xref: /openbmc/linux/fs/dlm/member.c (revision 56b5b1c7)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /******************************************************************************
3 *******************************************************************************
4 **
5 **  Copyright (C) 2005-2011 Red Hat, Inc.  All rights reserved.
6 **
7 **
8 *******************************************************************************
9 ******************************************************************************/
10 
11 #include "dlm_internal.h"
12 #include "lockspace.h"
13 #include "member.h"
14 #include "recoverd.h"
15 #include "recover.h"
16 #include "rcom.h"
17 #include "config.h"
18 #include "midcomms.h"
19 #include "lowcomms.h"
20 
21 int dlm_slots_version(struct dlm_header *h)
22 {
23 	if ((h->h_version & 0x0000FFFF) < DLM_HEADER_SLOTS)
24 		return 0;
25 	return 1;
26 }
27 
28 void dlm_slot_save(struct dlm_ls *ls, struct dlm_rcom *rc,
29 		   struct dlm_member *memb)
30 {
31 	struct rcom_config *rf = (struct rcom_config *)rc->rc_buf;
32 
33 	if (!dlm_slots_version(&rc->rc_header))
34 		return;
35 
36 	memb->slot = le16_to_cpu(rf->rf_our_slot);
37 	memb->generation = le32_to_cpu(rf->rf_generation);
38 }
39 
40 void dlm_slots_copy_out(struct dlm_ls *ls, struct dlm_rcom *rc)
41 {
42 	struct dlm_slot *slot;
43 	struct rcom_slot *ro;
44 	int i;
45 
46 	ro = (struct rcom_slot *)(rc->rc_buf + sizeof(struct rcom_config));
47 
48 	/* ls_slots array is sparse, but not rcom_slots */
49 
50 	for (i = 0; i < ls->ls_slots_size; i++) {
51 		slot = &ls->ls_slots[i];
52 		if (!slot->nodeid)
53 			continue;
54 		ro->ro_nodeid = cpu_to_le32(slot->nodeid);
55 		ro->ro_slot = cpu_to_le16(slot->slot);
56 		ro++;
57 	}
58 }
59 
60 #define SLOT_DEBUG_LINE 128
61 
62 static void log_slots(struct dlm_ls *ls, uint32_t gen, int num_slots,
63 		      struct rcom_slot *ro0, struct dlm_slot *array,
64 		      int array_size)
65 {
66 	char line[SLOT_DEBUG_LINE];
67 	int len = SLOT_DEBUG_LINE - 1;
68 	int pos = 0;
69 	int ret, i;
70 
71 	memset(line, 0, sizeof(line));
72 
73 	if (array) {
74 		for (i = 0; i < array_size; i++) {
75 			if (!array[i].nodeid)
76 				continue;
77 
78 			ret = snprintf(line + pos, len - pos, " %d:%d",
79 				       array[i].slot, array[i].nodeid);
80 			if (ret >= len - pos)
81 				break;
82 			pos += ret;
83 		}
84 	} else if (ro0) {
85 		for (i = 0; i < num_slots; i++) {
86 			ret = snprintf(line + pos, len - pos, " %d:%d",
87 				       ro0[i].ro_slot, ro0[i].ro_nodeid);
88 			if (ret >= len - pos)
89 				break;
90 			pos += ret;
91 		}
92 	}
93 
94 	log_rinfo(ls, "generation %u slots %d%s", gen, num_slots, line);
95 }
96 
97 int dlm_slots_copy_in(struct dlm_ls *ls)
98 {
99 	struct dlm_member *memb;
100 	struct dlm_rcom *rc = ls->ls_recover_buf;
101 	struct rcom_config *rf = (struct rcom_config *)rc->rc_buf;
102 	struct rcom_slot *ro0, *ro;
103 	int our_nodeid = dlm_our_nodeid();
104 	int i, num_slots;
105 	uint32_t gen;
106 
107 	if (!dlm_slots_version(&rc->rc_header))
108 		return -1;
109 
110 	gen = le32_to_cpu(rf->rf_generation);
111 	if (gen <= ls->ls_generation) {
112 		log_error(ls, "dlm_slots_copy_in gen %u old %u",
113 			  gen, ls->ls_generation);
114 	}
115 	ls->ls_generation = gen;
116 
117 	num_slots = le16_to_cpu(rf->rf_num_slots);
118 	if (!num_slots)
119 		return -1;
