[OpenBSD]

Manual Page Search Parameters
SELECT(2) System Calls Manual SELECT(2)

select, pselect, FD_SET, FD_CLR, FD_ISSET, FD_ZERO —
synchronous I/O multiplexing

#include <sys/select.h>

int
select(int nfds, fd_set *readfds, fd_set *writefds, fd_set *exceptfds, struct timeval *timeout);

int
pselect(int nfds, fd_set *readfds, fd_set *writefds, fd_set *exceptfds, const struct timespec *timeout, const sigset_t *mask);

FD_SET(fd, &fdset);

FD_CLR(fd, &fdset);

FD_ISSET(fd, &fdset);

FD_ZERO(&fdset);

select() examines the I/O descriptor sets whose addresses are passed in readfds, writefds, and exceptfds to see if some of their descriptors are ready for reading, are ready for writing, or have an exceptional condition pending, respectively. Exceptional conditions include the presence of out-of-band data on a socket. The first nfds descriptors are checked in each set; i.e., the descriptors from 0 through nfds-1 in the descriptor sets are examined. On return, select() replaces the given descriptor sets with subsets consisting of those descriptors that are ready for the requested operation. select() returns the total number of ready descriptors in all the sets.

The descriptor sets are stored as bit fields in arrays of integers. The following macros are provided for manipulating such descriptor sets: FD_ZERO(&fdset) initializes a descriptor set fdset to the null set. FD_SET(fd, &fdset) includes a particular descriptor fd in fdset. FD_CLR(fd, &fdset) removes fd from fdset. FD_ISSET(fd, &fdset) is non-zero if fd is a member of fdset, zero otherwise. The behavior of these macros is undefined if a descriptor value is less than zero or greater than or equal to FD_SETSIZE, which is normally at least equal to the maximum number of descriptors supported by the system.

If timeout is a non-null pointer, it specifies a maximum interval to wait for the selection to complete. If timeout is a null pointer, the select blocks indefinitely. To effect a poll, the timeout argument should be non-null, pointing to a zero-valued timeval structure. timeout is not changed by select(), and may be reused on subsequent calls; however, it is good style to re-initialize it before each invocation of select().

Any of readfds, writefds, and exceptfds may be given as null pointers if no descriptors are of interest.

The pselect() function is similar to select() except that it specifies the timeout using a timespec structure. Also, if mask is a non-null pointer, pselect() atomically sets the calling thread's signal mask to the signal set pointed to by mask for the duration of the function call. In this case, the original signal mask will be restored before pselect() returns.

If successful, select() and pselect() return the number of ready descriptors that are contained in the descriptor sets. If a descriptor is included in multiple descriptor sets, each inclusion is counted separately. If the time limit expires before any descriptors become ready, they return 0.

Otherwise, if select() or pselect() return with an error, including one due to an interrupted call, they return -1, and the descriptor sets will be unmodified.

An error return from select() or pselect() indicates:
[]
One or more of readfds, writefds, or exceptfds points outside the process's allocated address space.
[]
One of the descriptor sets specified an invalid descriptor.
[]
A signal was delivered before the time limit expired and before any of the selected descriptors became ready.
[]
The specified time limit is invalid. One of its components is negative or too large.
[]
nfds was less than 0.

accept(2), clock_gettime(2), connect(2), gettimeofday(2), poll(2), read(2), recv(2), send(2), write(2), getdtablesize(3)

The select() and pselect() functions conform to IEEE Std 1003.1-2008 (“POSIX.1”).

The select() system call first appeared in 4.1cBSD. The pselect() system call has been available since OpenBSD 5.4.

Although the provision of getdtablesize(3) was intended to allow user programs to be written independent of the kernel limit on the number of open files, the dimension of a sufficiently large bit field for select remains a problem. The default bit size of fd_set is based on the symbol FD_SETSIZE (currently 1024), but that is somewhat smaller than the current kernel limit to the number of open files. However, in order to accommodate programs which might potentially use a larger number of open files with select, it is possible to increase this size within a program by providing a larger definition of FD_SETSIZE before the inclusion of any headers. The kernel will cope, and the userland libraries provided with the system are also ready for large numbers of file descriptors.

Alternatively, to be really safe, it is possible to allocate fd_set bit-arrays dynamically. The idea is to permit a program to work properly even if it is execve(2)'d with 4000 file descriptors pre-allocated. The following illustrates the technique which is used by userland libraries:

fd_set *fdsr;
int max = fd;

fdsr = calloc(howmany(max+1, NFDBITS), sizeof(fd_mask));
if (fdsr == NULL) {
	...
	return (-1);
}
FD_SET(fd, fdsr);
n = select(max+1, fdsr, NULL, NULL, &tv);
...
free(fdsr);

Alternatively, it is possible to use the poll(2) interface. poll(2) is more efficient when the size of select()'s fd_set bit-arrays are very large, and for fixed numbers of file descriptors one need not size and dynamically allocate a memory object.

select() should probably have been designed to return the time remaining from the original timeout, if any, by modifying the time value in place. Even though some systems stupidly act in this different way, it is unlikely this semantic will ever be commonly implemented, as the change causes massive source code compatibility problems. Furthermore, recent new standards have dictated the current behaviour. In general, due to the existence of those brain-damaged non-conforming systems, it is unwise to assume that the timeout value will be unmodified by the select() call, and the caller should reinitialize it on each invocation. Calculating the delta is easily done by calling gettimeofday(2) before and after the call to select(), and using timersub() (as described in getitimer(2)).

Internally to the kernel, select() and pselect() work poorly if multiple processes wait on the same file descriptor.

September 17, 2016 OpenBSD-current