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
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
|
/*
* This file is part of the ZFS Event Daemon (ZED).
*
* Developed at Lawrence Livermore National Laboratory (LLNL-CODE-403049).
* Copyright (C) 2013-2014 Lawrence Livermore National Security, LLC.
* Refer to the ZoL git commit log for authoritative copyright attribution.
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License Version 1.0 (CDDL-1.0).
* You can obtain a copy of the license from the top-level file
* "OPENSOLARIS.LICENSE" or at <http://opensource.org/licenses/CDDL-1.0>.
* You may not use this file except in compliance with the license.
*/
#include <assert.h>
#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
#include <stdlib.h>
#include <string.h>
#include <stddef.h>
#include <sys/avl.h>
#include <sys/resource.h>
#include <sys/stat.h>
#include <sys/wait.h>
#include <time.h>
#include <unistd.h>
#include <pthread.h>
#include "zed_exec.h"
#include "zed_log.h"
#include "zed_strings.h"
#define ZEVENT_FILENO 3
struct launched_process_node {
avl_node_t node;
pid_t pid;
uint64_t eid;
char *name;
};
static int
_launched_process_node_compare(const void *x1, const void *x2)
{
pid_t p1;
pid_t p2;
assert(x1 != NULL);
assert(x2 != NULL);
p1 = ((const struct launched_process_node *) x1)->pid;
p2 = ((const struct launched_process_node *) x2)->pid;
if (p1 < p2)
return (-1);
else if (p1 == p2)
return (0);
else
return (1);
}
static pthread_t _reap_children_tid = (pthread_t)-1;
static volatile boolean_t _reap_children_stop;
static avl_tree_t _launched_processes;
static pthread_mutex_t _launched_processes_lock = PTHREAD_MUTEX_INITIALIZER;
static int16_t _launched_processes_limit;
/*
* Create an environment string array for passing to execve() using the
* NAME=VALUE strings in container [zsp].
* Return a newly-allocated environment, or NULL on error.
*/
static char **
_zed_exec_create_env(zed_strings_t *zsp)
{
int num_ptrs;
int buflen;
char *buf;
char **pp;
char *p;
const char *q;
int i;
int len;
num_ptrs = zed_strings_count(zsp) + 1;
buflen = num_ptrs * sizeof (char *);
for (q = zed_strings_first(zsp); q; q = zed_strings_next(zsp))
buflen += strlen(q) + 1;
buf = calloc(1, buflen);
if (!buf)
return (NULL);
pp = (char **)buf;
p = buf + (num_ptrs * sizeof (char *));
i = 0;
for (q = zed_strings_first(zsp); q; q = zed_strings_next(zsp)) {
pp[i] = p;
len = strlen(q) + 1;
memcpy(p, q, len);
p += len;
i++;
}
pp[i] = NULL;
assert(buf + buflen == p);
return ((char **)buf);
}
/*
* Fork a child process to handle event [eid]. The program [prog]
* in directory [dir] is executed with the environment [env].
*
* The file descriptor [zfd] is the zevent_fd used to track the
* current cursor location within the zevent nvlist.
*/
static void
_zed_exec_fork_child(uint64_t eid, const char *dir, const char *prog,
char *env[], int zfd, boolean_t in_foreground)
{
char path[PATH_MAX];
int n;
pid_t pid;
int fd;
struct launched_process_node *node;
sigset_t mask;
struct timespec launch_timeout =
{ .tv_sec = 0, .tv_nsec = 200 * 1000 * 1000, };
assert(dir != NULL);
assert(prog != NULL);
assert(env != NULL);
assert(zfd >= 0);
while (__atomic_load_n(&_launched_processes_limit,
__ATOMIC_SEQ_CST) <= 0)
(void) nanosleep(&launch_timeout, NULL);
n = snprintf(path, sizeof (path), "%s/%s", dir, prog);
if ((n < 0) || (n >= sizeof (path))) {
zed_log_msg(LOG_WARNING,
"Failed to fork \"%s\" for eid=%llu: %s",
prog, eid, strerror(ENAMETOOLONG));
return;
}
pid = fork();
if (pid < 0) {
zed_log_msg(LOG_WARNING,
"Failed to fork \"%s\" for eid=%llu: %s",
prog, eid, strerror(errno));
return;
} else if (pid == 0) {
(void) sigemptyset(&mask);
(void) sigprocmask(SIG_SETMASK, &mask, NULL);
(void) umask(022);
if (in_foreground && /* we're already devnulled if daemonised */
(fd = open("/dev/null", O_RDWR | O_CLOEXEC)) != -1) {
(void) dup2(fd, STDIN_FILENO);
(void) dup2(fd, STDOUT_FILENO);
(void) dup2(fd, STDERR_FILENO);
}
(void) dup2(zfd, ZEVENT_FILENO);
execle(path, prog, NULL, env);
_exit(127);
}
/* parent process */
__atomic_sub_fetch(&_launched_processes_limit, 1, __ATOMIC_SEQ_CST);
zed_log_msg(LOG_INFO, "Invoking \"%s\" eid=%llu pid=%d",
prog, eid, pid);
node = calloc(1, sizeof (*node));
if (node) {
node->pid = pid;
node->eid = eid;
node->name = strdup(prog);
(void) pthread_mutex_lock(&_launched_processes_lock);
avl_add(&_launched_processes, node);
