xref: /openbmc/linux/fs/ocfs2/cluster/heartbeat.c (revision b04b4f78)
1 /* -*- mode: c; c-basic-offset: 8; -*-
2  * vim: noexpandtab sw=8 ts=8 sts=0:
3  *
4  * Copyright (C) 2004, 2005 Oracle.  All rights reserved.
5  *
6  * This program is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU General Public
8  * License as published by the Free Software Foundation; either
9  * version 2 of the License, or (at your option) any later version.
10  *
11  * This program is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public
17  * License along with this program; if not, write to the
18  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19  * Boston, MA 021110-1307, USA.
20  */
21 
22 #include <linux/kernel.h>
23 #include <linux/sched.h>
24 #include <linux/jiffies.h>
25 #include <linux/module.h>
26 #include <linux/fs.h>
27 #include <linux/bio.h>
28 #include <linux/blkdev.h>
29 #include <linux/delay.h>
30 #include <linux/file.h>
31 #include <linux/kthread.h>
32 #include <linux/configfs.h>
33 #include <linux/random.h>
34 #include <linux/crc32.h>
35 #include <linux/time.h>
36 #include <linux/debugfs.h>
37 
38 #include "heartbeat.h"
39 #include "tcp.h"
40 #include "nodemanager.h"
41 #include "quorum.h"
42 
43 #include "masklog.h"
44 
45 
46 /*
47  * The first heartbeat pass had one global thread that would serialize all hb
48  * callback calls.  This global serializing sem should only be removed once
49  * we've made sure that all callees can deal with being called concurrently
50  * from multiple hb region threads.
51  */
52 static DECLARE_RWSEM(o2hb_callback_sem);
53 
54 /*
55  * multiple hb threads are watching multiple regions.  A node is live
56  * whenever any of the threads sees activity from the node in its region.
57  */
58 static DEFINE_SPINLOCK(o2hb_live_lock);
59 static struct list_head o2hb_live_slots[O2NM_MAX_NODES];
60 static unsigned long o2hb_live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)];
61 static LIST_HEAD(o2hb_node_events);
62 static DECLARE_WAIT_QUEUE_HEAD(o2hb_steady_queue);
63 
64 #define O2HB_DEBUG_DIR			"o2hb"
65 #define O2HB_DEBUG_LIVENODES		"livenodes"
66 static struct dentry *o2hb_debug_dir;
67 static struct dentry *o2hb_debug_livenodes;
68 
69 static LIST_HEAD(o2hb_all_regions);
70 
71 static struct o2hb_callback {
72 	struct list_head list;
73 } o2hb_callbacks[O2HB_NUM_CB];
74 
75 static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type);
76 
77 #define O2HB_DEFAULT_BLOCK_BITS       9
78 
79 unsigned int o2hb_dead_threshold = O2HB_DEFAULT_DEAD_THRESHOLD;
80 
81 /* Only sets a new threshold if there are no active regions.
82  *
83  * No locking or otherwise interesting code is required for reading
84  * o2hb_dead_threshold as it can't change once regions are active and
85  * it's not interesting to anyone until then anyway. */
86 static void o2hb_dead_threshold_set(unsigned int threshold)
87 {
88 	if (threshold > O2HB_MIN_DEAD_THRESHOLD) {
89 		spin_lock(&o2hb_live_lock);
90 		if (list_empty(&o2hb_all_regions))
91 			o2hb_dead_threshold = threshold;
92 		spin_unlock(&o2hb_live_lock);
93 	}
94 }
95 
96 struct o2hb_node_event {
97 	struct list_head        hn_item;
98 	enum o2hb_callback_type hn_event_type;
99 	struct o2nm_node        *hn_node;
100 	int                     hn_node_num;
101 };
102 
103 struct o2hb_disk_slot {
104 	struct o2hb_disk_heartbeat_block *ds_raw_block;
105 	u8			ds_node_num;
106 	u64			ds_last_time;
107 	u64			ds_last_generation;
108 	u16			ds_equal_samples;
109 	u16			ds_changed_samples;
110 	struct list_head	ds_live_item;
111 };
112 
113 /* each thread owns a region.. when we're asked to tear down the region
114  * we ask the thread to stop, who cleans up the region */
115 struct o2hb_region {
116 	struct config_item	hr_item;
117 
118 	struct list_head	hr_all_item;
