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 1.1 by root, Tue Oct 30 20:59:31 2007 UTC vs.
Revision 1.43 by root, Fri Nov 7 15:07:50 2008 UTC

1/*
2 * libev epoll fd activity backend
3 *
4 * Copyright (c) 2007,2008 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 *
53 * lots of "weird code" and complication handling in this file is due
54 * to these design problems with epoll, as we try very hard to avoid
55 * epoll_ctl syscalls for common usage patterns and handle the breakage
56 * ensuing from receiving events for closed and otherwise long gone
57 * file descriptors.
58 */
59
1#include <sys/epoll.h> 60#include <sys/epoll.h>
2 61
3static int epoll_fd = -1;
4
5static void 62static void
6epoll_reify_fd (int fd) 63epoll_modify (EV_P_ int fd, int oev, int nev)
7{ 64{
8 ANFD *anfd = anfds + fd;
9 struct ev_io *w;
10
11 int wev = 0;
12
13 for (w = anfd->head; w; w = w->next)
14 wev |= w->events;
15
16 if (anfd->wev != wev)
17 {
18 int mode = wev ? anfd->wev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD : EPOLL_CTL_DEL;
19 struct epoll_event ev; 65 struct epoll_event ev;
20 ev.events = wev; 66 unsigned char oldmask;
21 ev.data.fd = fd; 67
22 fprintf (stderr, "reify %d,%d,%d m%d (r=%d)\n", fd, anfd->wev, wev, mode,//D 68 /*
23 epoll_ctl (epoll_fd, mode, fd, &ev) 69 * we handle EPOLL_CTL_DEL by ignoring it here
24 );//D 70 * on the assumption that the fd is gone anyways
25 anfd->wev = wev; 71 * if that is wrong, we have to handle the spurious
72 * event in epoll_poll.
73 * if the fd is added again, we try to ADD it, and, if that
74 * fails, we assume it still has the same eventmask.
75 */
76 if (!nev)
77 return;
78
79 oldmask = anfds [fd].emask;
80 anfds [fd].emask = nev;
81
82 /* store the generation counter in the upper 32 bits, the fd in the lower 32 bits */
83 ev.data.u64 = (uint64_t)(uint32_t)fd
84 | ((uint64_t)(uint32_t)++anfds [fd].egen << 32);
85 ev.events = (nev & EV_READ ? EPOLLIN : 0)
86 | (nev & EV_WRITE ? EPOLLOUT : 0);
87
88 if (expect_true (!epoll_ctl (backend_fd, oev ? EPOLL_CTL_MOD : EPOLL_CTL_ADD, fd, &ev)))
89 return;
90
91 if (expect_true (errno == ENOENT))
26 } 92 {
27} 93 /* if ENOENT then the fd went away, so try to do the right thing */
94 if (!nev)
95 goto dec_egen;
28 96
29void epoll_postfork_child (void) 97 if (!epoll_ctl (backend_fd, EPOLL_CTL_ADD, fd, &ev))
98 return;
99 }
100 else if (expect_true (errno == EEXIST))
101 {
102 /* EEXIST means we ignored a previous DEL, but the fd is still active */
103 /* if the kernel mask is the same as the new mask, we assume it hasn't changed */
104 if (oldmask == nev)
105 goto dec_egen;
106
107 if (!epoll_ctl (backend_fd, EPOLL_CTL_MOD, fd, &ev))
108 return;
109 }
110
111 fd_kill (EV_A_ fd);
112
113dec_egen:
114 /* we didn't successfully call epoll_ctl, so decrement the generation counter again */
115 --anfds [fd].egen;
116}
117
118static void
119epoll_poll (EV_P_ ev_tstamp timeout)
30{ 120{
31 int i; 121 int i;
32
33 epoll_fd = epoll_create (256);
34
35 for (i = 0; i < anfdmax; ++i)
36 epoll_reify_fd (i);
37}
38
39static void epoll_reify (void)
40{
41 int i;
42 for (i = 0; i < fdchangecnt; ++i)
43 epoll_reify_fd (fdchanges [i]);
44}
45
46static struct epoll_event *events;
47static int eventmax;
48
49static void epoll_poll (ev_tstamp timeout)
50{
51 int eventcnt = epoll_wait (epoll_fd, events, eventmax, ceil (timeout * 1000.)); 122 int eventcnt = epoll_wait (backend_fd, epoll_events, epoll_eventmax, (int)ceil (timeout * 1000.));
