xref: /freebsd/sbin/hastd/secondary.c (revision 70e0bbedef95258a4dadc996d641a9bebd3f107d)
1 /*-
2  * Copyright (c) 2009-2010 The FreeBSD Foundation
3  * Copyright (c) 2010 Pawel Jakub Dawidek <pjd@FreeBSD.org>
4  * All rights reserved.
5  *
6  * This software was developed by Pawel Jakub Dawidek under sponsorship from
7  * the FreeBSD Foundation.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
22  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28  * SUCH DAMAGE.
29  */
30 
31 #include <sys/cdefs.h>
32 __FBSDID("$FreeBSD$");
33 
34 #include <sys/param.h>
35 #include <sys/time.h>
36 #include <sys/bio.h>
37 #include <sys/disk.h>
38 #include <sys/stat.h>
39 
40 #include <err.h>
41 #include <errno.h>
42 #include <fcntl.h>
43 #include <libgeom.h>
44 #include <pthread.h>
45 #include <signal.h>
46 #include <stdint.h>
47 #include <stdio.h>
48 #include <string.h>
49 #include <sysexits.h>
50 #include <unistd.h>
51 
52 #include <activemap.h>
53 #include <nv.h>
54 #include <pjdlog.h>
55 
56 #include "control.h"
57 #include "event.h"
58 #include "hast.h"
59 #include "hast_proto.h"
60 #include "hastd.h"
61 #include "hooks.h"
62 #include "metadata.h"
63 #include "proto.h"
64 #include "subr.h"
65 #include "synch.h"
66 
67 struct hio {
68 	uint64_t	 hio_seq;
69 	int		 hio_error;
70 	void		*hio_data;
71 	uint8_t		 hio_cmd;
72 	uint64_t	 hio_offset;
73 	uint64_t	 hio_length;
74 	TAILQ_ENTRY(hio) hio_next;
75 };
76 
77 static struct hast_resource *gres;
78 
79 /*
80  * Free list holds unused structures. When free list is empty, we have to wait
81  * until some in-progress requests are freed.
82  */
83 static TAILQ_HEAD(, hio) hio_free_list;
84 static pthread_mutex_t hio_free_list_lock;
85 static pthread_cond_t hio_free_list_cond;
86 /*
87  * Disk thread (the one that do I/O requests) takes requests from this list.
88  */
89 static TAILQ_HEAD(, hio) hio_disk_list;
90 static pthread_mutex_t hio_disk_list_lock;
91 static pthread_cond_t hio_disk_list_cond;
92 /*
93  * There is one recv list for every component, although local components don't
94  * use recv lists as local requests are done synchronously.
95  */
96 static TAILQ_HEAD(, hio) hio_send_list;
97 static pthread_mutex_t hio_send_list_lock;
98 static pthread_cond_t hio_send_list_cond;
99 
100 /*
101  * Maximum number of outstanding I/O requests.
102  */
103 #define	HAST_HIO_MAX	256
104 
105 static void *recv_thread(void *arg);
106 static void *disk_thread(void *arg);
107 static void *send_thread(void *arg);
108 
109 #define	QUEUE_INSERT(name, hio)	do {					\
110 	bool _wakeup;							\
111 									\
112 	mtx_lock(&hio_##name##_list_lock);				\
113 	_wakeup = TAILQ_EMPTY(&hio_##name##_list);			\
114 	TAILQ_INSERT_TAIL(&hio_##name##_list, (hio), hio_next);		\
115 	mtx_unlock(&hio_##name##_list_lock);				\
116 	if (_wakeup)							\
117 		cv_signal(&hio_##name##_list_cond);			\
118 } while (0)
119 #define	QUEUE_TAKE(name, hio)	do {					\
120 	mtx_lock(&hio_##name##_list_lock);				\
121 	while (((hio) = TAILQ_FIRST(&hio_##name##_list)) == NULL) {	\
122 		cv_wait(&hio_##name##_list_cond,			\
123 		    &hio_##name##_list_lock);				\
124 	}								\
125 	TAILQ_REMOVE(&hio_##name##_list, (hio), hio_next);		\
126 	mtx_unlock(&hio_##name##_list_lock);				\
127 } while (0)
128 
129 static void
130 hio_clear(struct hio *hio)
131 {
132 
133 	hio->hio_seq = 0;
134 	hio->hio_error = 0;
135 	hio->hio_cmd = HIO_UNDEF;
136 	hio->hio_offset = 0;
137 	hio->hio_length = 0;
138 }
139 
140 static void
141 init_environment(void)
142 {
143 	struct hio *hio;
144 	unsigned int ii;
145 
146 	/*
147 	 * Initialize lists, their locks and theirs condition variables.