119 	unsigned		hr_unclean_stop:1;
120 
121 	/* protected by the hr_callback_sem */
122 	struct task_struct 	*hr_task;
123 
124 	unsigned int		hr_blocks;
125 	unsigned long long	hr_start_block;
126 
127 	unsigned int		hr_block_bits;
128 	unsigned int		hr_block_bytes;
129 
130 	unsigned int		hr_slots_per_page;
131 	unsigned int		hr_num_pages;
132 
133 	struct page             **hr_slot_data;
134 	struct block_device	*hr_bdev;
135 	struct o2hb_disk_slot	*hr_slots;
136 
137 	/* let the person setting up hb wait for it to return until it
138 	 * has reached a 'steady' state.  This will be fixed when we have
139 	 * a more complete api that doesn't lead to this sort of fragility. */
140 	atomic_t		hr_steady_iterations;
141 
142 	char			hr_dev_name[BDEVNAME_SIZE];
143 
144 	unsigned int		hr_timeout_ms;
145 
146 	/* randomized as the region goes up and down so that a node
147 	 * recognizes a node going up and down in one iteration */
148 	u64			hr_generation;
149 
150 	struct delayed_work	hr_write_timeout_work;
151 	unsigned long		hr_last_timeout_start;
152 
153 	/* Used during o2hb_check_slot to hold a copy of the block
154 	 * being checked because we temporarily have to zero out the
155 	 * crc field. */
156 	struct o2hb_disk_heartbeat_block *hr_tmp_block;
157 };
158 
159 struct o2hb_bio_wait_ctxt {
160 	atomic_t          wc_num_reqs;
161 	struct completion wc_io_complete;
162 	int               wc_error;
163 };
164 
165 static void o2hb_write_timeout(struct work_struct *work)
166 {
167 	struct o2hb_region *reg =
168 		container_of(work, struct o2hb_region,
169 			     hr_write_timeout_work.work);
170 
171 	mlog(ML_ERROR, "Heartbeat write timeout to device %s after %u "
172 	     "milliseconds\n", reg->hr_dev_name,
173 	     jiffies_to_msecs(jiffies - reg->hr_last_timeout_start));
174 	o2quo_disk_timeout();
175 }
176 
177 static void o2hb_arm_write_timeout(struct o2hb_region *reg)
178 {
179 	mlog(0, "Queue write timeout for %u ms\n", O2HB_MAX_WRITE_TIMEOUT_MS);
180 
181 	cancel_delayed_work(&reg->hr_write_timeout_work);
182 	reg->hr_last_timeout_start = jiffies;
183 	schedule_delayed_work(&reg->hr_write_timeout_work,
184 			      msecs_to_jiffies(O2HB_MAX_WRITE_TIMEOUT_MS));
185 }
186 
187 static void o2hb_disarm_write_timeout(struct o2hb_region *reg)
188 {
189 	cancel_delayed_work(&reg->hr_write_timeout_work);
190 	flush_scheduled_work();
191 }
192 
193 static inline void o2hb_bio_wait_init(struct o2hb_bio_wait_ctxt *wc)
194 {
195 	atomic_set(&wc->wc_num_reqs, 1);
196 	init_completion(&wc->wc_io_complete);
197 	wc->wc_error = 0;
198 }
199 
200 /* Used in error paths too */
201 static inline void o2hb_bio_wait_dec(struct o2hb_bio_wait_ctxt *wc,
202 				     unsigned int num)
203 {
204 	/* sadly atomic_sub_and_test() isn't available on all platforms.  The
205 	 * good news is that the fast path only completes one at a time */
206 	while(num--) {
207 		if (atomic_dec_and_test(&wc->wc_num_reqs)) {
208 			BUG_ON(num > 0);
209 			complete(&wc->wc_io_complete);
210 		}
211 	}
212 }
213 
214 static void o2hb_wait_on_io(struct o2hb_region *reg,
215 			    struct o2hb_bio_wait_ctxt *wc)
216 {
217 	struct address_space *mapping = reg->hr_bdev->bd_inode->i_mapping;
218 
219 	blk_run_address_space(mapping);
220 	o2hb_bio_wait_dec(wc, 1);
221 
222 	wait_for_completion(&wc->wc_io_complete);
223 }
224 
225 static void o2hb_bio_end_io(struct bio *bio,
226 			   int error)
227 {
228 	struct o2hb_bio_wait_ctxt *wc = bio->bi_private;
229 
230 	if (error) {
231 		mlog(ML_ERROR, "IO Error %d\n", error);
232 		wc->wc_error = error;
233 	}
234 
235 	o2hb_bio_wait_dec(wc, 1);
236 	bio_put(bio);
237 }
238 
239 /* Setup a Bio to cover I/O against num_slots slots starting at
240  * start_slot. */
241 static struct bio *o2hb_setup_one_bio(struct o2hb_region *reg,
