summaryrefslogtreecommitdiffstats
path: root/mdmon.c
blob: cef5bbc8b0dd37d6e6fdfa42fa94ed41d2d19467 (plain)
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
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
/*
 * mdmon - monitor external metadata arrays
 *
 * Copyright (C) 2007-2009 Neil Brown <neilb@suse.de>
 * Copyright (C) 2007-2009 Intel Corporation
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 */

/*
 * md array manager.
 * When md arrays have user-space managed metadata, this is the program
 * that does the managing.
 *
 * Given one argument: the name of the array (e.g. /dev/md0) that is
 * the container.
 * We fork off a helper that runs high priority and mlocked.  It responds to
 * device failures and other events that might stop writeout, or that are
 * trivial to deal with.
 * The main thread then watches for new arrays being created in the container
 * and starts monitoring them too ... along with a few other tasks.
 *
 * The main thread communicates with the priority thread by writing over
 * a pipe.
 * Separate programs can communicate with the main thread via Unix-domain
 * socket.
 * The two threads share address space and open file table.
 *
 */

#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif

#include	<unistd.h>
#include	<stdlib.h>
#include	<sys/types.h>
#include	<sys/stat.h>
#include	<sys/socket.h>
#include	<sys/un.h>
#include	<sys/mman.h>
#include	<sys/syscall.h>
#include	<sys/wait.h>
#include	<stdio.h>
#include	<errno.h>
#include	<string.h>
#include	<fcntl.h>
#include	<dirent.h>
#ifdef USE_PTHREADS
#include	<pthread.h>
#else
#include	<sched.h>
#endif

#include	"mdadm.h"
#include	"mdmon.h"

char const Name[] = "mdmon";

struct active_array *discard_this;
struct active_array *pending_discard;

int mon_tid, mgr_tid;

int sigterm;

#ifdef USE_PTHREADS
static void *run_child(void *v)
{
	struct supertype *c = v;

	mon_tid = syscall(SYS_gettid);
	do_monitor(c);
	return 0;
}

static int clone_monitor(struct supertype *container)
{
	pthread_attr_t attr;
	pthread_t thread;
	int rc;

	mon_tid = -1;
	pthread_attr_init(&attr);
	pthread_attr_setstacksize(&attr, 4096);
	pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
	rc = pthread_create(&thread, &attr, run_child, container);
	if (rc)
		return rc;
	while (mon_tid == -1)
		sleep_for(0, USEC_TO_NSEC(10), true);
	pthread_attr_destroy(&attr);

	mgr_tid = syscall(SYS_gettid);

	return mon_tid;
}
#else /* USE_PTHREADS */
static int run_child(void *v)
{
	struct supertype *c = v;

	do_monitor(c);
	return 0;
}

#ifdef __ia64__
int __clone2(int (*fn)(void *),
	    void *child_stack_base, size_t stack_size,
	    int flags, void *arg, ...
	 /* pid_t *pid, struct user_desc *tls, pid_t *ctid */ );
#endif
static int clone_monitor(struct supertype *container)
{
	static char stack[4096];

#ifdef __ia64__
	mon_tid = __clone2(run_child, stack, sizeof(stack),
		   CLONE_FS|CLONE_FILES|CLONE_VM|CLONE_SIGHAND|CLONE_THREAD,
		   container);
#else
	mon_tid = clone(run_child, stack+4096-64,
		   CLONE_FS|CLONE_FILES|CLONE_VM|CLONE_SIGHAND|CLONE_THREAD,
		   container);
#endif

	mgr_tid = syscall(SYS_gettid);

	return mon_tid;
}
#endif /* USE_PTHREADS */

static int make_pidfile(char *devname)
{
	char path[100];
	char pid[10];
	int fd;
	int n;

	if (mkdir(MDMON_DIR, 0755) < 0 &&
	    errno != EEXIST)
		return -errno;
	sprintf(path, "%s/%s.pid", MDMON_DIR, devname);

	fd = open(path, O_RDWR|O_CREAT|O_EXCL, 0600);
	if (fd < 0)
		return -errno;
	sprintf(pid, "%d\n", getpid());
	n = write(fd, pid, strlen(pid));
	close(fd);
	if (n < 0)
		return -errno;
	return 0;
}

static void try_kill_monitor(pid_t pid, char *devname, int sock)
{
	char buf[100];
	int fd;
	int n;
	long fl;
	int rv;