148 	 */
149 	TAILQ_INIT(&hio_free_list);
150 	mtx_init(&hio_free_list_lock);
151 	cv_init(&hio_free_list_cond);
152 	TAILQ_INIT(&hio_disk_list);
153 	mtx_init(&hio_disk_list_lock);
154 	cv_init(&hio_disk_list_cond);
155 	TAILQ_INIT(&hio_send_list);
156 	mtx_init(&hio_send_list_lock);
157 	cv_init(&hio_send_list_cond);
158 
159 	/*
160 	 * Allocate requests pool and initialize requests.
161 	 */
162 	for (ii = 0; ii < HAST_HIO_MAX; ii++) {
163 		hio = malloc(sizeof(*hio));
164 		if (hio == NULL) {
165 			pjdlog_exitx(EX_TEMPFAIL,
166 			    "Unable to allocate memory (%zu bytes) for hio request.",
167 			    sizeof(*hio));
168 		}
169 		hio->hio_data = malloc(MAXPHYS);
170 		if (hio->hio_data == NULL) {
171 			pjdlog_exitx(EX_TEMPFAIL,
172 			    "Unable to allocate memory (%zu bytes) for gctl_data.",
173 			    (size_t)MAXPHYS);
174 		}
175 		hio_clear(hio);
176 		TAILQ_INSERT_HEAD(&hio_free_list, hio, hio_next);
177 	}
178 }
179 
180 static void
181 init_local(struct hast_resource *res)
182 {
183 
184 	if (metadata_read(res, true) == -1)
185 		exit(EX_NOINPUT);
186 }
187 
188 static void
189 init_remote(struct hast_resource *res, struct nv *nvin)
190 {
191 	uint64_t resuid;
192 	struct nv *nvout;
193 	unsigned char *map;
194 	size_t mapsize;
195 
196 #ifdef notyet
197 	/* Setup direction. */
198 	if (proto_send(res->hr_remoteout, NULL, 0) == -1)
199 		pjdlog_errno(LOG_WARNING, "Unable to set connection direction");
200 #endif
201 
202 	nvout = nv_alloc();
203 	nv_add_int64(nvout, (int64_t)res->hr_datasize, "datasize");
204 	nv_add_int32(nvout, (int32_t)res->hr_extentsize, "extentsize");
205 	resuid = nv_get_uint64(nvin, "resuid");
206 	res->hr_primary_localcnt = nv_get_uint64(nvin, "localcnt");
207 	res->hr_primary_remotecnt = nv_get_uint64(nvin, "remotecnt");
208 	nv_add_uint64(nvout, res->hr_secondary_localcnt, "localcnt");
209 	nv_add_uint64(nvout, res->hr_secondary_remotecnt, "remotecnt");
210 	mapsize = activemap_calc_ondisk_size(res->hr_local_mediasize -
211 	    METADATA_SIZE, res->hr_extentsize, res->hr_local_sectorsize);
212 	map = malloc(mapsize);
213 	if (map == NULL) {
214 		pjdlog_exitx(EX_TEMPFAIL,
215 		    "Unable to allocate memory (%zu bytes) for activemap.",
216 		    mapsize);
217 	}
218 	/*
219 	 * When we work as primary and secondary is missing we will increase
220 	 * localcnt in our metadata. When secondary is connected and synced
221 	 * we make localcnt be equal to remotecnt, which means nodes are more
222 	 * or less in sync.
223 	 * Split-brain condition is when both nodes are not able to communicate
224 	 * and are both configured as primary nodes. In turn, they can both
225 	 * make incompatible changes to the data and we have to detect that.
226 	 * Under split-brain condition we will increase our localcnt on first
227 	 * write and remote node will increase its localcnt on first write.
228 	 * When we connect we can see that primary's localcnt is greater than
229 	 * our remotecnt (primary was modified while we weren't watching) and
230 	 * our localcnt is greater than primary's remotecnt (we were modified
231 	 * while primary wasn't watching).
232 	 * There are many possible combinations which are all gathered below.
233 	 * Don't pay too much attention to exact numbers, the more important
234 	 * is to compare them. We compare secondary's local with primary's
235 	 * remote and secondary's remote with primary's local.
