1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
#![allow(deprecated)]
use std::os::unix::io::AsRawFd;
use std::os::unix::io::RawFd;
use std::sync::atomic::{AtomicUsize, Ordering, ATOMIC_USIZE_INIT};
use std::time::Duration;
use std::{cmp, i32};

use libc::{self, c_int};
use libc::{EPOLLERR, EPOLLHUP, EPOLLONESHOT};
use libc::{EPOLLET, EPOLLOUT, EPOLLIN, EPOLLPRI};

use {io, Ready, PollOpt, Token};
use event_imp::Event;
use sys::unix::{cvt, UnixReady};
use sys::unix::io::set_cloexec;

/// Each Selector has a globally unique(ish) ID associated with it. This ID
/// gets tracked by `TcpStream`, `TcpListener`, etc... when they are first
/// registered with the `Selector`. If a type that is previously associated with
/// a `Selector` attempts to register itself with a different `Selector`, the
/// operation will return with an error. This matches windows behavior.
static NEXT_ID: AtomicUsize = ATOMIC_USIZE_INIT;

#[derive(Debug)]
pub struct Selector {
    id: usize,
    epfd: RawFd,
}

impl Selector {
    pub fn new() -> io::Result<Selector> {
        let epfd = unsafe {
            // Emulate `epoll_create` by using `epoll_create1` if it's available
            // and otherwise falling back to `epoll_create` followed by a call to
            // set the CLOEXEC flag.
            dlsym!(fn epoll_create1(c_int) -> c_int);

            match epoll_create1.get() {
                Some(epoll_create1_fn) => {
                    cvt(epoll_create1_fn(libc::EPOLL_CLOEXEC))?
                }
                None => {
                    let fd = cvt(libc::epoll_create(1024))?;
                    drop(set_cloexec(fd));
                    fd
                }
            }
        };

        // offset by 1 to avoid choosing 0 as the id of a selector
        let id = NEXT_ID.fetch_add(1, Ordering::Relaxed) + 1;

        Ok(Selector {
            id: id,
            epfd: epfd,
        })
    }

    pub fn id(&self) -> usize {
        self.id
    }

    /// Wait for events from the OS
    pub fn select(&self, evts: &mut Events, awakener: Token, timeout: Option<Duration>) -> io::Result<bool> {
        // A bug in kernels < 2.6.37 makes timeouts larger than LONG_MAX / CONFIG_HZ
        // (approx. 30 minutes with CONFIG_HZ=1200) effectively infinite on 32 bits
        // architectures. The magic number is the same constant used by libuv.
        #[cfg(target_pointer_width = "32")]
        const MAX_SAFE_TIMEOUT: u64 = 1789569;
        #[cfg(not(target_pointer_width = "32"))]
        const MAX_SAFE_TIMEOUT: u64 = c_int::max_value() as u64;

        let timeout_ms = timeout
            .map(|to| cmp::min(millis(to), MAX_SAFE_TIMEOUT) as c_int)
            .unwrap_or(-1);

        // Wait for epoll events for at most timeout_ms milliseconds
        evts.clear();
        unsafe {
            let cnt = cvt(libc::epoll_wait(self.epfd,
                                           evts.events.as_mut_ptr(),
                                           evts.events.capacity() as i32,
                                           timeout_ms))?;
            let cnt = cnt as usize;
            evts.events.set_len(cnt);

            for i in 0..cnt {
                if evts.events[i].u64 as usize == awakener.into() {
                    evts.events.remove(i);
                    return Ok(true);
                }
            }
        }

        Ok(false)
    }

    /// Register event interests for the given IO handle with the OS
    pub fn register(&self, fd: RawFd, token: Token, interests: Ready, opts: PollOpt) -> io::Result<()> {
        let mut info = libc::epoll_event {
            events: ioevent_to_epoll(interests, opts),
            u64: usize::from(token) as u64
        };

        unsafe {
            cvt(libc::epoll_ctl(self.epfd, libc::EPOLL_CTL_ADD, fd, &mut info))?;
            Ok(())
        }
    }