	/* first rule of survival... don't off yourself */
	if (pid == getpid())
		return;

	/* kill this process if it is mdmon */
	sprintf(buf, "/proc/%lu/cmdline", (unsigned long) pid);
	fd = open(buf, O_RDONLY);
	if (fd < 0)
		return;

	n = read(fd, buf, sizeof(buf)-1);
	buf[sizeof(buf)-1] = 0;
	close(fd);

	if (n < 0 || !(strstr(buf, "mdmon") ||
		       strstr(buf, "@dmon")))
		return;

	kill(pid, SIGTERM);

	if (sock < 0)
		return;

	/* Wait for monitor to exit by reading from the socket, after
	 * clearing the non-blocking flag */
	fl = fcntl(sock, F_GETFL, 0);
	fl &= ~O_NONBLOCK;
	fcntl(sock, F_SETFL, fl);
	n = read(sock, buf, 100);

	/* If there is I/O going on it might took some time to get to
	 * clean state. Wait for monitor to exit fully to avoid races.
	 * Ping it with SIGUSR1 in case that it is sleeping  */
	for (n = 0; n < 25; n++) {
		rv = kill(pid, SIGUSR1);
		if (rv < 0)
			break;
		sleep_for(0, MSEC_TO_NSEC(200), true);
	}
}

void remove_pidfile(char *devname)
{
	char buf[100];

	sprintf(buf, "%s/%s.pid", MDMON_DIR, devname);
	unlink(buf);
	sprintf(buf, "%s/%s.sock", MDMON_DIR, devname);
	unlink(buf);
}

static int make_control_sock(char *devname)
{
	char path[100];
	int sfd;
	long fl;
	struct sockaddr_un addr;

	if (sigterm)
		return -1;

	sprintf(path, "%s/%s.sock", MDMON_DIR, devname);
	unlink(path);
	sfd = socket(PF_LOCAL, SOCK_STREAM, 0);
	if (sfd < 0)
		return -1;

	addr.sun_family = PF_LOCAL;
	strcpy(addr.sun_path, path);
	umask(077); /* ensure no world write access */
	if (bind(sfd, (struct sockaddr*)&addr, sizeof(addr)) < 0) {
		close(sfd);
		return -1;
	}
	listen(sfd, 10);
	fl = fcntl(sfd, F_GETFL, 0);
	fl |= O_NONBLOCK;
	fcntl(sfd, F_SETFL, fl);
	return sfd;
}

static void term(int sig)
{
	sigterm = 1;
}

static void wake_me(int sig)
{

}

/* if we are debugging and starting mdmon by hand then don't fork */
static int do_fork(void)
{
	#ifdef DEBUG
	if (check_env("MDADM_NO_MDMON"))
		return 0;
	#endif

	return 1;
}

void usage(void)
{
	fprintf(stderr,
"Usage: mdmon [options] CONTAINER\n"
"\n"
"Options are:\n"
"  --help        -h   : This message\n"
"  --all         -a   : All devices\n"
"  --foreground  -F   : Run in foreground (do not fork)\n"
"  --takeover    -t   : Takeover container\n"
);
	exit(2);
}

static bool is_duplicate_opt(const int opt, const int set_val, const char *long_name)
{
	if (opt == set_val) {
		pr_err("--%s option duplicated!\n", long_name);
		return true;
	}
	return false;
}

static int mdmon(char *devnm, int must_fork, int takeover);

int main(int argc, char *argv[])
{
	char *container_name = NULL;
	char *devnm = NULL;
	int status = 0;
	int opt;
	int all = 0;
	int takeover = 0;
	int dofork = 1;
	bool help = false;
	static struct option options[] = {
		{"all", 0, NULL, 'a'},
		{"takeover", 0, NULL, 't'},
		{"help", 0, NULL, 'h'},
		{"offroot", 0, NULL, OffRootOpt},
		{"foreground", 0, NULL, 'F'},
		{NULL, 0, NULL, 0}
	};