236 	 * Note that every case where primary's localcnt is smaller than
237 	 * secondary's remotecnt and where secondary's localcnt is smaller than
238 	 * primary's remotecnt should be impossible in practise. We will perform
239 	 * full synchronization then. Those cases are marked with an asterisk.
240 	 * Regular synchronization means that only extents marked as dirty are
241 	 * synchronized (regular synchronization).
242 	 *
243 	 * SECONDARY METADATA PRIMARY METADATA
244 	 * local=3 remote=3   local=2 remote=2*  ?! Full sync from secondary.
245 	 * local=3 remote=3   local=2 remote=3*  ?! Full sync from primary.
246 	 * local=3 remote=3   local=2 remote=4*  ?! Full sync from primary.
247 	 * local=3 remote=3   local=3 remote=2   Primary is out-of-date,
248 	 *                                       regular sync from secondary.
249 	 * local=3 remote=3   local=3 remote=3   Regular sync just in case.
250 	 * local=3 remote=3   local=3 remote=4*  ?! Full sync from primary.
251 	 * local=3 remote=3   local=4 remote=2   Split-brain condition.
252 	 * local=3 remote=3   local=4 remote=3   Secondary out-of-date,
253 	 *                                       regular sync from primary.
254 	 * local=3 remote=3   local=4 remote=4*  ?! Full sync from primary.
255 	 */
256 	if (res->hr_resuid == 0) {
257 		/*
258 		 * Provider is used for the first time. If primary node done no
259 		 * writes yet as well (we will find "virgin" argument) then
260 		 * there is no need to synchronize anything. If primary node
261 		 * done any writes already we have to synchronize everything.
262 		 */
263 		PJDLOG_ASSERT(res->hr_secondary_localcnt == 0);
264 		res->hr_resuid = resuid;
265 		if (metadata_write(res) == -1)
266 			exit(EX_NOINPUT);
267 		if (nv_exists(nvin, "virgin")) {
268 			free(map);
269 			map = NULL;
270 			mapsize = 0;
271 		} else {
272 			memset(map, 0xff, mapsize);
273 		}
274 		nv_add_int8(nvout, 1, "virgin");
275 		nv_add_uint8(nvout, HAST_SYNCSRC_PRIMARY, "syncsrc");
276 	} else if (res->hr_resuid != resuid) {
277 		char errmsg[256];
278 
279 		free(map);
280 		(void)snprintf(errmsg, sizeof(errmsg),
281 		    "Resource unique ID mismatch (primary=%ju, secondary=%ju).",
282 		    (uintmax_t)resuid, (uintmax_t)res->hr_resuid);
283 		pjdlog_error("%s", errmsg);
284 		nv_add_string(nvout, errmsg, "errmsg");
285 		if (hast_proto_send(res, res->hr_remotein, nvout, NULL, 0) == -1) {
286 			pjdlog_exit(EX_TEMPFAIL, "Unable to send response to %s",
287 			    res->hr_remoteaddr);
288 		}
289 		nv_free(nvout);
290 		exit(EX_CONFIG);
291 	} else if (
292 	    /* Is primary out-of-date? */
293 	    (res->hr_secondary_localcnt > res->hr_primary_remotecnt &&
294 	     res->hr_secondary_remotecnt == res->hr_primary_localcnt) ||
295 	    /* Are the nodes more or less in sync? */
296 	    (res->hr_secondary_localcnt == res->hr_primary_remotecnt &&
297 	     res->hr_secondary_remotecnt == res->hr_primary_localcnt) ||
298 	    /* Is secondary out-of-date? */
299 	    (res->hr_secondary_localcnt == res->hr_primary_remotecnt &&
300 	     res->hr_secondary_remotecnt < res->hr_primary_localcnt)) {
301 		/*
302 		 * Nodes are more or less in sync or one of the nodes is
303 		 * out-of-date.
304 		 * It doesn't matter at this point which one, we just have to
305 		 * send out local bitmap to the remote node.
306 		 */
307 		if (pread(res->hr_localfd, map, mapsize, METADATA_SIZE) !=
308 		    (ssize_t)mapsize) {
309 			pjdlog_exit(LOG_ERR, "Unable to read activemap");
310 		}
311 		if (res->hr_secondary_localcnt > res->hr_primary_remotecnt &&
312 		     res->hr_secondary_remotecnt == res->hr_primary_localcnt) {
313 			/* Primary is out-of-date, sync from secondary. */
314 			nv_add_uint8(nvout, HAST_SYNCSRC_SECONDARY, "syncsrc");
315 		} else {
316 			/*
317 			 * Secondary is out-of-date or counts match.