52 int i;
53 123
54 if (eventcnt < 0) 124 if (expect_false (eventcnt < 0))
125 {
126 if (errno != EINTR)
127 ev_syserr ("(libev) epoll_wait");
128
55 return; 129 return;
130 }
56 131
57 for (i = 0; i < eventcnt; ++i) 132 for (i = 0; i < eventcnt; ++i)
58 fd_event ( 133 {
59 events [i].data.fd, 134 struct epoll_event *ev = epoll_events + i;
135
136 int fd = (uint32_t)ev->data.u64; /* mask out the lower 32 bits */
137 int want = anfds [fd].events;
60 (events [i].events & (EPOLLOUT | EPOLLERR | EPOLLHUP | EPOLLPRI) ? EV_WRITE : 0) 138 int got = (ev->events & (EPOLLOUT | EPOLLERR | EPOLLHUP) ? EV_WRITE : 0)
61 | (events [i].events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0) 139 | (ev->events & (EPOLLIN | EPOLLERR | EPOLLHUP) ? EV_READ : 0);
62 ); 140
141 /* check for spurious notification */
142 if (expect_false ((uint32_t)anfds [fd].egen != (uint32_t)(ev->data.u64 >> 32)))
143 {
144 /* recreate kernel state */
145 postfork = 1;
146 continue;
147 }
148
149 if (expect_false (got & ~want))
150 {
151 anfds [fd].emask = want;
152
153 /* we received an event but are not interested in it, try mod or del */
154 /* I don't think we ever need MOD, but let's handle it anyways */
155 ev->events = (want & EV_READ ? EPOLLIN : 0)
156 | (want & EV_WRITE ? EPOLLOUT : 0);
157
158 if (epoll_ctl (backend_fd, want ? EPOLL_CTL_MOD : EPOLL_CTL_DEL, fd, ev))
159 {
160 postfork = 1; /* an error occured, recreate kernel state */
161 continue;
162 }
163 }
164
165 fd_event (EV_A_ fd, got);
166 }
63 167
64 /* if the receive array was full, increase its size */ 168 /* if the receive array was full, increase its size */
65 if (eventcnt == eventmax) 169 if (expect_false (eventcnt == epoll_eventmax))
66 { 170 {
67 free (events); 171 ev_free (epoll_events);
68 eventmax += eventmax >> 1; 172 epoll_eventmax = array_nextsize (sizeof (struct epoll_event), epoll_eventmax, epoll_eventmax + 1);
69 events = malloc (sizeof (struct epoll_event) * eventmax); 173 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
70 } 174 }
71} 175}
72 176
177int inline_size
73int epoll_init (int flags) 178epoll_init (EV_P_ int flags)
74{ 179{
75 epoll_fd = epoll_create (256); 180 backend_fd = epoll_create (256);
76 181
77 if (epoll_fd < 0) 182 if (backend_fd < 0)
78 return 0; 183 return 0;
79 184
80 ev_method = EVMETHOD_EPOLL; 185 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
81 method_fudge = 1e-3; /* needed to compensate fro epoll returning early */ 186
82 method_reify = epoll_reify; 187 backend_fudge = 0.; /* kernel sources seem to indicate this to be zero */
188 backend_modify = epoll_modify;
83 method_poll = epoll_poll; 189 backend_poll = epoll_poll;
84 190
85 eventmax = 64; /* intiial number of events receivable per poll */ 191 epoll_eventmax = 64; /* initial number of events receivable per poll */
86 events = malloc (sizeof (struct epoll_event) * eventmax); 192 epoll_events = (struct epoll_event *)ev_malloc (sizeof (struct epoll_event) * epoll_eventmax);
87 193
88 return 1; 194 return EVBACKEND_EPOLL;
89} 195}
196
197void inline_size
198epoll_destroy (EV_P)
199{
200 ev_free (epoll_events);
201}
202
203void inline_size
204epoll_fork (EV_P)
205{
206 close (backend_fd);
207
208 while ((backend_fd = epoll_create (256)) < 0)
209 ev_syserr ("(libev) epoll_create");
210
211 fcntl (backend_fd, F_SETFD, FD_CLOEXEC);
212
213 fd_rearm_all (EV_A);
214}
215

Diff Legend

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