242 				      struct o2hb_bio_wait_ctxt *wc,
243 				      unsigned int *current_slot,
244 				      unsigned int max_slots)
245 {
246 	int len, current_page;
247 	unsigned int vec_len, vec_start;
248 	unsigned int bits = reg->hr_block_bits;
249 	unsigned int spp = reg->hr_slots_per_page;
250 	unsigned int cs = *current_slot;
251 	struct bio *bio;
252 	struct page *page;
253 
254 	/* Testing has shown this allocation to take long enough under
255 	 * GFP_KERNEL that the local node can get fenced. It would be
256 	 * nicest if we could pre-allocate these bios and avoid this
257 	 * all together. */
258 	bio = bio_alloc(GFP_ATOMIC, 16);
259 	if (!bio) {
260 		mlog(ML_ERROR, "Could not alloc slots BIO!\n");
261 		bio = ERR_PTR(-ENOMEM);
262 		goto bail;
263 	}
264 
265 	/* Must put everything in 512 byte sectors for the bio... */
266 	bio->bi_sector = (reg->hr_start_block + cs) << (bits - 9);
267 	bio->bi_bdev = reg->hr_bdev;
268 	bio->bi_private = wc;
269 	bio->bi_end_io = o2hb_bio_end_io;
270 
271 	vec_start = (cs << bits) % PAGE_CACHE_SIZE;
272 	while(cs < max_slots) {
273 		current_page = cs / spp;
274 		page = reg->hr_slot_data[current_page];
275 
276 		vec_len = min(PAGE_CACHE_SIZE - vec_start,
277 			      (max_slots-cs) * (PAGE_CACHE_SIZE/spp) );
278 
279 		mlog(ML_HB_BIO, "page %d, vec_len = %u, vec_start = %u\n",
280 		     current_page, vec_len, vec_start);
281 
282 		len = bio_add_page(bio, page, vec_len, vec_start);
283 		if (len != vec_len) break;
284 
285 		cs += vec_len / (PAGE_CACHE_SIZE/spp);
286 		vec_start = 0;
287 	}
288 
289 bail:
290 	*current_slot = cs;
291 	return bio;
292 }
293 
294 static int o2hb_read_slots(struct o2hb_region *reg,
295 			   unsigned int max_slots)
296 {
297 	unsigned int current_slot=0;
298 	int status;
299 	struct o2hb_bio_wait_ctxt wc;
300 	struct bio *bio;
301 
302 	o2hb_bio_wait_init(&wc);
303 
304 	while(current_slot < max_slots) {
305 		bio = o2hb_setup_one_bio(reg, &wc, &current_slot, max_slots);
306 		if (IS_ERR(bio)) {
307 			status = PTR_ERR(bio);
308 			mlog_errno(status);
309 			goto bail_and_wait;
310 		}
311 
312 		atomic_inc(&wc.wc_num_reqs);
313 		submit_bio(READ, bio);
314 	}
315 
316 	status = 0;
317 
318 bail_and_wait:
319 	o2hb_wait_on_io(reg, &wc);
320 	if (wc.wc_error && !status)
321 		status = wc.wc_error;
322 
323 	return status;
324 }
325 
326 static int o2hb_issue_node_write(struct o2hb_region *reg,
327 				 struct o2hb_bio_wait_ctxt *write_wc)
328 {
329 	int status;
330 	unsigned int slot;
331 	struct bio *bio;
332 
333 	o2hb_bio_wait_init(write_wc);
334 
335 	slot = o2nm_this_node();
336 
337 	bio = o2hb_setup_one_bio(reg, write_wc, &slot, slot+1);
338 	if (IS_ERR(bio)) {
339 		status = PTR_ERR(bio);
340 		mlog_errno(status);
341 		goto bail;
342 	}
343 
344 	atomic_inc(&write_wc->wc_num_reqs);
345 	submit_bio(WRITE, bio);
346 
347 	status = 0;
348 bail:
349 	return status;
350 }
351 
352 static u32 o2hb_compute_block_crc_le(struct o2hb_region *reg,
353 				     struct o2hb_disk_heartbeat_block *hb_block)
354 {
355 	__le32 old_cksum;
356 	u32 ret;
357 
358 	/* We want to compute the block crc with a 0 value in the
359 	 * hb_cksum field. Save it off here and replace after the
360 	 * crc. */
361 	old_cksum = hb_block->hb_cksum;
362 	hb_block->hb_cksum = 0;
363 
364 	ret = crc32_le(0, (unsigned char *) hb_block, reg->hr_block_bytes);
365 
366 	hb_block->hb_cksum = old_cksum;
367 
368 	return ret;
369 }
370 
371 static void o2hb_dump_slot(struct o2hb_disk_heartbeat_block *hb_block)
372 {
373 	mlog(ML_ERROR, "Dump slot information: seq = 0x%llx, node = %u, "
374 	     "cksum = 0x%x, generation 0x%llx\n",
375 	     (long long)le64_to_cpu(hb_block->hb_seq),
376 	     hb_block->hb_node, le32_to_cpu(hb_block->hb_cksum),
377 	     (long long)le64_to_cpu(hb_block->hb_generation));
