select_tut(2)

—

Options · 11

1.	You should always try to use select() without a timeout. Your program should have nothing to do if there is no data available. Code that depends on timeouts is not usually portable and is difficult to debug.
2.	The value nfds must be properly calculated for efficiency as explained above.
3.	No file descriptor must be added to any set if you do not intend to check its result after the select() call, and respond appropriately. See next rule.
4.	After select() returns, all file descriptors in all sets should be checked to see if they are ready.
5.	The functions read(2), recv(2), write(2), and send(2) do not necessarily read/write the full amount of data that you have requested. If they do read/write the full amount, it's because you have a low traffic load and a fast stream. This is not always going to be the case. You should cope with the case of your functions managing to send or receive only a single byte.
6.	Never read/write only in single bytes at a time unless you are really sure that you have a small amount of data to process. It is extremely inefficient not to read/write as much data as you can buffer each time. The buffers in the example below are 1024 bytes although they could easily be made larger.
7.	Calls to read(2), recv(2), write(2), send(2), and select() can fail with the error EINTR, and calls to read(2), recv(2), write(2), and send(2) can fail with errno set to EAGAIN (EWOULDBLOCK). These results must be properly managed (not done properly above). If your program is not going to receive any signals, then it is unlikely you will get EINTR. If your program does not set nonblocking I/O, you will not get EAGAIN.
8.	Never call read(2), recv(2), write(2), or send(2) with a buffer length of zero.
9.	If the functions read(2), recv(2), write(2), and send(2) fail with errors other than those listed in 7., or one of the input functions returns 0, indicating end of file, then you should not pass that file descriptor to select() again. In the example below, I close the file descriptor immediately, and then set it to -1 to prevent it being included in a set.
10.	The timeout value must be initialized with each new call to select(), since some operating systems modify the structure. pselect() however does not modify its timeout structure.
11.	Since select() modifies its file descriptor sets, if the call is being used in a loop, then the sets must be reinitialized before each call.

NAME

select, pselect - synchronous I/O multiplexing

The select() and pselect() system calls are used to efficiently monitor multiple file descriptors, to see if any of them is, or becomes, "ready"; that is, to see whether I/O becomes possible, or an "exceptional condition" has occurred on any of the file descriptors.

This page provides background and tutorial information on the use of these system calls. For details of the arguments and semantics of select() and pselect(), see select(2).

Combining signal and data events

pselect() is useful if you are waiting for a signal as well as for file descriptor(s) to become ready for I/O. Programs that receive signals normally use the signal handler only to raise a global flag. The global flag will indicate that the event must be processed in the main loop of the program. A signal will cause the select() (or pselect()) call to return with errno set to EINTR. This behavior is essential so that signals can be processed in the main loop of the program, otherwise select() would block indefinitely.

Now, somewhere in the main loop will be a conditional to check the global flag. So we must ask: what if a signal arrives after the conditional, but before the select() call? The answer is that select() would block indefinitely, even though an event is actually pending. This race condition is solved by the pselect() call. This call can be used to set the signal mask to a set of signals that are to be received only within the pselect() call. For instance, let us say that the event in question was the exit of a child process. Before the start of the main loop, we would block SIGCHLD using sigprocmask(2). Our pselect() call would enable SIGCHLD by using an empty signal mask. Our program would look like:

bash

static volatile sig_atomic_t got_SIGCHLD = 0;
static void
child_sig_handler(int sig)
{
\n
    got_SIGCHLD = 1;
}
int
main(int argc, char *argv[])
{
\n
    sigset_t sigmask, empty_mask;
\n
    struct sigaction sa;
\n
    fd_set readfds, writefds, exceptfds;
\n
    int r;
\n
    sigemptyset(&sigmask);
\n
    sigaddset(&sigmask, SIGCHLD);
\n
    if (sigprocmask(SIG_BLOCK, &sigmask, NULL) == -1) {
\n
        perror("sigprocmask");
\n
        exit(EXIT_FAILURE);
\n
    }
\n
    sa.sa_flags = 0;
\n
    sa.sa_handler = child_sig_handler;
\n
    sigemptyset(&sa.sa_mask);
\n
    if (sigaction(SIGCHLD, &sa, NULL) == -1) {
\n
        perror("sigaction");
\n
        exit(EXIT_FAILURE);
\n
    }
\n
    sigemptyset(&empty_mask);
\n
    for (;;) {          /* main loop */
\n
        /* Initialize readfds, writefds, and exceptfds
\n
           before the pselect() call. (Code omitted.) */
\n
        r = pselect(nfds, &readfds, &writefds, &exceptfds,
\n
                    NULL, &empty_mask);
\n
        if (r == -1 && errno != EINTR) {
\n
            /* Handle error */
\n
        }
\n
        if (got_SIGCHLD) {
\n
            got_SIGCHLD = 0;
\n
            /* Handle signalled event here; e.g., wait() for all
\n
               terminated children. (Code omitted.) */
\n
        }
\n
        /* main body of program */
\n
    }
}

Practical

So what is the point of select()? Can't I just read and write to my file descriptors whenever I want? The point of select() is that it watches multiple descriptors at the same time and properly puts the process to sleep if there is no activity. UNIX programmers often find themselves in a position where they have to handle I/O from more than one file descriptor where the data flow may be intermittent. If you were to merely create a sequence of read(2) and write(2) calls, you would find that one of your calls may block waiting for data from/to a file descriptor, while another file descriptor is unused though ready for I/O. select() efficiently copes with this situation.

Select law

Many people who try to use select() come across behavior that is difficult to understand and produces nonportable or borderline results. For instance, the above program is carefully written not to block at any point, even though it does not set its file descriptors to nonblocking mode. It is easy to introduce subtle errors that will remove the advantage of using select(), so here is a list of essentials to watch for when using select().

1.: You should always try to use select() without a timeout. Your program should have nothing to do if there is no data available. Code that depends on timeouts is not usually portable and is difficult to debug.
2.: The value nfds must be properly calculated for efficiency as explained above.
3.: No file descriptor must be added to any set if you do not intend to check its result after the select() call, and respond appropriately. See next rule.
4.: After select() returns, all file descriptors in all sets should be checked to see if they are ready.
5.: The functions read(2), recv(2), write(2), and send(2) do not necessarily read/write the full amount of data that you have requested. If they do read/write the full amount, it's because you have a low traffic load and a fast stream. This is not always going to be the case. You should cope with the case of your functions managing to send or receive only a single byte.
6.: Never read/write only in single bytes at a time unless you are really sure that you have a small amount of data to process. It is extremely inefficient not to read/write as much data as you can buffer each time. The buffers in the example below are 1024 bytes although they could easily be made larger.
7.: Calls to read(2), recv(2), write(2), send(2), and select() can fail with the error EINTR, and calls to read(2), recv(2), write(2), and send(2) can fail with errno set to EAGAIN (EWOULDBLOCK). These results must be properly managed (not done properly above). If your program is not going to receive any signals, then it is unlikely you will get EINTR. If your program does not set nonblocking I/O, you will not get EAGAIN.
8.: Never call read(2), recv(2), write(2), or send(2) with a buffer length of zero.
9.: If the functions read(2), recv(2), write(2), and send(2) fail with errors other than those listed in 7., or one of the input functions returns 0, indicating end of file, then you should not pass that file descriptor to select() again. In the example below, I close the file descriptor immediately, and then set it to -1 to prevent it being included in a set.
10.: The timeout value must be initialized with each new call to select(), since some operating systems modify the structure. pselect() however does not modify its timeout structure.
11.: Since select() modifies its file descriptor sets, if the call is being used in a loop, then the sets must be reinitialized before each call.

RETURN VALUE

See select(2).

NOTES

Generally speaking, all operating systems that support sockets also support select(). select() can be used to solve many problems in a portable and efficient way that naive programmers try to solve in a more complicated manner using threads, forking, IPCs, signals, memory sharing, and so on.

The poll(2) system call has the same functionality as select(), and is somewhat more efficient when monitoring sparse file descriptor sets. It is nowadays widely available, but historically was less portable than select().

The Linux-specific epoll(7) API provides an interface that is more efficient than select(2) and poll(2) when monitoring large numbers of file descriptors.

EXAMPLES

Here is an example that better demonstrates the true utility of select(). The listing below is a TCP forwarding program that forwards from one TCP port to another.

