AIO_WRITE(2) System Calls Manual AIO_WRITE(2)

aio_write, aio_writev — asynchronous write to a file (REALTIME)

Standard C Library (libc, -lc)

#include <aio.h>

int
aio_write(struct aiocb *aiocbp);

#include <sys/uio.h>

int
aio_writev(struct aiocb *aiocbp);

The () system call allows the calling process to write aiocbp->aio_nbytes from the buffer pointed to by aiocbp->aio_buf to the descriptor aiocbp->aio_fildes.

The () system call allows the calling process to write data from multiple buffers (vectored I/O) to the descriptor aiocbp->aio_fildes. For aio_writev(), the aiocbp->aio_iov field points to an array of struct iovec structures, and aiocbp->aio_iovcnt specifies the number of elements in this array. Each iovec structure specifies a buffer and its length for the write operation.

The () function can also be used with lio_listio(2) by setting the aio_lio_opcode field to LIO_WRITEV.

Both calls return immediately after the write request has been enqueued to the descriptor; the write may or may not have completed at the time the call returns. If the request could not be enqueued, generally due to invalid arguments, the call returns without having enqueued the request.

If O_APPEND is set for aiocbp->aio_fildes, () and aio_writev() operations append to the file in the same order as the calls were made. If O_APPEND is not set for the file descriptor, the write operation will occur at the absolute position from the beginning of the file plus aiocbp->aio_offset.

If _POSIX_PRIORITIZED_IO is defined, and the descriptor supports it, then the enqueued operation is submitted at a priority equal to that of the calling process minus aiocbp->aio_reqprio.

The aiocbp pointer may be subsequently used as an argument to () and () in order to determine return or error status for the enqueued operation while it is in progress.

If the request is successfully enqueued, the value of aiocbp->aio_offset can be modified during the request as context, so this value must not be referenced after the request is enqueued.

All asynchronous I/O control buffers contain a sigevent structure in the aio_sigevent field which can be used to request notification when an operation completes.

For SIGEV_KEVENT notifications, the sigev_notify field should be set to SIGEV_KEVENT, and its sigevent's sigev_signo field should contain the descriptor of the kqueue that the event should be attached to. The posted kevent will contain:

ident asynchronous I/O control buffer pointer
filter
udata value stored in aiocbp->aio_sigevent.sigev_value

The Asynchronous I/O Control Block structure pointed to by aiocbp and the buffer that the aiocbp->aio_buf member of that structure references must remain valid until the operation has completed. For this reason, use of auto (stack) variables for these objects is discouraged.

For (), the array of struct iovec structures pointed to by aiocbp->aio_buf and all buffers referenced by the iov_base members of those structures must remain valid until the operation completes.

The asynchronous I/O control buffer aiocbp should be zeroed before the () or aio_writev() system call to avoid passing bogus context information to the kernel.

Modifications of the Asynchronous I/O Control Block structure or the buffer contents after the request has been enqueued, but before the request has completed, are not allowed.

If the file offset in aiocbp->aio_offset is past the offset maximum for aiocbp->aio_fildes, no I/O will occur.

The aio_write() and aio_writev() functions return the value 0 if successful; otherwise the value -1 is returned and the global variable errno is set to indicate the error.

The aio_write() and aio_writev() system calls will fail if:

[]
Due to system resource limitations, the request was not queued.
[]
The aio_write() or aio_writev() system call is not supported.

The following conditions may be synchronously detected when the aio_write() or aio_writev() system call is made, or asynchronously, at any time thereafter. If they are detected at call time, aio_write() or aio_writev() returns -1 and sets errno appropriately; otherwise the aio_return() system call must be called, and will return -1, and aio_error() must be called to determine the actual value that would have been returned in errno.

[]
Failed to transfer the asynchronous I/O control buffer aiocbp into the kernel's memory space, possibly due to an invalid or incorrect address.
[]
The aiocbp->aio_fildes argument is invalid, or is not opened for writing.
[]
The offset aiocbp->aio_offset is not valid.
[]
The priority specified by aiocbp->aio_reqprio is not a valid priority.
[]
The number of bytes specified by aiocbp->aio_nbytes is not valid.

The aio_writev() calls may also return one of the following errors:

[]
Part of the aio_iov points outside the process's allocated address space.
[]
aio_iovcnt was less than or equal to 0, or greater than AIO_UIO_MAXIOV.
[]
One of the iov_len values in the aio_iov array was negative.
[]
The sum of the iov_len values in the aio_iov array overflowed a 32-bit integer.
[]
The constant in aiocbp->aio_sigevent.sigev_notify is set to SIGEV_THREAD (SIGEV_THREAD is not supported).

If the request is successfully enqueued, but subsequently canceled or an error occurs, the value returned by the aio_return() system call is per the write(2) system call, and the value returned by the aio_error() system call is either one of the error returns from the write(2) system call, or one of:

[]
The aiocbp->aio_fildes argument is invalid for writing.
[]
The request was explicitly canceled via a call to aio_cancel().

aio_cancel(2), aio_error(2), aio_return(2), aio_suspend(2), aio(4)

The aio_write() system call is expected to conform to the IEEE Std 1003.1 (“POSIX.1”) standard. The aio_writev() system call is an extension to the IEEE Std 1003.1 (“POSIX.1”) standard.

The aio_write() system call first appeared in FreeBSD 3.0. The aio_writev() system call first appeared in Mac OS X 27.0.

This manual page was written by Wes Peters ⟨wes@softweyr.com⟩.

Invalid information in aiocbp->_aiocb_private may confuse the kernel.

September 18, 2008 Mac OS X 12