378 }
379 
380 static int o2hb_verify_crc(struct o2hb_region *reg,
381 			   struct o2hb_disk_heartbeat_block *hb_block)
382 {
383 	u32 read, computed;
384 
385 	read = le32_to_cpu(hb_block->hb_cksum);
386 	computed = o2hb_compute_block_crc_le(reg, hb_block);
387 
388 	return read == computed;
389 }
390 
391 /* We want to make sure that nobody is heartbeating on top of us --
392  * this will help detect an invalid configuration. */
393 static int o2hb_check_last_timestamp(struct o2hb_region *reg)
394 {
395 	int node_num, ret;
396 	struct o2hb_disk_slot *slot;
397 	struct o2hb_disk_heartbeat_block *hb_block;
398 
399 	node_num = o2nm_this_node();
400 
401 	ret = 1;
402 	slot = &reg->hr_slots[node_num];
403 	/* Don't check on our 1st timestamp */
404 	if (slot->ds_last_time) {
405 		hb_block = slot->ds_raw_block;
406 
407 		if (le64_to_cpu(hb_block->hb_seq) != slot->ds_last_time)
408 			ret = 0;
409 	}
410 
411 	return ret;
412 }
413 
414 static inline void o2hb_prepare_block(struct o2hb_region *reg,
415 				      u64 generation)
416 {
417 	int node_num;
418 	u64 cputime;
419 	struct o2hb_disk_slot *slot;
420 	struct o2hb_disk_heartbeat_block *hb_block;
421 
422 	node_num = o2nm_this_node();
423 	slot = &reg->hr_slots[node_num];
424 
425 	hb_block = (struct o2hb_disk_heartbeat_block *)slot->ds_raw_block;
426 	memset(hb_block, 0, reg->hr_block_bytes);
427 	/* TODO: time stuff */
428 	cputime = CURRENT_TIME.tv_sec;
429 	if (!cputime)
430 		cputime = 1;
431 
432 	hb_block->hb_seq = cpu_to_le64(cputime);
433 	hb_block->hb_node = node_num;
434 	hb_block->hb_generation = cpu_to_le64(generation);
435 	hb_block->hb_dead_ms = cpu_to_le32(o2hb_dead_threshold * O2HB_REGION_TIMEOUT_MS);
436 
437 	/* This step must always happen last! */
438 	hb_block->hb_cksum = cpu_to_le32(o2hb_compute_block_crc_le(reg,
439 								   hb_block));
440 
441 	mlog(ML_HB_BIO, "our node generation = 0x%llx, cksum = 0x%x\n",
442 	     (long long)generation,
443 	     le32_to_cpu(hb_block->hb_cksum));
444 }
445 
446 static void o2hb_fire_callbacks(struct o2hb_callback *hbcall,
447 				struct o2nm_node *node,
448 				int idx)
449 {
450 	struct list_head *iter;
451 	struct o2hb_callback_func *f;
452 
453 	list_for_each(iter, &hbcall->list) {
454 		f = list_entry(iter, struct o2hb_callback_func, hc_item);
455 		mlog(ML_HEARTBEAT, "calling funcs %p\n", f);
456 		(f->hc_func)(node, idx, f->hc_data);
457 	}
458 }
459 
460 /* Will run the list in order until we process the passed event */
461 static void o2hb_run_event_list(struct o2hb_node_event *queued_event)
462 {
463 	int empty;
464 	struct o2hb_callback *hbcall;
465 	struct o2hb_node_event *event;
466 
467 	spin_lock(&o2hb_live_lock);
468 	empty = list_empty(&queued_event->hn_item);
469 	spin_unlock(&o2hb_live_lock);
470 	if (empty)
471 		return;
472 
473 	/* Holding callback sem assures we don't alter the callback
474 	 * lists when doing this, and serializes ourselves with other
475 	 * processes wanting callbacks. */
476 	down_write(&o2hb_callback_sem);
477 
478 	spin_lock(&o2hb_live_lock);
479 	while (!list_empty(&o2hb_node_events)
480 	       && !list_empty(&queued_event->hn_item)) {
481 		event = list_entry(o2hb_node_events.next,
482 				   struct o2hb_node_event,
483 				   hn_item);
484 		list_del_init(&event->hn_item);
485 		spin_unlock(&o2hb_live_lock);
486 
487 		mlog(ML_HEARTBEAT, "Node %s event for %d\n",
488 		     event->hn_event_type == O2HB_NODE_UP_CB ? "UP" : "DOWN",
489 		     event->hn_node_num);
490 
491 		hbcall = hbcall_from_type(event->hn_event_type);
492 
493 		/* We should *never* have gotten on to the list with a
494 		 * bad type... This isn't something that we should try
495 		 * to recover from. */
496 		BUG_ON(IS_ERR(hbcall));
497 
498 		o2hb_fire_callbacks(hbcall, event->hn_node, event->hn_node_num);