	/*
	 * mdmon should never complain due to lack of a platform,
	 * that is mdadm's job if at all.
	 */
	imsm_set_no_platform(1);

	while ((opt = getopt_long(argc, argv, "thaF", options, NULL)) != -1) {
		switch (opt) {
		case 'a':
			if (is_duplicate_opt(all, 1, "all"))
				exit(1);
			container_name = argv[optind-1];
			all = 1;
			break;
		case 't':
			if (is_duplicate_opt(takeover, 1, "takeover"))
				exit(1);
			takeover = 1;
			break;
		case 'F':
			if (is_duplicate_opt(dofork, 0, "foreground"))
				exit(1);
			dofork = 0;
			break;
		case OffRootOpt:
			if (is_duplicate_opt(argv[0][0], '@', "offroot"))
				exit(1);
			argv[0][0] = '@';
			break;
		case 'h':
			if (is_duplicate_opt(help, true, "help"))
				exit(1);
			help = true;
			break;
		default:
			usage();
			break;
		}
	}

	if (in_initrd()) {
		/*
		 * set first char of argv[0] to @. This is used by
		 * systemd to signal that the task was launched from
		 * initrd/initramfs and should be preserved during shutdown
		 */
		argv[0][0] = '@';
	}

	if (!all && argv[optind]) {
		static const char prefix[] = "initrd/";
		container_name = argv[optind];
		if (strncmp(container_name, prefix,
			    sizeof(prefix) - 1) == 0)
			container_name += sizeof(prefix)-1;
		container_name = get_md_name(container_name);
		if (!container_name)
			return 1;
	}

	if (container_name == NULL || argc - optind > 1)
		usage();

	if (strcmp(container_name, "/proc/mdstat") == 0)
		all = 1;

	if (help)
		usage();

	if (all) {
		struct mdstat_ent *mdstat, *e;
		int container_len = strlen(container_name);

		/* launch an mdmon instance for each container found */
		mdstat = mdstat_read(0, 0);
		for (e = mdstat; e; e = e->next) {
			if (e->metadata_version &&
			    strncmp(e->metadata_version, "external:", 9) == 0 &&
			    !is_subarray(&e->metadata_version[9])) {
				/* update cmdline so this mdmon instance can be
				 * distinguished from others in a call to ps(1)
				 */
				if (strlen(e->devnm) <= (unsigned)container_len) {
					memset(container_name, 0, container_len);
					sprintf(container_name, "%s", e->devnm);
				}
				status |= mdmon(e->devnm, 1, takeover);
			}
		}
		free_mdstat(mdstat);

		return status;
	} else {
		int mdfd = open_mddev(container_name, 0);
		devnm = fd2devnm(mdfd);

		close(mdfd);
	}

	if (!devnm) {
		pr_err("%s is not a valid md device name\n",
			container_name);
		return 1;
	}
	return mdmon(devnm, dofork && do_fork(), takeover);
}

static int mdmon(char *devnm, int must_fork, int takeover)
{
	int mdfd;
	struct mdinfo *mdi, *di;
	struct supertype *container;
	sigset_t set;
	struct sigaction act;
	int pfd[2];
	int status;
	int ignore;
	pid_t victim = -1;
	int victim_sock = -1;

	dprintf("starting mdmon for %s\n", devnm);

	mdfd = open_dev(devnm);
	if (mdfd < 0) {
		pr_err("%s: %s\n", devnm, strerror(errno));
		return 1;
	}

	/* Fork, and have the child tell us when they are ready */
	if (must_fork) {
		if (pipe(pfd) != 0) {
			pr_err("failed to create pipe\n");
			return 1;
		}
		switch(fork()) {
		case -1:
			pr_err("failed to fork: %s\n", strerror(errno));
			return 1;
		case 0: /* child */
			close(pfd[0]);
			break;
		default: /* parent */
			close(pfd[1]);
			if (read(pfd[0], &status, sizeof(status)) != sizeof(status)) {
				wait(&status);
				status = WEXITSTATUS(status);
			}
			close(pfd[0]);
			return status;
		}
	} else
		pfd[0] = pfd[1] = -1;

	container = xcalloc(1, sizeof(*container));
	strcpy(container->devnm, devnm);
	container->arrays = NULL;
	container->sock = -1;