318 			 * Sync from primary.
319 			 */
320 			nv_add_uint8(nvout, HAST_SYNCSRC_PRIMARY, "syncsrc");
321 		}
322 	} else if (res->hr_secondary_localcnt > res->hr_primary_remotecnt &&
323 	     res->hr_primary_localcnt > res->hr_secondary_remotecnt) {
324 		/*
325 		 * Not good, we have split-brain condition.
326 		 */
327 		free(map);
328 		pjdlog_error("Split-brain detected, exiting.");
329 		nv_add_string(nvout, "Split-brain condition!", "errmsg");
330 		if (hast_proto_send(res, res->hr_remotein, nvout, NULL, 0) == -1) {
331 			pjdlog_exit(EX_TEMPFAIL, "Unable to send response to %s",
332 			    res->hr_remoteaddr);
333 		}
334 		nv_free(nvout);
335 		/* Exit on split-brain. */
336 		event_send(res, EVENT_SPLITBRAIN);
337 		exit(EX_CONFIG);
338 	} else /* if (res->hr_secondary_localcnt < res->hr_primary_remotecnt ||
339 	    res->hr_primary_localcnt < res->hr_secondary_remotecnt) */ {
340 		/*
341 		 * This should never happen in practise, but we will perform
342 		 * full synchronization.
343 		 */
344 		PJDLOG_ASSERT(res->hr_secondary_localcnt < res->hr_primary_remotecnt ||
345 		    res->hr_primary_localcnt < res->hr_secondary_remotecnt);
346 		mapsize = activemap_calc_ondisk_size(res->hr_local_mediasize -
347 		    METADATA_SIZE, res->hr_extentsize,
348 		    res->hr_local_sectorsize);
349 		memset(map, 0xff, mapsize);
350 		if (res->hr_secondary_localcnt > res->hr_primary_remotecnt) {
351 			/* In this one of five cases sync from secondary. */
352 			nv_add_uint8(nvout, HAST_SYNCSRC_SECONDARY, "syncsrc");
353 		} else {
354 			/* For the rest four cases sync from primary. */
355 			nv_add_uint8(nvout, HAST_SYNCSRC_PRIMARY, "syncsrc");
356 		}
357 		pjdlog_warning("This should never happen, asking for full synchronization (primary(local=%ju, remote=%ju), secondary(local=%ju, remote=%ju)).",
358 		    (uintmax_t)res->hr_primary_localcnt,
359 		    (uintmax_t)res->hr_primary_remotecnt,
360 		    (uintmax_t)res->hr_secondary_localcnt,
361 		    (uintmax_t)res->hr_secondary_remotecnt);
362 	}
363 	nv_add_uint32(nvout, (uint32_t)mapsize, "mapsize");
364 	if (hast_proto_send(res, res->hr_remotein, nvout, map, mapsize) == -1) {
365 		pjdlog_exit(EX_TEMPFAIL, "Unable to send activemap to %s",
366 		    res->hr_remoteaddr);
367 	}
368 	if (map != NULL)
369 		free(map);
370 	nv_free(nvout);
371 #ifdef notyet
372 	/* Setup direction. */
373 	if (proto_recv(res->hr_remotein, NULL, 0) == -1)
374 		pjdlog_errno(LOG_WARNING, "Unable to set connection direction");
375 #endif
376 }
377 
378 void
379 hastd_secondary(struct hast_resource *res, struct nv *nvin)
380 {
381 	sigset_t mask;
382 	pthread_t td;
383 	pid_t pid;
384 	int error, mode, debuglevel;
385 
386 	/*
387 	 * Create communication channel between parent and child.
388 	 */
389 	if (proto_client(NULL, "socketpair://", &res->hr_ctrl) == -1) {
390 		KEEP_ERRNO((void)pidfile_remove(pfh));
391 		pjdlog_exit(EX_OSERR,
392 		    "Unable to create control sockets between parent and child");
393 	}
394 	/*
395 	 * Create communication channel between child and parent.