499 
500 		spin_lock(&o2hb_live_lock);
501 	}
502 	spin_unlock(&o2hb_live_lock);
503 
504 	up_write(&o2hb_callback_sem);
505 }
506 
507 static void o2hb_queue_node_event(struct o2hb_node_event *event,
508 				  enum o2hb_callback_type type,
509 				  struct o2nm_node *node,
510 				  int node_num)
511 {
512 	assert_spin_locked(&o2hb_live_lock);
513 
514 	event->hn_event_type = type;
515 	event->hn_node = node;
516 	event->hn_node_num = node_num;
517 
518 	mlog(ML_HEARTBEAT, "Queue node %s event for node %d\n",
519 	     type == O2HB_NODE_UP_CB ? "UP" : "DOWN", node_num);
520 
521 	list_add_tail(&event->hn_item, &o2hb_node_events);
522 }
523 
524 static void o2hb_shutdown_slot(struct o2hb_disk_slot *slot)
525 {
526 	struct o2hb_node_event event =
527 		{ .hn_item = LIST_HEAD_INIT(event.hn_item), };
528 	struct o2nm_node *node;
529 
530 	node = o2nm_get_node_by_num(slot->ds_node_num);
531 	if (!node)
532 		return;
533 
534 	spin_lock(&o2hb_live_lock);
535 	if (!list_empty(&slot->ds_live_item)) {
536 		mlog(ML_HEARTBEAT, "Shutdown, node %d leaves region\n",
537 		     slot->ds_node_num);
538 
539 		list_del_init(&slot->ds_live_item);
540 
541 		if (list_empty(&o2hb_live_slots[slot->ds_node_num])) {
542 			clear_bit(slot->ds_node_num, o2hb_live_node_bitmap);
543 
544 			o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node,
545 					      slot->ds_node_num);
546 		}
547 	}
548 	spin_unlock(&o2hb_live_lock);
549 
550 	o2hb_run_event_list(&event);
551 
552 	o2nm_node_put(node);
553 }
554 
555 static int o2hb_check_slot(struct o2hb_region *reg,
556 			   struct o2hb_disk_slot *slot)
557 {
558 	int changed = 0, gen_changed = 0;
559 	struct o2hb_node_event event =
560 		{ .hn_item = LIST_HEAD_INIT(event.hn_item), };
561 	struct o2nm_node *node;
562 	struct o2hb_disk_heartbeat_block *hb_block = reg->hr_tmp_block;
563 	u64 cputime;
564 	unsigned int dead_ms = o2hb_dead_threshold * O2HB_REGION_TIMEOUT_MS;
565 	unsigned int slot_dead_ms;
566 
567 	memcpy(hb_block, slot->ds_raw_block, reg->hr_block_bytes);
568 
569 	/* Is this correct? Do we assume that the node doesn't exist
570 	 * if we're not configured for him? */
571 	node = o2nm_get_node_by_num(slot->ds_node_num);
572 	if (!node)
573 		return 0;
574 
575 	if (!o2hb_verify_crc(reg, hb_block)) {
576 		/* all paths from here will drop o2hb_live_lock for
577 		 * us. */
578 		spin_lock(&o2hb_live_lock);
579 
580 		/* Don't print an error on the console in this case -
581 		 * a freshly formatted heartbeat area will not have a
582 		 * crc set on it. */
583 		if (list_empty(&slot->ds_live_item))
584 			goto out;
585 
586 		/* The node is live but pushed out a bad crc. We
587 		 * consider it a transient miss but don't populate any
588 		 * other values as they may be junk. */
589 		mlog(ML_ERROR, "Node %d has written a bad crc to %s\n",
590 		     slot->ds_node_num, reg->hr_dev_name);
591 		o2hb_dump_slot(hb_block);
592 
593 		slot->ds_equal_samples++;
594 		goto fire_callbacks;
595 	}
596 
597 	/* we don't care if these wrap.. the state transitions below
598 	 * clear at the right places */
599 	cputime = le64_to_cpu(hb_block->hb_seq);
600 	if (slot->ds_last_time != cputime)
601 		slot->ds_changed_samples++;
602 	else
603 		slot->ds_equal_samples++;
604 	slot->ds_last_time = cputime;
605 
606 	/* The node changed heartbeat generations. We assume this to
607 	 * mean it dropped off but came back before we timed out. We
608 	 * want to consider it down for the time being but don't want
609 	 * to lose any changed_samples state we might build up to
610 	 * considering it live again. */
611 	if (slot->ds_last_generation != le64_to_cpu(hb_block->hb_generation)) {
612 		gen_changed = 1;
613 		slot->ds_equal_samples = 0;
614 		mlog(ML_HEARTBEAT, "Node %d changed generation (0x%llx "