120 
121 	ro0 = (struct rcom_slot *)(rc->rc_buf + sizeof(struct rcom_config));
122 
123 	for (i = 0, ro = ro0; i < num_slots; i++, ro++) {
124 		ro->ro_nodeid = le32_to_cpu(ro->ro_nodeid);
125 		ro->ro_slot = le16_to_cpu(ro->ro_slot);
126 	}
127 
128 	log_slots(ls, gen, num_slots, ro0, NULL, 0);
129 
130 	list_for_each_entry(memb, &ls->ls_nodes, list) {
131 		for (i = 0, ro = ro0; i < num_slots; i++, ro++) {
132 			if (ro->ro_nodeid != memb->nodeid)
133 				continue;
134 			memb->slot = ro->ro_slot;
135 			memb->slot_prev = memb->slot;
136 			break;
137 		}
138 
139 		if (memb->nodeid == our_nodeid) {
140 			if (ls->ls_slot && ls->ls_slot != memb->slot) {
141 				log_error(ls, "dlm_slots_copy_in our slot "
142 					  "changed %d %d", ls->ls_slot,
143 					  memb->slot);
144 				return -1;
145 			}
146 
147 			if (!ls->ls_slot)
148 				ls->ls_slot = memb->slot;
149 		}
150 
151 		if (!memb->slot) {
152 			log_error(ls, "dlm_slots_copy_in nodeid %d no slot",
153 				   memb->nodeid);
154 			return -1;
155 		}
156 	}
157 
158 	return 0;
159 }
160 
161 /* for any nodes that do not support slots, we will not have set memb->slot
162    in wait_status_all(), so memb->slot will remain -1, and we will not
163    assign slots or set ls_num_slots here */
164 
165 int dlm_slots_assign(struct dlm_ls *ls, int *num_slots, int *slots_size,
166 		     struct dlm_slot **slots_out, uint32_t *gen_out)
167 {
168 	struct dlm_member *memb;
169 	struct dlm_slot *array;
170 	int our_nodeid = dlm_our_nodeid();
171 	int array_size, max_slots, i;
172 	int need = 0;
173 	int max = 0;
174 	int num = 0;
175 	uint32_t gen = 0;
176 
177 	/* our own memb struct will have slot -1 gen 0 */
178 
179 	list_for_each_entry(memb, &ls->ls_nodes, list) {
180 		if (memb->nodeid == our_nodeid) {
181 			memb->slot = ls->ls_slot;
182 			memb->generation = ls->ls_generation;
183 			break;
184 		}
185 	}
186 
187 	list_for_each_entry(memb, &ls->ls_nodes, list) {
188 		if (memb->generation > gen)
189 			gen = memb->generation;
190 
191 		/* node doesn't support slots */
192 
193 		if (memb->slot == -1)
194 			return -1;
195 
196 		/* node needs a slot assigned */
197 
198 		if (!memb->slot)
199 			need++;
200 
201 		/* node has a slot assigned */
202 
203 		num++;
204 
205 		if (!max || max < memb->slot)
206 			max = memb->slot;
207 
208 		/* sanity check, once slot is assigned it shouldn't change */
209 
210 		if (memb->slot_prev && memb->slot && memb->slot_prev != memb->slot) {
211 			log_error(ls, "nodeid %d slot changed %d %d",
212 				  memb->nodeid, memb->slot_prev, memb->slot);
213 			return -1;
214 		}
215 		memb->slot_prev = memb->slot;
216 	}
217 
218 	array_size = max + need;
219 	array = kcalloc(array_size, sizeof(*array), GFP_NOFS);
220 	if (!array)
221 		return -ENOMEM;
222 
223 	num = 0;
224 
225 	/* fill in slots (offsets) that are used */
226 
227 	list_for_each_entry(memb, &ls->ls_nodes, list) {
228 		if (!memb->slot)
229 			continue;
230 
231 		if (memb->slot > array_size) {
232 			log_error(ls, "invalid slot number %d", memb->slot);
233 			kfree(array);
234 			return -1;
235 		}
236 
237 		array[memb->slot - 1].nodeid = memb->nodeid;
238 		array[memb->slot - 1].slot = memb->slot;
239 		num++;
240 	}
241 
242 	/* assign new slots from unused offsets */
243 
244 	list_for_each_entry(memb, &ls->ls_nodes, list) {