396 	 */
397 	if (proto_client(NULL, "socketpair://", &res->hr_event) == -1) {
398 		KEEP_ERRNO((void)pidfile_remove(pfh));
399 		pjdlog_exit(EX_OSERR,
400 		    "Unable to create event sockets between child and parent");
401 	}
402 
403 	pid = fork();
404 	if (pid == -1) {
405 		KEEP_ERRNO((void)pidfile_remove(pfh));
406 		pjdlog_exit(EX_OSERR, "Unable to fork");
407 	}
408 
409 	if (pid > 0) {
410 		/* This is parent. */
411 		proto_close(res->hr_remotein);
412 		res->hr_remotein = NULL;
413 		proto_close(res->hr_remoteout);
414 		res->hr_remoteout = NULL;
415 		/* Declare that we are receiver. */
416 		proto_recv(res->hr_event, NULL, 0);
417 		/* Declare that we are sender. */
418 		proto_send(res->hr_ctrl, NULL, 0);
419 		res->hr_workerpid = pid;
420 		return;
421 	}
422 
423 	gres = res;
424 	mode = pjdlog_mode_get();
425 	debuglevel = pjdlog_debug_get();
426 
427 	/* Declare that we are sender. */
428 	proto_send(res->hr_event, NULL, 0);
429 	/* Declare that we are receiver. */
430 	proto_recv(res->hr_ctrl, NULL, 0);
431 	descriptors_cleanup(res);
432 
433 	descriptors_assert(res, mode);
434 
435 	pjdlog_init(mode);
436 	pjdlog_debug_set(debuglevel);
437 	pjdlog_prefix_set("[%s] (%s) ", res->hr_name, role2str(res->hr_role));
438 	setproctitle("%s (%s)", res->hr_name, role2str(res->hr_role));
439 
440 	PJDLOG_VERIFY(sigemptyset(&mask) == 0);
441 	PJDLOG_VERIFY(sigprocmask(SIG_SETMASK, &mask, NULL) == 0);
442 
443 	/* Error in setting timeout is not critical, but why should it fail? */
444 	if (proto_timeout(res->hr_remotein, 2 * HAST_KEEPALIVE) == -1)
445 		pjdlog_errno(LOG_WARNING, "Unable to set connection timeout");
446 	if (proto_timeout(res->hr_remoteout, res->hr_timeout) == -1)
447 		pjdlog_errno(LOG_WARNING, "Unable to set connection timeout");
448 
449 	init_local(res);
450 	init_environment();
451 
452 	if (drop_privs(res) != 0)
453 		exit(EX_CONFIG);
454 	pjdlog_info("Privileges successfully dropped.");
455 
456 	/*
457 	 * Create the control thread before sending any event to the parent,
458 	 * as we can deadlock when parent sends control request to worker,
459 	 * but worker has no control thread started yet, so parent waits.
460 	 * In the meantime worker sends an event to the parent, but parent
461 	 * is unable to handle the event, because it waits for control
462 	 * request response.
463 	 */
464 	error = pthread_create(&td, NULL, ctrl_thread, res);
465 	PJDLOG_ASSERT(error == 0);
466 
467 	init_remote(res, nvin);
468 	event_send(res, EVENT_CONNECT);
469 
470 	error = pthread_create(&td, NULL, recv_thread, res);
471 	PJDLOG_ASSERT(error == 0);
472 	error = pthread_create(&td, NULL, disk_thread, res);
473 	PJDLOG_ASSERT(error == 0);
474 	(void)send_thread(res);
475 }
476 
477 static void
478 reqlog(int loglevel, int debuglevel, int error, struct hio *hio, const char *fmt, ...)