	mdi = sysfs_read(mdfd, container->devnm, GET_VERSION|GET_LEVEL|GET_DEVS);

	if (!mdi) {
		pr_err("failed to load sysfs info for %s\n", container->devnm);
		exit(3);
	}
	if (mdi->array.level != UnSet) {
		pr_err("%s is not a container - cannot monitor\n", devnm);
		exit(3);
	}
	if (mdi->array.major_version != -1 ||
	    mdi->array.minor_version != -2) {
		pr_err("%s does not use external metadata - cannot monitor\n",
			devnm);
		exit(3);
	}

	container->ss = version_to_superswitch(mdi->text_version);
	if (container->ss == NULL) {
		pr_err("%s uses unsupported metadata: %s\n",
			devnm, mdi->text_version);
		exit(3);
	}

	container->devs = NULL;
	for (di = mdi->devs; di; di = di->next) {
		struct mdinfo *cd = xmalloc(sizeof(*cd));
		*cd = *di;
		cd->next = container->devs;
		container->devs = cd;
	}
	sysfs_free(mdi);

	/* SIGUSR is sent between parent and child.  So both block it
	 * and enable it only with pselect.
	 */
	sigemptyset(&set);
	sigaddset(&set, SIGUSR1);
	sigaddset(&set, SIGTERM);
	sigprocmask(SIG_BLOCK, &set, NULL);
	act.sa_handler = wake_me;
	act.sa_flags = 0;
	sigaction(SIGUSR1, &act, NULL);
	act.sa_handler = term;
	sigaction(SIGTERM, &act, NULL);
	act.sa_handler = SIG_IGN;
	sigaction(SIGPIPE, &act, NULL);

	victim = mdmon_pid(container->devnm);
	if (victim >= 0)
		victim_sock = connect_monitor(container->devnm);

	ignore = chdir("/");
	if (!takeover && victim > 0 && victim_sock >= 0) {
		if (fping_monitor(victim_sock) == 0) {
			pr_err("%s already managed\n", container->devnm);
			exit(3);
		}
		close(victim_sock);
		victim_sock = -1;
	}
	if (container->ss->load_container(container, mdfd, devnm)) {
		pr_err("Cannot load metadata for %s\n", devnm);
		exit(3);
	}
	close(mdfd);

	/* Ok, this is close enough.  We can say goodbye to our parent now.
	 */
	if (victim > 0)
		remove_pidfile(devnm);
	if (make_pidfile(devnm) < 0) {
		exit(3);
	}
	container->sock = make_control_sock(devnm);

	status = 0;
	if (pfd[1] >= 0) {
		if (write(pfd[1], &status, sizeof(status)) < 0)
			pr_err("failed to notify our parent: %d\n",
			       getppid());
		close(pfd[1]);
	}

	mlockall(MCL_CURRENT | MCL_FUTURE);

	if (clone_monitor(container) < 0) {
		pr_err("failed to start monitor process: %s\n",
			strerror(errno));
		exit(2);
	}

	if (victim > 0) {
		try_kill_monitor(victim, container->devnm, victim_sock);
		if (victim_sock >= 0)
			close(victim_sock);
	}

	setsid();
	manage_fork_fds(0);

	/* This silliness is to stop the compiler complaining
	 * that we ignore 'ignore'
	 */
	if (ignore)
		ignore++;

	do_manager(container);

	exit(0);
}

/* Some stub functions so super-* can link with us */
int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
		  struct supertype *st, unsigned long blocks,
		  int *fds, unsigned long long *offsets,
		  int dests, int *destfd, unsigned long long *destoffsets)
{
	return 0;
}

int restore_stripes(int *dest, unsigned long long *offsets,
		    int raid_disks, int chunk_size, int level, int layout,
		    int source, unsigned long long read_offset,
		    unsigned long long start, unsigned long long length,
		    char *src_buf)
{
	return 1;
}

int save_stripes(int *source, unsigned long long *offsets,
		 int raid_disks, int chunk_size, int level, int layout,
		 int nwrites, int *dest,
		 unsigned long long start, unsigned long long length,
		 char *buf)
{
	return 0;
}

struct superswitch super0 = {
	.name = "0.90",
};
struct superswitch super1 = {
	.name = "1.x",
};