615 		     "to 0x%llx)\n", slot->ds_node_num,
616 		     (long long)slot->ds_last_generation,
617 		     (long long)le64_to_cpu(hb_block->hb_generation));
618 	}
619 
620 	slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation);
621 
622 	mlog(ML_HEARTBEAT, "Slot %d gen 0x%llx cksum 0x%x "
623 	     "seq %llu last %llu changed %u equal %u\n",
624 	     slot->ds_node_num, (long long)slot->ds_last_generation,
625 	     le32_to_cpu(hb_block->hb_cksum),
626 	     (unsigned long long)le64_to_cpu(hb_block->hb_seq),
627 	     (unsigned long long)slot->ds_last_time, slot->ds_changed_samples,
628 	     slot->ds_equal_samples);
629 
630 	spin_lock(&o2hb_live_lock);
631 
632 fire_callbacks:
633 	/* dead nodes only come to life after some number of
634 	 * changes at any time during their dead time */
635 	if (list_empty(&slot->ds_live_item) &&
636 	    slot->ds_changed_samples >= O2HB_LIVE_THRESHOLD) {
637 		mlog(ML_HEARTBEAT, "Node %d (id 0x%llx) joined my region\n",
638 		     slot->ds_node_num, (long long)slot->ds_last_generation);
639 
640 		/* first on the list generates a callback */
641 		if (list_empty(&o2hb_live_slots[slot->ds_node_num])) {
642 			set_bit(slot->ds_node_num, o2hb_live_node_bitmap);
643 
644 			o2hb_queue_node_event(&event, O2HB_NODE_UP_CB, node,
645 					      slot->ds_node_num);
646 
647 			changed = 1;
648 		}
649 
650 		list_add_tail(&slot->ds_live_item,
651 			      &o2hb_live_slots[slot->ds_node_num]);
652 
653 		slot->ds_equal_samples = 0;
654 
655 		/* We want to be sure that all nodes agree on the
656 		 * number of milliseconds before a node will be
657 		 * considered dead. The self-fencing timeout is
658 		 * computed from this value, and a discrepancy might
659 		 * result in heartbeat calling a node dead when it
660 		 * hasn't self-fenced yet. */
661 		slot_dead_ms = le32_to_cpu(hb_block->hb_dead_ms);
662 		if (slot_dead_ms && slot_dead_ms != dead_ms) {
663 			/* TODO: Perhaps we can fail the region here. */
664 			mlog(ML_ERROR, "Node %d on device %s has a dead count "
665 			     "of %u ms, but our count is %u ms.\n"
666 			     "Please double check your configuration values "
667 			     "for 'O2CB_HEARTBEAT_THRESHOLD'\n",
668 			     slot->ds_node_num, reg->hr_dev_name, slot_dead_ms,
669 			     dead_ms);
670 		}
671 		goto out;
672 	}
673 
674 	/* if the list is dead, we're done.. */
675 	if (list_empty(&slot->ds_live_item))
676 		goto out;
677 
678 	/* live nodes only go dead after enough consequtive missed
679 	 * samples..  reset the missed counter whenever we see
680 	 * activity */
681 	if (slot->ds_equal_samples >= o2hb_dead_threshold || gen_changed) {
682 		mlog(ML_HEARTBEAT, "Node %d left my region\n",
683 		     slot->ds_node_num);
684 
685 		/* last off the live_slot generates a callback */
686 		list_del_init(&slot->ds_live_item);
687 		if (list_empty(&o2hb_live_slots[slot->ds_node_num])) {
688 			clear_bit(slot->ds_node_num, o2hb_live_node_bitmap);
689 
690 			o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node,
691 					      slot->ds_node_num);
692 
693 			changed = 1;
694 		}
695 
696 		/* We don't clear this because the node is still
697 		 * actually writing new blocks. */
698 		if (!gen_changed)
699 			slot->ds_changed_samples = 0;
700 		goto out;
701 	}
702 	if (slot->ds_changed_samples) {
703 		slot->ds_changed_samples = 0;
704 		slot->ds_equal_samples = 0;
705 	}
706 out:
707 	spin_unlock(&o2hb_live_lock);
708 
709 	o2hb_run_event_list(&event);
710 
711 	o2nm_node_put(node);
712 	return changed;
713 }
714 
715 /* This could be faster if we just implmented a find_last_bit, but I
716  * don't think the circumstances warrant it. */
717 static int o2hb_highest_node(unsigned long *nodes,
718 			     int numbits)
719 {
720 	int highest, node;
721 
722 	highest = numbits;
723 	node = -1;