    /// Register event interests for the given IO handle with the OS
    pub fn reregister(&self, fd: RawFd, token: Token, interests: Ready, opts: PollOpt) -> io::Result<()> {
        let mut info = libc::epoll_event {
            events: ioevent_to_epoll(interests, opts),
            u64: usize::from(token) as u64
        };

        unsafe {
            cvt(libc::epoll_ctl(self.epfd, libc::EPOLL_CTL_MOD, fd, &mut info))?;
            Ok(())
        }
    }

    /// Deregister event interests for the given IO handle with the OS
    pub fn deregister(&self, fd: RawFd) -> io::Result<()> {
        // The &info argument should be ignored by the system,
        // but linux < 2.6.9 required it to be not null.
        // For compatibility, we provide a dummy EpollEvent.
        let mut info = libc::epoll_event {
            events: 0,
            u64: 0,
        };

        unsafe {
            cvt(libc::epoll_ctl(self.epfd, libc::EPOLL_CTL_DEL, fd, &mut info))?;
            Ok(())
        }
    }
}

fn ioevent_to_epoll(interest: Ready, opts: PollOpt) -> u32 {
    let mut kind = 0;

    if interest.is_readable() {
        kind |= EPOLLIN;
    }

    if interest.is_writable() {
        kind |= EPOLLOUT;
    }

    if UnixReady::from(interest).is_priority() {
        kind |= EPOLLPRI;
    }

    if opts.is_edge() {
        kind |= EPOLLET;
    }

    if opts.is_oneshot() {
        kind |= EPOLLONESHOT;
    }

    if opts.is_level() {
        kind &= !EPOLLET;
    }

    kind as u32
}

impl AsRawFd for Selector {
    fn as_raw_fd(&self) -> RawFd {
        self.epfd
    }
}

impl Drop for Selector {
    fn drop(&mut self) {
        unsafe {
            let _ = libc::close(self.epfd);
        }
    }
}

pub struct Events {
    events: Vec<libc::epoll_event>,
}

impl Events {
    pub fn with_capacity(u: usize) -> Events {
        Events {
            events: Vec::with_capacity(u)
        }
    }

    #[inline]
    pub fn len(&self) -> usize {
        self.events.len()
    }

    #[inline]
    pub fn capacity(&self) -> usize {
        self.events.capacity()
    }

    #[inline]
    pub fn is_empty(&self) -> bool {
        self.events.is_empty()
    }

    #[inline]
    pub fn get(&self, idx: usize) -> Option<Event> {
        self.events.get(idx).map(|event| {
            let epoll = event.events as c_int;
            let mut kind = Ready::empty();

            if (epoll & EPOLLIN) != 0 {
                kind = kind | Ready::readable();
            }

            if (epoll & EPOLLPRI) != 0 {
                kind = kind | Ready::readable() | UnixReady::priority();
            }

            if (epoll & EPOLLOUT) != 0 {
                kind = kind | Ready::writable();
            }

            // EPOLLHUP - Usually means a socket error happened
            if (epoll & EPOLLERR) != 0 {
                kind = kind | UnixReady::error();
            }

            if (epoll & EPOLLHUP) != 0 {
                kind = kind | UnixReady::hup();
            }

            let token = self.events[idx].u64;

            Event::new(kind, Token(token as usize))
        })
    }

    pub fn push_event(&mut self, event: Event) {
        self.events.push(libc::epoll_event {
            events: ioevent_to_epoll(event.readiness(), PollOpt::empty()),
            u64: usize::from(event.token()) as u64
        });
    }

    pub fn clear(&mut self) {
        unsafe { self.events.set_len(0); }
    }
}

const NANOS_PER_MILLI: u32 = 1_000_000;
const MILLIS_PER_SEC: u64 = 1_000;

/// Convert a `Duration` to milliseconds, rounding up and saturating at
/// `u64::MAX`.
///
/// The saturating is fine because `u64::MAX` milliseconds are still many
/// million years.
pub fn millis(duration: Duration) -> u64 {
    // Round up.
    let millis = (duration.subsec_nanos() + NANOS_PER_MILLI - 1) / NANOS_PER_MILLI;
    duration.as_secs().saturating_mul(MILLIS_PER_SEC).saturating_add(millis as u64)
}