245 		if (memb->slot)
246 			continue;
247 
248 		for (i = 0; i < array_size; i++) {
249 			if (array[i].nodeid)
250 				continue;
251 
252 			memb->slot = i + 1;
253 			memb->slot_prev = memb->slot;
254 			array[i].nodeid = memb->nodeid;
255 			array[i].slot = memb->slot;
256 			num++;
257 
258 			if (!ls->ls_slot && memb->nodeid == our_nodeid)
259 				ls->ls_slot = memb->slot;
260 			break;
261 		}
262 
263 		if (!memb->slot) {
264 			log_error(ls, "no free slot found");
265 			kfree(array);
266 			return -1;
267 		}
268 	}
269 
270 	gen++;
271 
272 	log_slots(ls, gen, num, NULL, array, array_size);
273 
274 	max_slots = (DLM_MAX_APP_BUFSIZE - sizeof(struct dlm_rcom) -
275 		     sizeof(struct rcom_config)) / sizeof(struct rcom_slot);
276 
277 	if (num > max_slots) {
278 		log_error(ls, "num_slots %d exceeds max_slots %d",
279 			  num, max_slots);
280 		kfree(array);
281 		return -1;
282 	}
283 
284 	*gen_out = gen;
285 	*slots_out = array;
286 	*slots_size = array_size;
287 	*num_slots = num;
288 	return 0;
289 }
290 
291 static void add_ordered_member(struct dlm_ls *ls, struct dlm_member *new)
292 {
293 	struct dlm_member *memb = NULL;
294 	struct list_head *tmp;
295 	struct list_head *newlist = &new->list;
296 	struct list_head *head = &ls->ls_nodes;
297 
298 	list_for_each(tmp, head) {
299 		memb = list_entry(tmp, struct dlm_member, list);
300 		if (new->nodeid < memb->nodeid)
301 			break;
302 	}
303 
304 	if (!memb)
305 		list_add_tail(newlist, head);
306 	else {
307 		/* FIXME: can use list macro here */
308 		newlist->prev = tmp->prev;
309 		newlist->next = tmp;
310 		tmp->prev->next = newlist;
311 		tmp->prev = newlist;
312 	}
313 }
314 
315 static int dlm_add_member(struct dlm_ls *ls, struct dlm_config_node *node)
316 {
317 	struct dlm_member *memb;
318 	int error;
319 
320 	memb = kzalloc(sizeof(*memb), GFP_NOFS);
321 	if (!memb)
322 		return -ENOMEM;
323 
324 	error = dlm_lowcomms_connect_node(node->nodeid);
325 	if (error < 0) {
326 		kfree(memb);
327 		return error;
328 	}
329 
330 	memb->nodeid = node->nodeid;
331 	memb->weight = node->weight;
332 	memb->comm_seq = node->comm_seq;
333 	dlm_midcomms_add_member(node->nodeid);
334 	add_ordered_member(ls, memb);
335 	ls->ls_num_nodes++;
336 	return 0;
337 }
338 
339 static struct dlm_member *find_memb(struct list_head *head, int nodeid)
340 {
341 	struct dlm_member *memb;
342 
343 	list_for_each_entry(memb, head, list) {
344 		if (memb->nodeid == nodeid)
345 			return memb;
346 	}
347 	return NULL;
348 }
349 
350 int dlm_is_member(struct dlm_ls *ls, int nodeid)
351 {
352 	if (find_memb(&ls->ls_nodes, nodeid))
353 		return 1;
354 	return 0;
355 }
356 
357 int dlm_is_removed(struct dlm_ls *ls, int nodeid)
358 {
359 	if (find_memb(&ls->ls_nodes_gone, nodeid))
360 		return 1;
361 	return 0;
362 }
363 
364 static void clear_memb_list(struct list_head *head,
365 			    void (*after_del)(int nodeid))
366 {
367 	struct dlm_member *memb;
368 
369 	while (!list_empty(head)) {
370 		memb = list_entry(head->next, struct dlm_member, list);
371 		list_del(&memb->list);
372 		if (after_del)
373 			after_del(memb->nodeid);
374 		kfree(memb);
375 	}
376 }
377 
378 static void clear_members_cb(int nodeid)
379 {
380 	dlm_midcomms_remove_member(nodeid);
381 }
382 
383 void dlm_clear_members(struct dlm_ls *ls)
384 {
