Open System Services System Calls Reference Manual (G06.29+, H06.08+, J06.03+)

writev(2) OSS System Calls Reference Manual
With devices incapable of seeking, writing always takes place starting at the current position.
For such devices, the value of the file pointer after a call to the writev( ) function is always 0
(zero).
Fewer bytes than requested can be written if there is not enough room to satisfy the request. In
this case, the number of bytes written is returned. For example, suppose there is space for 20
bytes more in a file before reaching a limit. A write request of 512 bytes returns a value of 20.
The limit reached can be either the end of the physical medium or the value that has been set by
the ulimit() function. The next write of a nonzero number of bytes gives a failure return (except
as noted later).
Upon successful completion, the writev() function returns the number of bytes actually written
to the file associated with filedes.
If the O_APPEND status flag of the file is set, the file offset is set to the end of the file prior to
each write.
Write requests to a pipe or FIFO file are handled the same as writes to a regular file with these
exceptions:
• No file offset is associated with a pipe; therfore, each writev() request appends to the
end of the pipe.
• If the size of the writev() request is less than or equal to the value of the PIPE_BUF
system variable, the writev() function is guaranteed to be atomic. The data is not inter-
leaved with data from other processes doing writes on the same pipe.
• If the size of the writev() request is greater than the value of the PIPE_BUF system
variable, the file system attempts to resize the pipe buffer from 2 * PIPE_BUF to 65,536
bytes. If the resizing is successful, the file system performs atomic writes of up to 32,768
bytes and can transfer up to 52 kilobytes of data from the pipe buffer on subsequent
read() or readv() calls by the client.
If the file system cannot resize the buffer, it continues to use the existing buffer. A
second attempt at resizing occurs after approximately a minute elapses.
Writes of greater than PIPE_BUF bytes can have data interleaved, on arbitrary boun-
daries, with writes by other processes, whether or not the O_NONBLOCK flag is set.
• If the O_NONBLOCK flag is not set, a writev() request to a full pipe causes the process
to block until enough space becomes available to handle the entire request.
• If the O_NONBLOCK flag is set, writev() requests are handled differently in these
ways:
— The writev( ) function does block the process.
— The writev( ) requests for PIPE_BUF or fewer bytes either succeed completely
and return the number of bytes written, or return the value -1 and set errno to
[EAGAIN].
—Awritev() request for greater than PIPE_BUF bytes either transfers what it can
and returns the number of bytes written, or transfers no data and returns the value
-1 with errno set to [EAGAIN]. Also, if a request is greater than PIPE_BUF
bytes and all data previously written to the pipe has been read, writev() transfers
at least PIPE_BUF bytes.
When attempting to write to a file descriptor for a special character device (a terminal) that can-
not accept data immediately:
10−26
Hewlett-Packard Company 527186-023