724 	while ((node = find_next_bit(nodes, numbits, node + 1)) != -1) {
725 		if (node >= numbits)
726 			break;
727 
728 		highest = node;
729 	}
730 
731 	return highest;
732 }
733 
734 static int o2hb_do_disk_heartbeat(struct o2hb_region *reg)
735 {
736 	int i, ret, highest_node, change = 0;
737 	unsigned long configured_nodes[BITS_TO_LONGS(O2NM_MAX_NODES)];
738 	struct o2hb_bio_wait_ctxt write_wc;
739 
740 	ret = o2nm_configured_node_map(configured_nodes,
741 				       sizeof(configured_nodes));
742 	if (ret) {
743 		mlog_errno(ret);
744 		return ret;
745 	}
746 
747 	highest_node = o2hb_highest_node(configured_nodes, O2NM_MAX_NODES);
748 	if (highest_node >= O2NM_MAX_NODES) {
749 		mlog(ML_NOTICE, "ocfs2_heartbeat: no configured nodes found!\n");
750 		return -EINVAL;
751 	}
752 
753 	/* No sense in reading the slots of nodes that don't exist
754 	 * yet. Of course, if the node definitions have holes in them
755 	 * then we're reading an empty slot anyway... Consider this
756 	 * best-effort. */
757 	ret = o2hb_read_slots(reg, highest_node + 1);
758 	if (ret < 0) {
759 		mlog_errno(ret);
760 		return ret;
761 	}
762 
763 	/* With an up to date view of the slots, we can check that no
764 	 * other node has been improperly configured to heartbeat in
765 	 * our slot. */
766 	if (!o2hb_check_last_timestamp(reg))
767 		mlog(ML_ERROR, "Device \"%s\": another node is heartbeating "
768 		     "in our slot!\n", reg->hr_dev_name);
769 
770 	/* fill in the proper info for our next heartbeat */
771 	o2hb_prepare_block(reg, reg->hr_generation);
772 
773 	/* And fire off the write. Note that we don't wait on this I/O
774 	 * until later. */
775 	ret = o2hb_issue_node_write(reg, &write_wc);
776 	if (ret < 0) {
777 		mlog_errno(ret);
778 		return ret;
779 	}
780 
781 	i = -1;
782 	while((i = find_next_bit(configured_nodes, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) {
783 
784 		change |= o2hb_check_slot(reg, &reg->hr_slots[i]);
785 	}
786 
787 	/*
788 	 * We have to be sure we've advertised ourselves on disk
789 	 * before we can go to steady state.  This ensures that
790 	 * people we find in our steady state have seen us.
791 	 */
792 	o2hb_wait_on_io(reg, &write_wc);
793 	if (write_wc.wc_error) {
794 		/* Do not re-arm the write timeout on I/O error - we
795 		 * can't be sure that the new block ever made it to
796 		 * disk */
797 		mlog(ML_ERROR, "Write error %d on device \"%s\"\n",
798 		     write_wc.wc_error, reg->hr_dev_name);
799 		return write_wc.wc_error;
800 	}
801 
802 	o2hb_arm_write_timeout(reg);
803 
804 	/* let the person who launched us know when things are steady */
805 	if (!change && (atomic_read(&reg->hr_steady_iterations) != 0)) {
806 		if (atomic_dec_and_test(&reg->hr_steady_iterations))
807 			wake_up(&o2hb_steady_queue);
808 	}
809 
810 	return 0;
811 }
812 
813 /* Subtract b from a, storing the result in a. a *must* have a larger
814  * value than b. */
815 static void o2hb_tv_subtract(struct timeval *a,
816 			     struct timeval *b)
817 {
818 	/* just return 0 when a is after b */
819 	if (a->tv_sec < b->tv_sec ||
820 	    (a->tv_sec == b->tv_sec && a->tv_usec < b->tv_usec)) {
821 		a->tv_sec = 0;
822 		a->tv_usec = 0;
823 		return;
824 	}
825 
826 	a->tv_sec -= b->tv_sec;
827 	a->tv_usec -= b->tv_usec;
828 	while ( a->tv_usec < 0 ) {
829 		a->tv_sec--;
830 		a->tv_usec += 1000000;
831 	}
832 }
833 
834 static unsigned int o2hb_elapsed_msecs(struct timeval *start,
835 				       struct timeval *end)
836 {
837 	struct timeval res = *end;
838 
839 	o2hb_tv_subtract(&res, start);
840 
841 	return res.tv_sec * 1000 + res.tv_usec / 1000;
842 }
843 
844 /*
845  * we ride the region ref that the region dir holds.  before the region
846  * dir is removed and drops it ref it will wait to tear down this
847  * thread.
