ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev_epoll.c
(Generate patch)

Comparing libev/ev_epoll.c (file contents):
Revision 1.2 by root, Tue Oct 30 23:54:38 2007 UTC vs.
Revision 1.54 by sf-exg, Fri Oct 22 10:50:24 2010 UTC

1/*
2 * libev epoll fd activity backend
3 *
4 * Copyright (c) 2007,2008,2009,2010 Marc Alexander Lehmann <libev@schmorp.de>
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without modifica-
8 * tion, are permitted provided that the following conditions are met:
9 *
10 * 1. Redistributions of source code must retain the above copyright notice,
11 * this list of conditions and the following disclaimer.
12 *
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
18 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
19 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
20 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
21 * CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
22 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
23 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
24 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
25 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
26 * OF THE POSSIBILITY OF SUCH DAMAGE.
27 *
28 * Alternatively, the contents of this file may be used under the terms of
29 * the GNU General Public License ("GPL") version 2 or any later version,
30 * in which case the provisions of the GPL are applicable instead of
31 * the above. If you wish to allow the use of your version of this file
32 * only under the terms of the GPL and not to allow others to use your
33 * version of this file under the BSD license, indicate your decision
34 * by deleting the provisions above and replace them with the notice
35 * and other provisions required by the GPL. If you do not delete the
36 * provisions above, a recipient may use your version of this file under
37 * either the BSD or the GPL.
38 */
39
40/*
41 * general notes about epoll:
42 *
43 * a) epoll silently removes fds from the fd set. as nothing tells us
44 * that an fd has been removed otherwise, we have to continually
45 * "rearm" fds that we suspect *might* have changed (same
46 * problem with kqueue, but much less costly there).
47 * b) the fact that ADD != MOD creates a lot of extra syscalls due to a)
48 * and seems not to have any advantage.
49 * c) the inability to handle fork or file descriptors (think dup)
50 * limits the applicability over poll, so this is not a generic
51 * poll replacement.
52 * d) epoll doesn't work the same as select with many file descriptors
53 * (such as files). while not critical, no other advanced interface
54 * seems to share this (rather non-unixy) limitation.
55 * e) epoll claims to be embeddable, but in practise you never get
56 * a ready event for the epoll fd.
57 *
58 * lots of "weird code" and complication handling in this file is due
59 * to these design problems with epoll, as we try very hard to avoid
60 * epoll_ctl syscalls for common usage patterns and handle the breakage
61 * ensuing from receiving events for closed and otherwise long gone
62 * file descriptors.
63 */
64
1#include <sys/epoll.h> 65#include <sys/epoll.h>
2 66
3static int epoll_fd = -1;
4
5static void 67static void
6epoll_modify (int fd, int oev, int nev) 68epoll_modify (EV_P_ int fd, int oev, int nev)
7{ 69{
8 int mode = nev ? oev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD : EPOLL_CTL_DEL;
9
10 struct epoll_event ev; 70 struct epoll_event ev;
11 ev.data.fd = fd; 71 unsigned char oldmask;
12 ev.events = 72
73 /*
74 * we handle EPOLL_CTL_DEL by ignoring it here
75 * on the assumption that the fd is gone anyways
76 * if that is wrong, we have to handle the spurious
77 * event in epoll_poll.
78 * if the fd is added again, we try to ADD it, and, if that
79 * fails, we assume it still has the same eventmask.
80 */
81 if (!nev)
82 return;
83
84 oldmask = anfds [fd].emask;
85 anfds [fd].emask = nev;
86
87 /* store the generation counter in the upper 32 bits, the fd in the lower 32 bits */
88 ev.data.u64 = (uint64_t)(uint32_t)fd
89 | ((uint64_t)(uint32_t)++anfds [fd].egen << 32);
13 (nev & EV_READ ? EPOLLIN : 0) 90 ev.events = (nev & EV_READ ? EPOLLIN : 0)
14 | (nev & EV_WRITE ? EPOLLOUT : 0); 91 | (nev & EV_WRITE ? EPOLLOUT : 0);
15 92
16 fprintf (stderr, "reify %d,%d,%d m%d (r=%d)\n", fd, oev, nev, mode,//D 93 if (expect_true (!epoll_ctl (backend_fd, oev && oldmask != nev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD, fd, &ev)))
17 epoll_ctl (epoll_fd, mode, fd, &ev) 94 return;
18 );//D
19}
20 95
21void epoll_postfork_child (void) 96 if (expect_true (errno == ENOENT))
22{ 97 {
23 int fd; 98 /* if ENOENT then the fd went away, so try to do the right thing */
99 if (!nev)
100 goto dec_egen;
24 101
25 epoll_fd = epoll_create (256); 102 if (!epoll_ctl (backend_fd, EPOLL_CTL_ADD, fd, &ev))
103 return;
104 }
105 else if (expect_true (errno == EEXIST))
