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 1.21 by root, Tue Nov 6 00:52:33 2007 UTC vs.
Revision 1.50 by root, Tue Mar 23 23:45:12 2010 UTC

1/* 1/*
2 * libev epoll fd activity backend 2 * libev epoll fd activity backend
3 * 3 *
4 * Copyright (c) 2007 Marc Alexander Lehmann <libev@schmorp.de> 4 * Copyright (c) 2007,2008,2009,2010 Marc Alexander Lehmann <libev@schmorp.de>
5 * All rights reserved. 5 * All rights reserved.
6 * 6 *
7 * Redistribution and use in source and binary forms, with or without 7 * Redistribution and use in source and binary forms, with or without modifica-
8 * modification, are permitted provided that the following conditions are 8 * tion, are permitted provided that the following conditions are met:
9 * met:
10 * 9 *
11 * * Redistributions of source code must retain the above copyright 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
12 * notice, this list of conditions and the following disclaimer. 14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
13 * 16 *
14 * * Redistributions in binary form must reproduce the above 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
15 * copyright notice, this list of conditions and the following 18 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
16 * disclaimer in the documentation and/or other materials provided 19 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
17 * with the distribution. 20 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 21 * CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
25 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 25 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
29 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 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.
30 */ 63 */
31 64
32#include <sys/epoll.h> 65#include <sys/epoll.h>
33 66
34static void 67static void
35epoll_modify (EV_P_ int fd, int oev, int nev) 68epoll_modify (EV_P_ int fd, int oev, int nev)
36{ 69{
37 int mode = nev ? oev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD : EPOLL_CTL_DEL;
38
39 struct epoll_event ev; 70 struct epoll_event ev;
40 ev.data.u64 = fd; /* use u64 to fully initialise the struct, for nicer strace etc. */ 71 unsigned char oldmask;
41 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);
42 (nev & EV_READ ? EPOLLIN : 0) 90 ev.events = (nev & EV_READ ? EPOLLIN : 0)
43 | (nev & EV_WRITE ? EPOLLOUT : 0); 91 | (nev & EV_WRITE ? EPOLLOUT : 0);
44 92
45 if (epoll_ctl (epoll_fd, mode, fd, &ev)) 93 if (expect_true (!epoll_ctl (backend_fd, oev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD, fd, &ev)))
94 return;
95
96 if (expect_true (errno == ENOENT))
97 {
46 if (errno != ENOENT /* on ENOENT the fd went away, so try to do the right thing */ 98 /* if ENOENT then the fd went away, so try to do the right thing */
99 if (!nev)
100 goto dec_egen;
101
47 || (nev && epoll_ctl (epoll_fd, EPOLL_CTL_ADD, fd, &ev))) 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;
111
112 if (!epoll_ctl (backend_fd, EPOLL_CTL_MOD, fd, &ev))
113 return;
114 }
115
48 fd_kill (EV_A_ fd); 116 fd_kill (EV_A_ fd);
117
118dec_egen:
119 /* we didn't successfully call epoll_ctl, so decrement the generation counter again */
120 --anfds [fd].egen;
49} 121}
50 122
51static void 123static void
52epoll_poll (EV_P_ ev_tstamp timeout) 124epoll_poll (EV_P_ ev_tstamp timeout)
53{ 125{
54 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;
55 int eventcnt = epoll_wait (epoll_fd, epoll_events, epoll_eventmax, ceil (timeout * 1000.)); 132 eventcnt = epoll_wait (backend_fd, epoll_events, epoll_eventmax, (int)ceil (timeout * 1000.));
133 EV_ACQUIRE_CB;
56 134
57 if (eventcnt < 0) 135 if (expect_false (eventcnt < 0))
58 { 136 {
59 if (errno != EINTR) 137 if (errno != EINTR)
60 syserr ("(libev) epoll_wait"); 138 ev_syserr ("(libev) epoll_wait");
61 139
62 return; 140 return;
63 } 141 }
64 142
65 for (i = 0; i < eventcnt; ++i) 143 for (i = 0; i < eventcnt; ++i)