479 {
480 	char msg[1024];
481 	va_list ap;
482 	int len;
483 
484 	va_start(ap, fmt);
485 	len = vsnprintf(msg, sizeof(msg), fmt, ap);
486 	va_end(ap);
487 	if ((size_t)len < sizeof(msg)) {
488 		switch (hio->hio_cmd) {
489 		case HIO_READ:
490 			(void)snprintf(msg + len, sizeof(msg) - len,
491 			    "READ(%ju, %ju).", (uintmax_t)hio->hio_offset,
492 			    (uintmax_t)hio->hio_length);
493 			break;
494 		case HIO_DELETE:
495 			(void)snprintf(msg + len, sizeof(msg) - len,
496 			    "DELETE(%ju, %ju).", (uintmax_t)hio->hio_offset,
497 			    (uintmax_t)hio->hio_length);
498 			break;
499 		case HIO_FLUSH:
500 			(void)snprintf(msg + len, sizeof(msg) - len, "FLUSH.");
501 			break;
502 		case HIO_WRITE:
503 			(void)snprintf(msg + len, sizeof(msg) - len,
504 			    "WRITE(%ju, %ju).", (uintmax_t)hio->hio_offset,
505 			    (uintmax_t)hio->hio_length);
506 			break;
507 		case HIO_KEEPALIVE:
508 			(void)snprintf(msg + len, sizeof(msg) - len, "KEEPALIVE.");
509 			break;
510 		default:
511 			(void)snprintf(msg + len, sizeof(msg) - len,
512 			    "UNKNOWN(%u).", (unsigned int)hio->hio_cmd);
513 			break;
514 		}
515 	}
516 	pjdlog_common(loglevel, debuglevel, error, "%s", msg);
517 }
518 
519 static int
520 requnpack(struct hast_resource *res, struct hio *hio, struct nv *nv)
521 {
522 
523 	hio->hio_cmd = nv_get_uint8(nv, "cmd");
524 	if (hio->hio_cmd == 0) {
525 		pjdlog_error("Header contains no 'cmd' field.");
526 		hio->hio_error = EINVAL;
527 		goto end;
528 	}
529 	if (hio->hio_cmd != HIO_KEEPALIVE) {
530 		hio->hio_seq = nv_get_uint64(nv, "seq");
531 		if (hio->hio_seq == 0) {
532 			pjdlog_error("Header contains no 'seq' field.");
533 			hio->hio_error = EINVAL;
534 			goto end;
535 		}
536 	}
537 	switch (hio->hio_cmd) {
538 	case HIO_FLUSH:
539 	case HIO_KEEPALIVE:
540 		break;
541 	case HIO_READ:
542 	case HIO_WRITE:
543 	case HIO_DELETE:
544 		hio->hio_offset = nv_get_uint64(nv, "offset");
545 		if (nv_error(nv) != 0) {
546 			pjdlog_error("Header is missing 'offset' field.");
547 			hio->hio_error = EINVAL;
548 			goto end;
549 		}
550 		hio->hio_length = nv_get_uint64(nv, "length");
551 		if (nv_error(nv) != 0) {
552 			pjdlog_error("Header is missing 'length' field.");
553 			hio->hio_error = EINVAL;
554 			goto end;
555 		}
556 		if (hio->hio_length == 0) {
557 			pjdlog_error("Data length is zero.");
558 			hio->hio_error = EINVAL;
559 			goto end;
560 		}
561 		if (hio->hio_length > MAXPHYS) {
562 			pjdlog_error("Data length is too large (%ju > %ju).",
563 			    (uintmax_t)hio->hio_length, (uintmax_t)MAXPHYS);
564 			hio->hio_error = EINVAL;
565 			goto end;
566 		}
567 		if ((hio->hio_offset % res->hr_local_sectorsize) != 0) {
568 			pjdlog_error("Offset %ju is not multiple of sector size.",
569 			    (uintmax_t)hio->hio_offset);
570 			hio->hio_error = EINVAL;
571 			goto end;
572 		}
573 		if ((hio->hio_length % res->hr_local_sectorsize) != 0) {
574 			pjdlog_error("Length %ju is not multiple of sector size.",
575 			    (uintmax_t)hio->hio_length);
576 			hio->hio_error = EINVAL;
577 			goto end;
578 		}
579 		if (hio->hio_offset + hio->hio_length >
580 		    (uint64_t)res->hr_datasize) {
581 			pjdlog_error("Data offset is too large (%ju > %ju).",
582 			    (uintmax_t)(hio->hio_offset + hio->hio_length),
583 			    (uintmax_t)res->hr_datasize);
584 			hio->hio_error = EINVAL;
585 			goto end;
586 		}
587 		break;
588 	default:
589 		pjdlog_error("Header contains invalid 'cmd' (%hhu).",
590 		    hio->hio_cmd);
591 		hio->hio_error = EINVAL;
592 		goto end;
593 	}
594 	hio->hio_error = 0;
595 end:
596 	return (hio->hio_error);
597 }
598 
599 static __dead2 void
600 secondary_exit(int exitcode, const char *fmt, ...)
601 {
602 	va_list ap;
603 
604 	PJDLOG_ASSERT(exitcode != EX_OK);
605 	va_start(ap, fmt);
606 	pjdlogv_errno(LOG_ERR, fmt, ap);
607 	va_end(ap);
608 	event_send(gres, EVENT_DISCONNECT);
609 	exit(exitcode);
610 }
611 
612 /*
613  * Thread receives requests from the primary node.