bash

#include <arpa/inet.h>
#include <errno.h>
#include <netinet/in.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/select.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <unistd.h>
static int forward_port;
#undef max
#define max(x, y) ((x) > (y) ? (x) : (y))
static int
listen_socket(int listen_port)
{
\n
    int                 lfd;
\n
    int                 yes;
\n
    struct sockaddr_in  addr;
\n
    lfd = socket(AF_INET, SOCK_STREAM, 0);
\n
    if (lfd == -1) {
\n
        perror("socket");
\n
        return -1;
\n
    }
\n
    yes = 1;
\n
    if (setsockopt(lfd, SOL_SOCKET, SO_REUSEADDR,
\n
                   &yes, sizeof(yes)) == -1)
\n
    {
\n
        perror("setsockopt");
\n
        close(lfd);
\n
        return -1;
\n
    }
\n
    memset(&addr, 0, sizeof(addr));
\n
    addr.sin_port = htons(listen_port);
\n
    addr.sin_family = AF_INET;
\n
    if (bind(lfd, (struct sockaddr *) &addr, sizeof(addr)) == -1) {
\n
        perror("bind");
\n
        close(lfd);
\n
        return -1;
\n
    }
\n
    printf("accepting connections on port %d\n", listen_port);
\n
    listen(lfd, 10);
\n
    return lfd;
}
static int
connect_socket(int connect_port, char *address)
{
\n
    int                 cfd;
\n
    struct sockaddr_in  addr;
\n
    cfd = socket(AF_INET, SOCK_STREAM, 0);
\n
    if (cfd == -1) {
\n
        perror("socket");
\n
        return -1;
\n
    }
\n
    memset(&addr, 0, sizeof(addr));
\n
    addr.sin_port = htons(connect_port);
\n
    addr.sin_family = AF_INET;
\n
    if (!inet_aton(address, (struct in_addr *) &addr.sin_addr.s_addr)) {
\n
        fprintf(stderr, "inet_aton(): bad IP address format\n");
\n
        close(cfd);
\n
        return -1;
\n
    }
\n
    if (connect(cfd, (struct sockaddr *) &addr, sizeof(addr)) == -1) {
\n
        perror("connect()");
\n
        shutdown(cfd, SHUT_RDWR);
\n
        close(cfd);
\n
        return -1;
\n
    }
\n
    return cfd;
}
#define SHUT_FD1 do {                                \
\n
                     if (fd1 >= 0) {                 \
\n
                         shutdown(fd1, SHUT_RDWR);   \
\n
                         close(fd1);                 \
\n
                         fd1 = -1;                   \
\n
                     }                               \
\n
                 } while (0)
#define SHUT_FD2 do {                                \
\n
                     if (fd2 >= 0) {                 \
\n
                         shutdown(fd2, SHUT_RDWR);   \
\n
                         close(fd2);                 \
\n
                         fd2 = -1;                   \
\n
                     }                               \
\n
                 } while (0)
#define BUF_SIZE 1024
int
main(int argc, char *argv[])
{
\n
    int      h;
\n
    int      ready, nfds;
\n
    int      fd1 = -1, fd2 = -1;
\n
    int      buf1_avail = 0, buf1_written = 0;
\n
    int      buf2_avail = 0, buf2_written = 0;
\n
    char     buf1[BUF_SIZE], buf2[BUF_SIZE];
\n
    fd_set   readfds, writefds, exceptfds;
\n
    ssize_t  nbytes;
\n
    if (argc != 4) {
\n
        fprintf(stderr, "Usage\n\tfwd <listen-port> "
\n
                "<forward-to-port> <forward-to-ip-address>\n");
\n
        exit(EXIT_FAILURE);
\n
    }
\n
    signal(SIGPIPE, SIG_IGN);
\n
    forward_port = atoi(argv[2]);
\n
    h = listen_socket(atoi(argv[1]));
\n
    if (h == -1)
\n
        exit(EXIT_FAILURE);
\n
    for (;;) {
\n
        nfds = 0;
\n
        FD_ZERO(&readfds);
\n
        FD_ZERO(&writefds);
\n
        FD_ZERO(&exceptfds);
\n
        FD_SET(h, &readfds);
\n
        nfds = max(nfds, h);
\n
        if (fd1 > 0 && buf1_avail < BUF_SIZE)
\n
            FD_SET(fd1, &readfds);
\n
            /* Note: nfds is updated below, when fd1 is added to
\n
               exceptfds. */
\n
        if (fd2 > 0 && buf2_avail < BUF_SIZE)
\n
            FD_SET(fd2, &readfds);
\n
        if (fd1 > 0 && buf2_avail - buf2_written > 0)
\n
            FD_SET(fd1, &writefds);
\n
        if (fd2 > 0 && buf1_avail - buf1_written > 0)
\n
            FD_SET(fd2, &writefds);
\n
        if (fd1 > 0) {
\n
            FD_SET(fd1, &exceptfds);
\n
            nfds = max(nfds, fd1);
\n
        }
\n
        if (fd2 > 0) {
\n
            FD_SET(fd2, &exceptfds);
\n
            nfds = max(nfds, fd2);
\n
        }
\n
        ready = select(nfds + 1, &readfds, &writefds, &exceptfds, NULL);
\n
        if (ready == -1 && errno == EINTR)
\n
            continue;
\n
        if (ready == -1) {
\n
            perror("select()");
\n
            exit(EXIT_FAILURE);
\n
        }
\n
        if (FD_ISSET(h, &readfds)) {
\n
            socklen_t addrlen;
\n
            struct sockaddr_in client_addr;
\n
            int fd;
\n
            addrlen = sizeof(client_addr);
\n
            memset(&client_addr, 0, addrlen);
\n
            fd = accept(h, (struct sockaddr *) &client_addr, &addrlen);
\n
            if (fd == -1) {
\n
                perror("accept()");
\n
            } else {
\n
                SHUT_FD1;
\n
                SHUT_FD2;
\n
                buf1_avail = buf1_written = 0;
\n
                buf2_avail = buf2_written = 0;
\n
                fd1 = fd;
\n
                fd2 = connect_socket(forward_port, argv[3]);
\n
                if (fd2 == -1)
\n
                    SHUT_FD1;
\n
                else
\n
                    printf("connect from %s\n",
\n
                           inet_ntoa(client_addr.sin_addr));
\n
                /* Skip any events on the old, closed file
\n
                   descriptors. */
\n
                continue;
\n
            }
\n
        }
\n
        /* NB: read OOB data before normal reads. */
\n
        if (fd1 > 0 && FD_ISSET(fd1, &exceptfds)) {
\n
            char c;
\n
            nbytes = recv(fd1, &c, 1, MSG_OOB);
\n
            if (nbytes < 1)
\n
                SHUT_FD1;
\n
            else
\n
                send(fd2, &c, 1, MSG_OOB);
\n
        }
\n
        if (fd2 > 0 && FD_ISSET(fd2, &exceptfds)) {
\n
            char c;
\n
            nbytes = recv(fd2, &c, 1, MSG_OOB);
\n
            if (nbytes < 1)
\n
                SHUT_FD2;
\n
            else
\n
                send(fd1, &c, 1, MSG_OOB);
\n
        }
\n
        if (fd1 > 0 && FD_ISSET(fd1, &readfds)) {
\n
            nbytes = read(fd1, buf1 + buf1_avail,
\n
                          BUF_SIZE - buf1_avail);
\n
            if (nbytes < 1)
\n
                SHUT_FD1;
\n
            else
\n
                buf1_avail += nbytes;
\n
        }
\n
        if (fd2 > 0 && FD_ISSET(fd2, &readfds)) {
\n
            nbytes = read(fd2, buf2 + buf2_avail,
\n
                          BUF_SIZE - buf2_avail);
\n
            if (nbytes < 1)
\n
                SHUT_FD2;
\n
            else
\n
                buf2_avail += nbytes;
\n
        }
\n
        if (fd1 > 0 && FD_ISSET(fd1, &writefds) && buf2_avail > 0) {
\n
            nbytes = write(fd1, buf2 + buf2_written,
\n
                           buf2_avail - buf2_written);
\n
            if (nbytes < 1)
\n
                SHUT_FD1;
\n
            else
\n
                buf2_written += nbytes;
\n
        }
\n
        if (fd2 > 0 && FD_ISSET(fd2, &writefds) && buf1_avail > 0) {
\n
            nbytes = write(fd2, buf1 + buf1_written,
\n
                           buf1_avail - buf1_written);
\n
            if (nbytes < 1)
\n
                SHUT_FD2;
\n
            else
\n
                buf1_written += nbytes;
\n
        }
\n
        /* Check if write data has caught read data. */
\n
        if (buf1_written == buf1_avail)
\n
            buf1_written = buf1_avail = 0;
\n
        if (buf2_written == buf2_avail)
\n
            buf2_written = buf2_avail = 0;
\n
        /* One side has closed the connection, keep
\n
           writing to the other side until empty. */
\n
        if (fd1 < 0 && buf1_avail - buf1_written == 0)
\n
            SHUT_FD2;
\n
        if (fd2 < 0 && buf2_avail - buf2_written == 0)
\n
            SHUT_FD1;
\n
    }
\n
    exit(EXIT_SUCCESS);
}

The above program properly forwards most kinds of TCP connections including OOB signal data transmitted by telnet servers. It handles the tricky problem of having data flow in both directions simultaneously. You might think it more efficient to use a fork(2) call and devote a thread to each stream. This becomes more tricky than you might suspect. Another idea is to set nonblocking I/O using fcntl(2). This also has its problems because you end up using inefficient timeouts.

The program does not handle more than one simultaneous connection at a time, although it could easily be extended to do this with a linked list of buffers—one for each connection. At the moment, new connections cause the current connection to be dropped.

NAME

LIBRARY

SYNOPSIS

DESCRIPTION