66 fd_event ( 144 {
67 EV_A_ 145 struct epoll_event *ev = epoll_events + i;
68 epoll_events [i].data.u64, 146
147 int fd = (uint32_t)ev->data.u64; /* mask out the lower 32 bits */
148 int want = anfds [fd].events;
69 (epoll_events [i].events & (EPOLLOUT | EPOLLERR | EPOLLHUP) ? EV_WRITE : 0) 149 int got = (ev->events & (EPOLLOUT | EPOLLERR | EPOLLHUP) ? EV_WRITE : 0)
70 | (epoll_events [i].events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0) 150 | (ev->events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0);
71 ); 151
152 /* check for spurious notification */
153 if (expect_false ((uint32_t)anfds [fd].egen != (uint32_t)(ev->data.u64 >> 32)))
154 {
155 /* recreate kernel state */
156 postfork = 1;
157 continue;
158 }
159
160 if (expect_false (got & ~want))
161 {
162 anfds [fd].emask = want;
163
164 /* we received an event but are not interested in it, try mod or del */
165 /* I don't think we ever need MOD, but let's handle it anyways */
166 ev->events = (want & EV_READ ? EPOLLIN : 0)
167 | (want & EV_WRITE ? EPOLLOUT : 0);
168
169 /* pre-2.6.9 kernels require a non-null pointer with EPOLL_CTL_DEL, */
170 /* which is fortunately easy to do for us. */
171 if (epoll_ctl (backend_fd, want ? EPOLL_CTL_MOD : EPOLL_CTL_DEL, fd, ev))
172 {
173 postfork = 1; /* an error occured, recreate kernel state */
174 continue;
175 }
176 }
177
178 fd_event (EV_A_ fd, got);
179 }
72 180
73 /* if the receive array was full, increase its size */ 181 /* if the receive array was full, increase its size */
74 if (expect_false (eventcnt == epoll_eventmax)) 182 if (expect_false (eventcnt == epoll_eventmax))
75 { 183 {
76 ev_free (epoll_events); 184 ev_free (epoll_events);
77 epoll_eventmax = array_roundsize (epoll_events, epoll_eventmax << 1); 185 epoll_eventmax = array_nextsize (sizeof (struct epoll_event), epoll_eventmax, epoll_eventmax + 1);
78 epoll_events = ev_malloc (sizeof (struct epoll_event) * epoll_eventmax); 186 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
79 } 187 }
80} 188}
81 189
82static int 190int inline_size
83epoll_init (EV_P_ int flags) 191epoll_init (EV_P_ int flags)
84{ 192{
193#ifdef EPOLL_CLOEXEC
194 backend_fd = epoll_create1 (EPOLL_CLOEXEC);
195
196 if (backend_fd <= 0)
197#endif
85 epoll_fd = epoll_create (256); 198 backend_fd = epoll_create (256);
86 199
87 if (epoll_fd < 0) 200 if (backend_fd < 0)
88 return 0; 201 return 0;
89 202
90 fcntl (epoll_fd, F_SETFD, FD_CLOEXEC); 203 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
91 204
92 method_fudge = 1e-3; /* needed to compensate for epoll returning early */ 205 backend_fudge = 0.; /* kernel sources seem to indicate this to be zero */
93 method_modify = epoll_modify; 206 backend_modify = epoll_modify;
94 method_poll = epoll_poll; 207 backend_poll = epoll_poll;
95 208
96 epoll_eventmax = 64; /* intiial number of events receivable per poll */ 209 epoll_eventmax = 64; /* initial number of events receivable per poll */
97 epoll_events = ev_malloc (sizeof (struct epoll_event) * epoll_eventmax); 210 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
98 211
99 return EVMETHOD_EPOLL; 212 return EVBACKEND_EPOLL;
100} 213}
101 214
102static void 215void inline_size
103epoll_destroy (EV_P) 216epoll_destroy (EV_P)
104{ 217{
105 close (epoll_fd);
106
107 ev_free (epoll_events); 218 ev_free (epoll_events);
108} 219}
109 220
110static void 221void inline_size
111epoll_fork (EV_P) 222epoll_fork (EV_P)
112{ 223{
113 for (;;) 224 close (backend_fd);
114 {
115 epoll_fd = epoll_create (256);
116 225
117 if (epoll_fd >= 0) 226 while ((backend_fd = epoll_create (256)) < 0)
118 break;
119
120 syserr ("(libev) epoll_create"); 227 ev_syserr ("(libev) epoll_create");
121 }
122 228
123 fcntl (epoll_fd, F_SETFD, FD_CLOEXEC); 229 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
124 230
125 fd_rearm_all (EV_A); 231 fd_rearm_all (EV_A);
126} 232}
127 233

Diff Legend

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