614  */
615 static void *
616 recv_thread(void *arg)
617 {
618 	struct hast_resource *res = arg;
619 	struct hio *hio;
620 	struct nv *nv;
621 
622 	for (;;) {
623 		pjdlog_debug(2, "recv: Taking free request.");
624 		QUEUE_TAKE(free, hio);
625 		pjdlog_debug(2, "recv: (%p) Got request.", hio);
626 		if (hast_proto_recv_hdr(res->hr_remotein, &nv) == -1) {
627 			secondary_exit(EX_TEMPFAIL,
628 			    "Unable to receive request header");
629 		}
630 		if (requnpack(res, hio, nv) != 0) {
631 			nv_free(nv);
632 			pjdlog_debug(2,
633 			    "recv: (%p) Moving request to the send queue.",
634 			    hio);
635 			QUEUE_INSERT(send, hio);
636 			continue;
637 		}
638 		switch (hio->hio_cmd) {
639 		case HIO_READ:
640 			res->hr_stat_read++;
641 			break;
642 		case HIO_WRITE:
643 			res->hr_stat_write++;
644 			break;
645 		case HIO_DELETE:
646 			res->hr_stat_delete++;
647 			break;
648 		case HIO_FLUSH:
649 			res->hr_stat_flush++;
650 			break;
651 		case HIO_KEEPALIVE:
652 			break;
653 		default:
654 			PJDLOG_ABORT("Unexpected command (cmd=%hhu).",
655 			    hio->hio_cmd);
656 		}
657 		reqlog(LOG_DEBUG, 2, -1, hio,
658 		    "recv: (%p) Got request header: ", hio);
659 		if (hio->hio_cmd == HIO_KEEPALIVE) {
660 			nv_free(nv);
661 			pjdlog_debug(2,
662 			    "recv: (%p) Moving request to the free queue.",
663 			    hio);
664 			hio_clear(hio);
665 			QUEUE_INSERT(free, hio);
666 			continue;
667 		} else if (hio->hio_cmd == HIO_WRITE) {
668 			if (hast_proto_recv_data(res, res->hr_remotein, nv,
669 			    hio->hio_data, MAXPHYS) == -1) {
670 				secondary_exit(EX_TEMPFAIL,
671 				    "Unable to receive request data");
672 			}
673 		}
674 		nv_free(nv);
675 		pjdlog_debug(2, "recv: (%p) Moving request to the disk queue.",
676 		    hio);
677 		QUEUE_INSERT(disk, hio);
678 	}
679 	/* NOTREACHED */
680 	return (NULL);
681 }
682 
683 /*
684  * Thread reads from or writes to local component and also handles DELETE and
685  * FLUSH requests.
686  */
687 static void *
688 disk_thread(void *arg)
689 {
690 	struct hast_resource *res = arg;
691 	struct hio *hio;
692 	ssize_t ret;
693 	bool clear_activemap, logerror;
694 
695 	clear_activemap = true;
696 
697 	for (;;) {
698 		pjdlog_debug(2, "disk: Taking request.");
699 		QUEUE_TAKE(disk, hio);
700 		while (clear_activemap) {
701 			unsigned char *map;
702 			size_t mapsize;
703 
704 			/*
705 			 * When first request is received, it means that primary
706 			 * already received our activemap, merged it and stored
707 			 * locally. We can now safely clear our activemap.