106 {
107 /* EEXIST means we ignored a previous DEL, but the fd is still active */
108 /* if the kernel mask is the same as the new mask, we assume it hasn't changed */
109 if (oldmask == nev)
110 goto dec_egen;
26 111
27 /* re-register interest in fds */ 112 if (!epoll_ctl (backend_fd, EPOLL_CTL_MOD, fd, &ev))
28 for (fd = 0; fd < anfdmax; ++fd) 113 return;
29 if (anfds [fd].wev) 114 }
30 epoll_modify (fd, EV_NONE, anfds [fd].wev);
31}
32 115
33static struct epoll_event *events; 116 fd_kill (EV_A_ fd);
34static int eventmax;
35 117
118dec_egen:
119 /* we didn't successfully call epoll_ctl, so decrement the generation counter again */
120 --anfds [fd].egen;
121}
122
123static void
36static void epoll_poll (ev_tstamp timeout) 124epoll_poll (EV_P_ ev_tstamp timeout)
37{ 125{
38 int eventcnt = epoll_wait (epoll_fd, events, eventmax, ceil (timeout * 1000.));
39 int i; 126 int i;
127 int eventcnt;
128
129 /* epoll wait times cannot be larger than (LONG_MAX - 999UL) / HZ msecs, which is below */
130 /* the default libev max wait time, however. */
131 EV_RELEASE_CB;
132 eventcnt = epoll_wait (backend_fd, epoll_events, epoll_eventmax, (int)ceil (timeout * 1000.));
133 EV_ACQUIRE_CB;
40 134
41 if (eventcnt < 0) 135 if (expect_false (eventcnt < 0))
136 {
137 if (errno != EINTR)
138 ev_syserr ("(libev) epoll_wait");
139
42 return; 140 return;
141 }
43 142
44 for (i = 0; i < eventcnt; ++i) 143 for (i = 0; i < eventcnt; ++i)
45 fd_event ( 144 {
46 events [i].data.fd, 145 struct epoll_event *ev = epoll_events + i;
146
147 int fd = (uint32_t)ev->data.u64; /* mask out the lower 32 bits */
148 int want = anfds [fd].events;
47 (events [i].events & (EPOLLOUT | EPOLLERR | EPOLLHUP) ? EV_WRITE : 0) 149 int got = (ev->events & (EPOLLOUT | EPOLLERR | EPOLLHUP) ? EV_WRITE : 0)
48 | (events [i].events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0) 150 | (ev->events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0);
49 ); 151
152 /* check for spurious notification */
153 /* we assume that fd is always in range, as we never shrink the anfds array */
154 if (expect_false ((uint32_t)anfds [fd].egen != (uint32_t)(ev->data.u64 >> 32)))
155 {
156 /* recreate kernel state */
157 postfork = 1;
158 continue;
159 }
160
161 if (expect_false (got & ~want))
162 {
163 anfds [fd].emask = want;
164
165 /* we received an event but are not interested in it, try mod or del */
166 /* I don't think we ever need MOD, but let's handle it anyways */
167 ev->events = (want & EV_READ ? EPOLLIN : 0)
168 | (want & EV_WRITE ? EPOLLOUT : 0);
169
170 /* pre-2.6.9 kernels require a non-null pointer with EPOLL_CTL_DEL, */
171 /* which is fortunately easy to do for us. */
172 if (epoll_ctl (backend_fd, want ? EPOLL_CTL_MOD : EPOLL_CTL_DEL, fd, ev))
173 {
174 postfork = 1; /* an error occurred, recreate kernel state */
175 continue;
176 }
177 }
178
179 fd_event (EV_A_ fd, got);
180 }
50 181
51 /* if the receive array was full, increase its size */ 182 /* if the receive array was full, increase its size */
52 if (eventcnt == eventmax) 183 if (expect_false (eventcnt == epoll_eventmax))
53 { 184 {
54 free (events); 185 ev_free (epoll_events);
55 eventmax += eventmax >> 1; 186 epoll_eventmax = array_nextsize (sizeof (struct epoll_event), epoll_eventmax, epoll_eventmax + 1);
56 events = malloc (sizeof (struct epoll_event) * eventmax); 187 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
57 } 188 }
58} 189}
59 190
191int inline_size
60int epoll_init (int flags) 192epoll_init (EV_P_ int flags)
61{ 193{
194#ifdef EPOLL_CLOEXEC
195 backend_fd = epoll_create1 (EPOLL_CLOEXEC);
196
197 if (backend_fd <= 0)
198#endif
62 epoll_fd = epoll_create (256); 199 backend_fd = epoll_create (256);
63 200
64 if (epoll_fd < 0) 201 if (backend_fd < 0)
65 return 0; 202 return 0;
66 203
67 ev_method = EVMETHOD_EPOLL; 204 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
68 method_fudge = 1e-3; /* needed to compensate for epoll returning early */ 205
206 backend_fudge = 0.; /* kernel sources seem to indicate this to be zero */
69 method_modify = epoll_modify; 207 backend_modify = epoll_modify;
70 method_poll = epoll_poll; 208 backend_poll = epoll_poll;
71 209
72 eventmax = 64; /* intiial number of events receivable per poll */ 210 epoll_eventmax = 64; /* initial number of events receivable per poll */
73 events = malloc (sizeof (struct epoll_event) * eventmax); 211 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
74 212
75 return 1; 213 return EVBACKEND_EPOLL;
76} 214}
215
216void inline_size
217epoll_destroy (EV_P)
218{
219 ev_free (epoll_events);
220}
221
222void inline_size
223epoll_fork (EV_P)
224{
225 close (backend_fd);
226
227 while ((backend_fd = epoll_create (256)) < 0)
228 ev_syserr ("(libev) epoll_create");
229
230 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
231
232 fd_rearm_all (EV_A);
233}
234

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines