man poll

Name

poll, ppoll - wait for some event on a file descriptor

Synopsis

#include <poll.h>

int poll(struct pollfd *fds, nfds_t nfds, int timeout);

#define _GNU_SOURCE         /* See feature_test_macros(7) */
#include <poll.h>

int ppoll(struct pollfd *fds, nfds_t nfds,
        const struct timespec *timeout_ts, const sigset_t *sigmask);

Description

poll() performs a similar task to select(2): it waits for one of a set of file descriptors to become ready to perform I/O.

The set of file descriptors to be monitored is specified in the fds argument, which is an array of structures of the following form:

struct pollfd {
    int   fd;         /* file descriptor */
    short events;     /* requested events */
    short revents;    /* returned events */
};

The caller should specify the number of items in the  fds array in  nfds.

The field fd contains a file descriptor for an open file. If this field is negative, then the corresponding events field is ignored and the revents field returns zero. (This provides an easy way of ignoring a file descriptor for a single poll() call: simply negate the fd field.)

The field events is an input parameter, a bit mask specifying the events the application is interested in for the file descriptor fd. If this field is specified as zero, then all events are ignored for fd and revents returns zero.

The field revents is an output parameter, filled by the kernel with the events that actually occurred. The bits returned in revents can include any of those specified in events, or one of the values POLLERR, POLLHUP, or POLLNVAL. (These three bits are meaningless in the events field, and will be set in the revents field whenever the corresponding condition is true.)

If none of the events requested (and no error) has occurred for any of the file descriptors, then poll() blocks until one of the events occurs.

The timeout argument specifies the minimum number of milliseconds that poll() will block. (This interval will be rounded up to the system clock granularity, and kernel scheduling delays mean that the blocking interval may overrun by a small amount.) Specifying a negative value in timeout means an infinite timeout. Specifying a timeout of zero causespoll() to return immediately, even if no file descriptors are ready.

The bits that may be set/returned in events and revents are defined in <poll.h>:

POLLIN

There is data to read.

POLLPRI

There is urgent data to read (e.g., out-of-band data on TCP socket; pseudoterminal master in packet mode has seen state change in slave).

POLLOUT

Writing now will not block.

POLLRDHUP (since Linux 2.6.17)

Stream socket peer closed connection, or shut down writing half of connection. The _GNU_SOURCE feature test macro must be defined (before including any header files) in order to obtain this definition.

POLLERR

Error condition (output only).

POLLHUP

Hang up (output only).

POLLNVAL

Invalid request: fd not open (output only).

When compiling with  _XOPEN_SOURCE defined, one also has the following, which convey no further information beyond the bits listed above:

POLLRDNORM

Equivalent to POLLIN.

POLLRDBAND

Priority band data can be read (generally unused on Linux).

POLLWRNORM

Equivalent to POLLOUT.

POLLWRBAND

Priority data may be written.

Linux also knows about, but does not use  POLLMSG.

ppoll()

The relationship between  poll() and  ppoll() is analogous to the relationship between  select(2) and  pselect(2): like  pselect(2),  ppoll() allows an application to safely wait until either a file descriptor becomes ready or until a signal is caught.

Other than the difference in the precision of the timeout argument, the following ppoll() call:

ready = ppoll(&fds, nfds, timeout_ts, &sigmask);

is equivalent to  atomically executing the following calls:

 sigset_t origmask;
 int timeout;

timeout = (timeout_ts == NULL) ? -1 :
           (timeout_ts.tv_sec * 1000 + timeout_ts.tv_nsec / 1000000);
 sigprocmask(SIG_SETMASK, &sigmask, &origmask);
 ready = poll(&fds, nfds, timeout);
 sigprocmask(SIG_SETMASK, &origmask, NULL);

See the description of  pselect(2) for an explanation of why  ppoll() is necessary.

If the sigmask argument is specified as NULL, then no signal mask manipulation is performed (and thus ppoll() differs from poll() only in the precision of the timeout argument).

The timeout_ts argument specifies an upper limit on the amount of time that ppoll() will block. This argument is a pointer to a structure of the following form:

struct timespec {
    long    tv_sec;         /* seconds */
    long    tv_nsec;        /* nanoseconds */
};

If  timeout_ts is specified as NULL, then  ppoll() can block indefinitely.

Return Value

On success, a positive number is returned; this is the number of structures which have nonzero revents fields (in other words, those descriptors with events or errors reported). A value of 0 indicates that the call timed out and no file descriptors were ready. On error, -1 is returned, and errno is set appropriately.

Errors

EFAULT

The array given as argument was not contained in the calling program's address space.

EINTR

A signal occurred before any requested event; see signal(7).

EINVAL

The nfds value exceeds the RLIMIT_NOFILE value.

ENOMEM

There was no space to allocate file descriptor tables.

Versions

The poll() system call was introduced in Linux 2.1.23. On older kernels that lack this system call, the glibc (and the old Linux libc) poll() wrapper function provides emulation using select(2).

The ppoll() system call was added to Linux in kernel 2.6.16. The ppoll() library call was added in glibc 2.4.

Conforming To

poll() conforms to POSIX.1-2001. ppoll() is Linux-specific.

Notes

Some implementations define the nonstandard constant INFTIM with the value -1 for use as a timeout for poll(). This constant is not provided in glibc.

For a discussion of what may happen if a file descriptor being monitored by poll() is closed in another thread, see select(2).

Linux notes

The Linux  ppoll() system call modifies its  timeout_ts argument. However, the glibc wrapper function hides this behavior by using a local variable for the timeout argument that is passed to the system call. Thus, the glibc  ppoll() function does not modify its  timeout_ts argument.

Bugs

See the discussion of spurious readiness notifications under the BUGS section of select(2).

See Also

select(2), select_tut(2), time(7)

Referenced By