp
If
.Pf non- Dv NULL ,
the system will write the total number of bytes sent on the socket to the
variable pointed to by
.Fa sbytes .
.Pp
The
.Fa flags
argument has one possible value:
.Dv SF_NODISKIO .
This flag causes any
.Fn sendfile
call which would block on disk I/O to instead
return
.Er EBUSY .
Busy servers may benefit by transferring requests that would
block to a separate I/O worker thread.
.Pp
When using a socket marked for non-blocking I/O,
.Fn sendfile
may send fewer bytes than requested.
In this case, the number of bytes successfully
written is returned in
.Fa *sbytes
(if specified),
and the error
.Er EAGAIN
is returned.
.Sh IMPLEMENTATION NOTES
The
.Fx
implementation of
.Fn sendfile
is "zero-copy", meaning that it has been optimized so that copying of the file data is avoided.
.Sh TUNING
On some architectures, this system call internally uses a special
.Fn sendfile
buffer
.Pq Vt "struct sf_buf"
to handle sending file data to the client.
If the sending socket is
blocking, and there are not enough
.Fn sendfile
buffers available,
.Fn sendfile
will block and report a state of
.Dq Li sfbufa .
If the sending socket is non-blocking and there are not enough
.Fn sendfile
buffers available, the call will block and wait for the
necessary buffers to become available before finishing the call.
.Pp
The number of
.Vt sf_buf Ns 's
allocated should be proportional to the number of nmbclusters used to
send data to a client via
.Fn sendfile .
Tune accordingly to avoid blocking!
Busy installations that make extensive use of
.Fn sendfile
may want to increase these values to be inline with their
.Va kern.ipc.nmbclusters
(see
.Xr tuning 7
for details).
.Pp
The number of
.Fn sendfile
buffers available is determined at boot time by either the
.Va kern.ipc.nsfbufs
.Xr loader.conf 5
variable or the
.Dv NSFBUFS
kernel configuration tunable.
The number of
.Fn sendfile
buffers scales with
.Va kern.maxusers .
The
.Va kern.ipc.nsfbufsused
and
.Va kern.ipc.nsfbufspeak
read-only
.Xr sysctl 8
variables show current and peak
.Fn sendfile
buffers usage respectively.
These values may also be viewed through
.Nm netstat Fl m .
.Pp
If a value of zero is reported for
.Va kern.ipc.nsfbufs ,
your architecture does not need to use
.Fn sendfile
buffers because their task can be efficiently performed
by the generic virtual memory structures.
.Sh RETURN VALUES
.Rv -std sendfile
.Sh ERRORS
.Bl -tag -width Er
.It Bq Er EAGAIN
The socket is marked for non-blocking I/O and not all data was sent due to
the socket buffer being filled.
If specified, the number of bytes successfully sent will be returned in
.Fa *sbytes .
.It Bq Er EBADF
The
.Fa fd
argument
is not a valid file descriptor.
.It Bq Er EBADF
The
.Fa s
argument
is not a valid socket descriptor.
.It Bq Er EBUSY
Completing the entire transfer would have required disk I/O, so
it was aborted.
Partial data may have been sent.
(This error can only occur when
.Dv SF_NODISKIO
is specified.)
.It Bq Er EFAULT
An invalid address was specified for an argument.
.It Bq Er EINTR
A signal interrupted
.Fn sendfile
before it could be completed.
If specified, the number
of bytes successfully sent will be returned in
.Fa *sbytes .
.It Bq Er EINVAL
The
.Fa fd
argument
is not a regular file.
.It Bq Er EINVAL
The
.Fa s
argument
is not a SOCK_STREAM type socket.
.It Bq Er EINVAL
The
.Fa offset
argument
is negative.
.It Bq Er EIO
An error occurred while reading from
.Fa fd .
.It Bq Er ENOTCONN
The
.Fa s
argument
points to an unconnected socket.
.It Bq Er ENOTSOCK
The
.Fa s
argument
is not a socket.
.It Bq Er EOPNOTSUPP
The file system for descriptor
.Fa fd
does not support
.Fn sendfile .
.It Bq Er EPIPE
The socket peer has closed the connection.
.El
.Sh SEE ALSO
.Xr netstat 1 ,
.Xr open 2 ,
.Xr send 2 ,
.Xr socket 2 ,
.Xr writev 2 ,
.Xr tuning 7
.Rs
.%A K. Elmeleegy
.%A A. Chanda
.%A A. L. Cox
.%A W. Zwaenepoel
.%T A Portable Kernel Abstraction for Low-Overhead Ephemeral Mapping Management
.%J The Proceedings of the 2005 USENIX Annual Technical Conference
.%P pp 223-236
.%D 2005
.Re
.Sh HISTORY
The
.Fn sendfile
system call
first appeared in
.Fx 3.0 .
This manual page first appeared in
.Fx 3.1 .
.Sh AUTHORS
The
.Fn sendfile
system call
and this manual page were written by
.An David G. Lawrence Aq dg@dglawrence.com .