708 			 */
709 			mapsize =
710 			    activemap_calc_ondisk_size(res->hr_local_mediasize -
711 			    METADATA_SIZE, res->hr_extentsize,
712 			    res->hr_local_sectorsize);
713 			map = calloc(1, mapsize);
714 			if (map == NULL) {
715 				pjdlog_warning("Unable to allocate memory to clear local activemap.");
716 				break;
717 			}
718 			if (pwrite(res->hr_localfd, map, mapsize,
719 			    METADATA_SIZE) != (ssize_t)mapsize) {
720 				pjdlog_errno(LOG_WARNING,
721 				    "Unable to store cleared activemap");
722 				free(map);
723 				break;
724 			}
725 			free(map);
726 			clear_activemap = false;
727 			pjdlog_debug(1, "Local activemap cleared.");
728 			break;
729 		}
730 		reqlog(LOG_DEBUG, 2, -1, hio, "disk: (%p) Got request: ", hio);
731 		logerror = true;
732 		/* Handle the actual request. */
733 		switch (hio->hio_cmd) {
734 		case HIO_READ:
735 			ret = pread(res->hr_localfd, hio->hio_data,
736 			    hio->hio_length,
737 			    hio->hio_offset + res->hr_localoff);
738 			if (ret == -1)
739 				hio->hio_error = errno;
740 			else if (ret != (int64_t)hio->hio_length)
741 				hio->hio_error = EIO;
742 			else
743 				hio->hio_error = 0;
744 			break;
745 		case HIO_WRITE:
746 			ret = pwrite(res->hr_localfd, hio->hio_data,
747 			    hio->hio_length,
748 			    hio->hio_offset + res->hr_localoff);
749 			if (ret == -1)
750 				hio->hio_error = errno;
751 			else if (ret != (int64_t)hio->hio_length)
752 				hio->hio_error = EIO;
753 			else
754 				hio->hio_error = 0;
755 			break;
756 		case HIO_DELETE:
757 			ret = g_delete(res->hr_localfd,
758 			    hio->hio_offset + res->hr_localoff,
759 			    hio->hio_length);
760 			if (ret == -1)
761 				hio->hio_error = errno;
762 			else
763 				hio->hio_error = 0;
764 			break;
765 		case HIO_FLUSH:
766 			if (!res->hr_localflush) {
767 				ret = -1;
768 				hio->hio_error = EOPNOTSUPP;
769 				logerror = false;
770 				break;
771 			}
772 			ret = g_flush(res->hr_localfd);
773 			if (ret == -1) {
774 				if (errno == EOPNOTSUPP)
775 					res->hr_localflush = false;
776 				hio->hio_error = errno;
777 			} else {
778 				hio->hio_error = 0;
779 			}
780 			break;
781 		default:
782 			PJDLOG_ABORT("Unexpected command (cmd=%hhu).",
783 			    hio->hio_cmd);
784 		}
785 		if (logerror && hio->hio_error != 0) {
786 			reqlog(LOG_ERR, 0, hio->hio_error, hio,
787 			    "Request failed: ");
788 		}
789 		pjdlog_debug(2, "disk: (%p) Moving request to the send queue.",
790 		    hio);
791 		QUEUE_INSERT(send, hio);
792 	}
793 	/* NOTREACHED */
794 	return (NULL);
795 }
796 
797 /*
798  * Thread sends requests back to primary node.
799  */
800 static void *
801 send_thread(void *arg)
802 {
803 	struct hast_resource *res = arg;
804 	struct nv *nvout;
805 	struct hio *hio;
806 	void *data;
807 	size_t length;
808 
809 	for (;;) {
810 		pjdlog_debug(2, "send: Taking request.");
811 		QUEUE_TAKE(send, hio);
812 		reqlog(LOG_DEBUG, 2, -1, hio, "send: (%p) Got request: ", hio);
813 		nvout = nv_alloc();
814 		/* Copy sequence number. */
815 		nv_add_uint64(nvout, hio->hio_seq, "seq");
816 		switch (hio->hio_cmd) {
817 		case HIO_READ:
818 			if (hio->hio_error == 0) {
819 				data = hio->hio_data;
820 				length = hio->hio_length;
821 				break;
822 			}
823 			/*
824 			 * We send no data in case of an error.
825 			 */
826 			/* FALLTHROUGH */
827 		case HIO_DELETE:
828 		case HIO_FLUSH:
829 		case HIO_WRITE:
830 			data = NULL;
831 			length = 0;
832 			break;
833 		default:
834 			PJDLOG_ABORT("Unexpected command (cmd=%hhu).",
835 			    hio->hio_cmd);
836 		}
837 		if (hio->hio_error != 0)
838 			nv_add_int16(nvout, hio->hio_error, "error");
839 		if (hast_proto_send(res, res->hr_remoteout, nvout, data,
840 		    length) == -1) {
841 			secondary_exit(EX_TEMPFAIL, "Unable to send reply.");
842 		}
843 		nv_free(nvout);
844 		pjdlog_debug(2, "send: (%p) Moving request to the free queue.",
845 		    hio);
846 		hio_clear(hio);
847 		QUEUE_INSERT(free, hio);
848 	}
849 	/* NOTREACHED */
850 	return (NULL);
851 }
852