(void) pthread_mutex_unlock(&_launched_processes_lock);
}
}
static void
_nop(int sig)
{}
static void *
_reap_children(void *arg)
{
struct launched_process_node node, *pnode;
pid_t pid;
int status;
struct rusage usage;
struct sigaction sa = {};
(void) sigfillset(&sa.sa_mask);
(void) sigdelset(&sa.sa_mask, SIGCHLD);
(void) pthread_sigmask(SIG_SETMASK, &sa.sa_mask, NULL);
(void) sigemptyset(&sa.sa_mask);
sa.sa_handler = _nop;
sa.sa_flags = SA_NOCLDSTOP;
(void) sigaction(SIGCHLD, &sa, NULL);
for (_reap_children_stop = B_FALSE; !_reap_children_stop; ) {
pid = wait4(0, &status, 0, &usage);
if (pid == (pid_t)-1) {
if (errno == ECHILD)
pause();
else if (errno != EINTR)
zed_log_msg(LOG_WARNING,
"Failed to wait for children: %s",
strerror(errno));
} else {
memset(&node, 0, sizeof (node));
node.pid = pid;
(void) pthread_mutex_lock(&_launched_processes_lock);
pnode = avl_find(&_launched_processes, &node, NULL);
if (pnode) {
memcpy(&node, pnode, sizeof (node));
avl_remove(&_launched_processes, pnode);
free(pnode);
}
(void) pthread_mutex_unlock(&_launched_processes_lock);
__atomic_add_fetch(&_launched_processes_limit, 1,
__ATOMIC_SEQ_CST);
usage.ru_utime.tv_sec += usage.ru_stime.tv_sec;
usage.ru_utime.tv_usec += usage.ru_stime.tv_usec;
usage.ru_utime.tv_sec +=
usage.ru_utime.tv_usec / (1000 * 1000);
usage.ru_utime.tv_usec %= 1000 * 1000;
if (WIFEXITED(status)) {
zed_log_msg(LOG_INFO,
"Finished \"%s\" eid=%llu pid=%d "
"time=%llu.%06us exit=%d",
node.name, node.eid, pid,
(unsigned long long) usage.ru_utime.tv_sec,
(unsigned int) usage.ru_utime.tv_usec,
WEXITSTATUS(status));
} else if (WIFSIGNALED(status)) {
zed_log_msg(LOG_INFO,
"Finished \"%s\" eid=%llu pid=%d "
"time=%llu.%06us sig=%d/%s",
node.name, node.eid, pid,
(unsigned long long) usage.ru_utime.tv_sec,
(unsigned int) usage.ru_utime.tv_usec,
WTERMSIG(status),
strsignal(WTERMSIG(status)));
} else {
zed_log_msg(LOG_INFO,
"Finished \"%s\" eid=%llu pid=%d "
"time=%llu.%06us status=0x%X",
node.name, node.eid,
(unsigned long long) usage.ru_utime.tv_sec,
(unsigned int) usage.ru_utime.tv_usec,
(unsigned int) status);
}
free(node.name);
}
}
return (NULL);
}
void
zed_exec_fini(void)
{
struct launched_process_node *node;
void *ck = NULL;
if (_reap_children_tid == (pthread_t)-1)
return;
_reap_children_stop = B_TRUE;
(void) pthread_kill(_reap_children_tid, SIGCHLD);
(void) pthread_join(_reap_children_tid, NULL);
while ((node = avl_destroy_nodes(&_launched_processes, &ck)) != NULL) {
free(node->name);
free(node);
}
avl_destroy(&_launched_processes);
(void) pthread_mutex_destroy(&_launched_processes_lock);
(void) pthread_mutex_init(&_launched_processes_lock, NULL);
_reap_children_tid = (pthread_t)-1;
}
/*
* Process the event [eid] by synchronously invoking all zedlets with a
* matching class prefix.
*
* Each executable in [zcp->zedlets] from the directory [zcp->zedlet_dir]
* is matched against the event's [class], [subclass], and the "all" class
* (which matches all events).
* Every zedlet with a matching class prefix is invoked.
* The NAME=VALUE strings in [envs] will be passed to the zedlet as
* environment variables.
*
* The file descriptor [zcp->zevent_fd] is the zevent_fd used to track the
* current cursor location within the zevent nvlist.
*
* Return 0 on success, -1 on error.
*/
int
zed_exec_process(uint64_t eid, const char *class, const char *subclass,
struct zed_conf *zcp, zed_strings_t *envs)
{
const char *class_strings[4];
const char *allclass = "all";
const char **csp;
const char *z;
char **e;
int n;
if (!zcp->zedlet_dir || !zcp->zedlets || !envs || zcp->zevent_fd < 0)
return (-1);
if (_reap_children_tid == (pthread_t)-1) {
_launched_processes_limit = zcp->max_jobs;
if (pthread_create(&_reap_children_tid, NULL,
_reap_children, NULL) != 0)
return (-1);
pthread_setname_np(_reap_children_tid, "reap ZEDLETs");
avl_create(&_launched_processes, _launched_process_node_compare,
sizeof (struct launched_process_node),
offsetof(struct launched_process_node, node));
}
csp = class_strings;
if (class)
*csp++ = class;
if (subclass)
*csp++ = subclass;
if (allclass)
*csp++ = allclass;
*csp = NULL;
e = _zed_exec_create_env(envs);
for (z = zed_strings_first(zcp->zedlets); z;
z = zed_strings_next(zcp->zedlets)) {
for (csp = class_strings; *csp; csp++) {
n = strlen(*csp);
if ((strncmp(z, *csp, n) == 0) && !isalpha(z[n]))
_zed_exec_fork_child(eid, zcp->zedlet_dir,
z, e, zcp->zevent_fd, zcp->do_foreground);
}
}
free(e);
return (0);
}
|