1EPOLL_WAIT(2) Linux Programmer's Manual EPOLL_WAIT(2)
2
6 epoll_wait, epoll_pwait - wait for an I/O event on an epoll file de‐
7 scriptor
8
10 #include <sys/epoll.h>
11 int epoll_wait(int epfd, struct epoll_event *events,
13 int maxevents, int timeout);
14 int epoll_pwait(int epfd, struct epoll_event *events,
15 int maxevents, int timeout,
16 const sigset_t *sigmask);
17
19 The epoll_wait() system call waits for events on the epoll(7) instance
20 referred to by the file descriptor epfd. The buffer pointed to by
21 events is used to return information from the ready list about file de‐
22 scriptors in the interest list that have some events available. Up to
23 maxevents are returned by epoll_wait(). The maxevents argument must be
24 greater than zero.
25 The timeout argument specifies the number of milliseconds that
27 epoll_wait() will block. Time is measured against the CLOCK_MONOTONIC
28 clock.
29 A call to epoll_wait() will block until either:
31 • a file descriptor delivers an event;
33 • the call is interrupted by a signal handler; or
35 • the timeout expires.
37 Note that the timeout interval will be rounded up to the system clock
39 granularity, and kernel scheduling delays mean that the blocking inter‐
40 val may overrun by a small amount. Specifying a timeout of -1 causes
41 epoll_wait() to block indefinitely, while specifying a timeout equal to
42 zero cause epoll_wait() to return immediately, even if no events are
43 available.
44 The struct epoll_event is defined as:
46 typedef union epoll_data {
48 void *ptr;
49 int fd;
50 uint32_t u32;
51 uint64_t u64;
52 } epoll_data_t;
53 struct epoll_event {
55 uint32_t events; /* Epoll events */
56 epoll_data_t data; /* User data variable */
57 };
58 The data field of each returned epoll_event structure contains the same
60 data as was specified in the most recent call to epoll_ctl(2)
61 (EPOLL_CTL_ADD, EPOLL_CTL_MOD) for the corresponding open file descrip‐
62 tor.
63 The events field is a bit mask that indicates the events that have oc‐
65 curred for the corresponding open file description. See epoll_ctl(2)
66 for a list of the bits that may appear in this mask.
67 epoll_pwait()
69 The relationship between epoll_wait() and epoll_pwait() is analogous to
70 the relationship between select(2) and pselect(2): like pselect(2),
71 epoll_pwait() allows an application to safely wait until either a file
72 descriptor becomes ready or until a signal is caught.
73 The following epoll_pwait() call:
75 ready = epoll_pwait(epfd, &events, maxevents, timeout, &sigmask);
77 is equivalent to atomically executing the following calls:
79 sigset_t origmask;
81 pthread_sigmask(SIG_SETMASK, &sigmask, &origmask);
83 ready = epoll_wait(epfd, &events, maxevents, timeout);
84 pthread_sigmask(SIG_SETMASK, &origmask, NULL);
85 The sigmask argument may be specified as NULL, in which case
87 epoll_pwait() is equivalent to epoll_wait().
88
90 When successful, epoll_wait() returns the number of file descriptors
91 ready for the requested I/O, or zero if no file descriptor became ready
92 during the requested timeout milliseconds. When an error occurs,
93 epoll_wait() returns -1 and errno is set appropriately.
94
96 EBADF epfd is not a valid file descriptor.
97 EFAULT The memory area pointed to by events is not accessible with
99 write permissions.
100 EINTR The call was interrupted by a signal handler before either (1)
102 any of the requested events occurred or (2) the timeout expired;
103 see signal(7).
104 EINVAL epfd is not an epoll file descriptor, or maxevents is less than
106 or equal to zero.
107
109 epoll_wait() was added to the kernel in version 2.6. Library support
110 is provided in glibc starting with version 2.3.2.
111 epoll_pwait() was added to Linux in kernel 2.6.19. Library support is
113 provided in glibc starting with version 2.6.
114
116 epoll_wait() is Linux-specific.
117
119 While one thread is blocked in a call to epoll_wait(), it is possible
120 for another thread to add a file descriptor to the waited-upon epoll
121 instance. If the new file descriptor becomes ready, it will cause the
122 epoll_wait() call to unblock.
123 If more than maxevents file descriptors are ready when epoll_wait() is
125 called, then successive epoll_wait() calls will round robin through the
126 set of ready file descriptors. This behavior helps avoid starvation
127 scenarios, where a process fails to notice that additional file de‐
128 scriptors are ready because it focuses on a set of file descriptors
129 that are already known to be ready.
130 Note that it is possible to call epoll_wait() on an epoll instance
132 whose interest list is currently empty (or whose interest list becomes
133 empty because file descriptors are closed or removed from the interest
134 in another thread). The call will block until some file descriptor is
135 later added to the interest list (in another thread) and that file de‐
136 scriptor becomes ready.
137
139 In kernels before 2.6.37, a timeout value larger than approximately
140 LONG_MAX / HZ milliseconds is treated as -1 (i.e., infinity). Thus,
141 for example, on a system where sizeof(long) is 4 and the kernel HZ
142 value is 1000, this means that timeouts greater than 35.79 minutes are
143 treated as infinity.
144 C library/kernel differences
146 The raw epoll_pwait() system call has a sixth argument, size_t sigset‐
147 size, which specifies the size in bytes of the sigmask argument. The
148 glibc epoll_pwait() wrapper function specifies this argument as a fixed
149 value (equal to sizeof(sigset_t)).
150
152 epoll_create(2), epoll_ctl(2), epoll(7)
153
155 This page is part of release 5.10 of the Linux man-pages project. A
156 description of the project, information about reporting bugs, and the
157 latest version of this page, can be found at
158 https://www.kernel.org/doc/man-pages/.
159Linux 2020-04-11 EPOLL_WAIT(2)