385 	clear_memb_list(&ls->ls_nodes, clear_members_cb);
386 	ls->ls_num_nodes = 0;
387 }
388 
389 void dlm_clear_members_gone(struct dlm_ls *ls)
390 {
391 	clear_memb_list(&ls->ls_nodes_gone, NULL);
392 }
393 
394 static void make_member_array(struct dlm_ls *ls)
395 {
396 	struct dlm_member *memb;
397 	int i, w, x = 0, total = 0, all_zero = 0, *array;
398 
399 	kfree(ls->ls_node_array);
400 	ls->ls_node_array = NULL;
401 
402 	list_for_each_entry(memb, &ls->ls_nodes, list) {
403 		if (memb->weight)
404 			total += memb->weight;
405 	}
406 
407 	/* all nodes revert to weight of 1 if all have weight 0 */
408 
409 	if (!total) {
410 		total = ls->ls_num_nodes;
411 		all_zero = 1;
412 	}
413 
414 	ls->ls_total_weight = total;
415 	array = kmalloc_array(total, sizeof(*array), GFP_NOFS);
416 	if (!array)
417 		return;
418 
419 	list_for_each_entry(memb, &ls->ls_nodes, list) {
420 		if (!all_zero && !memb->weight)
421 			continue;
422 
423 		if (all_zero)
424 			w = 1;
425 		else
426 			w = memb->weight;
427 
428 		DLM_ASSERT(x < total, printk("total %d x %d\n", total, x););
429 
430 		for (i = 0; i < w; i++)
431 			array[x++] = memb->nodeid;
432 	}
433 
434 	ls->ls_node_array = array;
435 }
436 
437 /* send a status request to all members just to establish comms connections */
438 
439 static int ping_members(struct dlm_ls *ls)
440 {
441 	struct dlm_member *memb;
442 	int error = 0;
443 
444 	list_for_each_entry(memb, &ls->ls_nodes, list) {
445 		if (dlm_recovery_stopped(ls)) {
446 			error = -EINTR;
447 			break;
448 		}
449 		error = dlm_rcom_status(ls, memb->nodeid, 0);
450 		if (error)
451 			break;
452 	}
453 	if (error)
454 		log_rinfo(ls, "ping_members aborted %d last nodeid %d",
455 			  error, ls->ls_recover_nodeid);
456 	return error;
457 }
458 
459 static void dlm_lsop_recover_prep(struct dlm_ls *ls)
460 {
461 	if (!ls->ls_ops || !ls->ls_ops->recover_prep)
462 		return;
463 	ls->ls_ops->recover_prep(ls->ls_ops_arg);
464 }
465 
466 static void dlm_lsop_recover_slot(struct dlm_ls *ls, struct dlm_member *memb)
467 {
468 	struct dlm_slot slot;
469 	uint32_t seq;
470 	int error;
471 
472 	if (!ls->ls_ops || !ls->ls_ops->recover_slot)
473 		return;
474 
475 	/* if there is no comms connection with this node
476 	   or the present comms connection is newer
477 	   than the one when this member was added, then
478 	   we consider the node to have failed (versus
479 	   being removed due to dlm_release_lockspace) */
480 
481 	error = dlm_comm_seq(memb->nodeid, &seq);
482 
483 	if (!error && seq == memb->comm_seq)
484 		return;
485 
486 	slot.nodeid = memb->nodeid;
487 	slot.slot = memb->slot;
488 
489 	ls->ls_ops->recover_slot(ls->ls_ops_arg, &slot);
490 }
491 
492 void dlm_lsop_recover_done(struct dlm_ls *ls)
493 {
494 	struct dlm_member *memb;
495 	struct dlm_slot *slots;
496 	int i, num;
497 
498 	if (!ls->ls_ops || !ls->ls_ops->recover_done)
499 		return;
500 
501 	num = ls->ls_num_nodes;
502 	slots = kcalloc(num, sizeof(*slots), GFP_KERNEL);
503 	if (!slots)
504 		return;
505 
506 	i = 0;
507 	list_for_each_entry(memb, &ls->ls_nodes, list) {
508 		if (i == num) {
509 			log_error(ls, "dlm_lsop_recover_done bad num %d", num);
510 			goto out;
511 		}
512 		slots[i].nodeid = memb->nodeid;
513 		slots[i].slot = memb->slot;
514 		i++;
515 	}
516 
517 	ls->ls_ops->recover_done(ls->ls_ops_arg, slots, num,
518 				 ls->ls_slot, ls->ls_generation);
519  out:
520 	kfree(slots);
521 }
522 
523 static struct dlm_config_node *find_config_node(struct dlm_recover *rv,
524 						int nodeid)
525 {
526 	int i;
527 
528 	for (i = 0; i < rv->nodes_count; i++) {
529 		if (rv->nodes[i].nodeid == nodeid)
530 			return &rv->nodes[i];
531 	}
532 	return NULL;
533 }
534 
535 int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
536 {
537 	struct dlm_member *memb, *safe;
538 	struct dlm_config_node *node;
539 	int i, error, neg = 0, low = -1;
540 
541 	/* previously removed members that we've not finished removing need to
542 	   count as a negative change so the "neg" recovery steps will happen */
543 
544 	list_for_each_entry(memb, &ls->ls_nodes_gone, list) {
545 		log_rinfo(ls, "prev removed member %d", memb->nodeid);
546 		neg++;
547 	}
548 
549 	/* move departed members from ls_nodes to ls_nodes_gone */
550 
551 	list_for_each_entry_safe(memb, safe, &ls->ls_nodes, list) {
552 		node = find_config_node(rv, memb->nodeid);
553 		if (node && !node->new)
554 			continue;
555 
556 		if (!node) {
557 			log_rinfo(ls, "remove member %d", memb->nodeid);
558 		} else {
559 			/* removed and re-added */
560 			log_rinfo(ls, "remove member %d comm_seq %u %u",
561 				  memb->nodeid, memb->comm_seq, node->comm_seq);
562 		}
563 
564 		neg++;
565 		list_move(&memb->list, &ls->ls_nodes_gone);
566 		dlm_midcomms_remove_member(memb->nodeid);
567 		ls->ls_num_nodes--;
568 		dlm_lsop_recover_slot(ls, memb);
569 	}
570 
571 	/* add new members to ls_nodes */
572 
573 	for (i = 0; i < rv->nodes_count; i++) {
574 		node = &rv->nodes[i];
575 		if (dlm_is_member(ls, node->nodeid))
576 			continue;
577 		dlm_add_member(ls, node);
578 		log_rinfo(ls, "add member %d", node->nodeid);
579 	}
580 
581 	list_for_each_entry(memb, &ls->ls_nodes, list) {
582 		if (low == -1 || memb->nodeid < low)
583 			low = memb->nodeid;
584 	}
585 	ls->ls_low_nodeid = low;
586 
587 	make_member_array(ls);
588 	*neg_out = neg;
589 
590 	error = ping_members(ls);
591 	/* error -EINTR means that a new recovery action is triggered.
592 	 * We ignore this recovery action and let run the new one which might
593 	 * have new member configuration.
594 	 */
595 	if (error == -EINTR)
596 		error = 0;
597 
598 	/* new_lockspace() may be waiting to know if the config
599 	 * is good or bad
600 	 */
601 	ls->ls_members_result = error;
602 	complete(&ls->ls_members_done);
603 
604 	log_rinfo(ls, "dlm_recover_members %d nodes", ls->ls_num_nodes);
605 	return error;
606 }
607 
608 /* Userspace guarantees that dlm_ls_stop() has completed on all nodes before
609    dlm_ls_start() is called on any of them to start the new recovery. */
610 
611 int dlm_ls_stop(struct dlm_ls *ls)
612 {
613 	int new;
614 
615 	/*
616 	 * Prevent dlm_recv from being in the middle of something when we do
617 	 * the stop.  This includes ensuring dlm_recv isn't processing a
618 	 * recovery message (rcom), while dlm_recoverd is aborting and
619 	 * resetting things from an in-progress recovery.  i.e. we want
620 	 * dlm_recoverd to abort its recovery without worrying about dlm_recv
621 	 * processing an rcom at the same time.  Stopping dlm_recv also makes
622 	 * it easy for dlm_receive_message() to check locking stopped and add a
623 	 * message to the requestqueue without races.
624 	 */
625 
626 	down_write(&ls->ls_recv_active);
627 
628 	/*
629 	 * Abort any recovery that's in progress (see RECOVER_STOP,
630 	 * dlm_recovery_stopped()) and tell any other threads running in the
631 	 * dlm to quit any processing (see RUNNING, dlm_locking_stopped()).
632 	 */
633 
634 	spin_lock(&ls->ls_recover_lock);
635 	set_bit(LSFL_RECOVER_STOP, &ls->ls_flags);
636 	new = test_and_clear_bit(LSFL_RUNNING, &ls->ls_flags);
637 	ls->ls_recover_seq++;
638 	spin_unlock(&ls->ls_recover_lock);
639 
640 	/*
641 	 * Let dlm_recv run again, now any normal messages will be saved on the
642 	 * requestqueue for later.
643 	 */
644 
645 	up_write(&ls->ls_recv_active);
646 
647 	/*
648 	 * This in_recovery lock does two things:
649 	 * 1) Keeps this function from returning until all threads are out
650 	 *    of locking routines and locking is truly stopped.
651 	 * 2) Keeps any new requests from being processed until it's unlocked
652 	 *    when recovery is complete.
653 	 */
654 
655 	if (new) {
656 		set_bit(LSFL_RECOVER_DOWN, &ls->ls_flags);
657 		wake_up_process(ls->ls_recoverd_task);
658 		wait_event(ls->ls_recover_lock_wait,
659 			   test_bit(LSFL_RECOVER_LOCK, &ls->ls_flags));
660 	}
661 
662 	/*
663 	 * The recoverd suspend/resume makes sure that dlm_recoverd (if
664 	 * running) has noticed RECOVER_STOP above and quit processing the
665 	 * previous recovery.
666 	 */
667 
668 	dlm_recoverd_suspend(ls);
669 
670 	spin_lock(&ls->ls_recover_lock);
671 	kfree(ls->ls_slots);
672 	ls->ls_slots = NULL;
673 	ls->ls_num_slots = 0;
674 	ls->ls_slots_size = 0;
675 	ls->ls_recover_status = 0;
676 	spin_unlock(&ls->ls_recover_lock);
677 
678 	dlm_recoverd_resume(ls);
679 
680 	if (!ls->ls_recover_begin)
681 		ls->ls_recover_begin = jiffies;
682 
683 	dlm_lsop_recover_prep(ls);
684 	return 0;
685 }
686 
687 int dlm_ls_start(struct dlm_ls *ls)
688 {
689 	struct dlm_recover *rv, *rv_old;
690 	struct dlm_config_node *nodes = NULL;
691 	int error, count;
692 
693 	rv = kzalloc(sizeof(*rv), GFP_NOFS);
694 	if (!rv)
695 		return -ENOMEM;
696 
697 	error = dlm_config_nodes(ls->ls_name, &nodes, &count);
698 	if (error < 0)
699 		goto fail_rv;
700 
701 	spin_lock(&ls->ls_recover_lock);
702 
703 	/* the lockspace needs to be stopped before it can be started */
704 
705 	if (!dlm_locking_stopped(ls)) {
706 		spin_unlock(&ls->ls_recover_lock);
707 		log_error(ls, "start ignored: lockspace running");
708 		error = -EINVAL;
709 		goto fail;
710 	}
711 
712 	rv->nodes = nodes;
713 	rv->nodes_count = count;
714 	rv->seq = ++ls->ls_recover_seq;
715 	rv_old = ls->ls_recover_args;
716 	ls->ls_recover_args = rv;
717 	spin_unlock(&ls->ls_recover_lock);
718 
719 	if (rv_old) {
720 		log_error(ls, "unused recovery %llx %d",
721 			  (unsigned long long)rv_old->seq, rv_old->nodes_count);
722 		kfree(rv_old->nodes);
723 		kfree(rv_old);
724 	}
725 
726 	set_bit(LSFL_RECOVER_WORK, &ls->ls_flags);
727 	wake_up_process(ls->ls_recoverd_task);
728 	return 0;
729 
730  fail:
731 	kfree(nodes);
732  fail_rv:
733 	kfree(rv);
734 	return error;
735 }
736 
737