848  */
849 static int o2hb_thread(void *data)
850 {
851 	int i, ret;
852 	struct o2hb_region *reg = data;
853 	struct o2hb_bio_wait_ctxt write_wc;
854 	struct timeval before_hb, after_hb;
855 	unsigned int elapsed_msec;
856 
857 	mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread running\n");
858 
859 	set_user_nice(current, -20);
860 
861 	while (!kthread_should_stop() && !reg->hr_unclean_stop) {
862 		/* We track the time spent inside
863 		 * o2hb_do_disk_heartbeat so that we avoid more than
864 		 * hr_timeout_ms between disk writes. On busy systems
865 		 * this should result in a heartbeat which is less
866 		 * likely to time itself out. */
867 		do_gettimeofday(&before_hb);
868 
869 		i = 0;
870 		do {
871 			ret = o2hb_do_disk_heartbeat(reg);
872 		} while (ret && ++i < 2);
873 
874 		do_gettimeofday(&after_hb);
875 		elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb);
876 
877 		mlog(0, "start = %lu.%lu, end = %lu.%lu, msec = %u\n",
878 		     before_hb.tv_sec, (unsigned long) before_hb.tv_usec,
879 		     after_hb.tv_sec, (unsigned long) after_hb.tv_usec,
880 		     elapsed_msec);
881 
882 		if (elapsed_msec < reg->hr_timeout_ms) {
883 			/* the kthread api has blocked signals for us so no
884 			 * need to record the return value. */
885 			msleep_interruptible(reg->hr_timeout_ms - elapsed_msec);
886 		}
887 	}
888 
889 	o2hb_disarm_write_timeout(reg);
890 
891 	/* unclean stop is only used in very bad situation */
892 	for(i = 0; !reg->hr_unclean_stop && i < reg->hr_blocks; i++)
893 		o2hb_shutdown_slot(&reg->hr_slots[i]);
894 
895 	/* Explicit down notification - avoid forcing the other nodes
896 	 * to timeout on this region when we could just as easily
897 	 * write a clear generation - thus indicating to them that
898 	 * this node has left this region.
899 	 *
900 	 * XXX: Should we skip this on unclean_stop? */
901 	o2hb_prepare_block(reg, 0);
902 	ret = o2hb_issue_node_write(reg, &write_wc);
903 	if (ret == 0) {
904 		o2hb_wait_on_io(reg, &write_wc);
905 	} else {
906 		mlog_errno(ret);
907 	}
908 
909 	mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread exiting\n");
910 
911 	return 0;
912 }
913 
914 #ifdef CONFIG_DEBUG_FS
915 static int o2hb_debug_open(struct inode *inode, struct file *file)
916 {
917 	unsigned long map[BITS_TO_LONGS(O2NM_MAX_NODES)];
918 	char *buf = NULL;
919 	int i = -1;
920 	int out = 0;
921 
922 	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
923 	if (!buf)
924 		goto bail;
925 
926 	o2hb_fill_node_map(map, sizeof(map));
927 
928 	while ((i = find_next_bit(map, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES)
929 		out += snprintf(buf + out, PAGE_SIZE - out, "%d ", i);
930 	out += snprintf(buf + out, PAGE_SIZE - out, "\n");
931 
932 	i_size_write(inode, out);
933 
934 	file->private_data = buf;
935 
936 	return 0;
937 bail:
938 	return -ENOMEM;
939 }
940 
941 static int o2hb_debug_release(struct inode *inode, struct file *file)
942 {
943 	kfree(file->private_data);
944 	return 0;
945 }
946 
947 static ssize_t o2hb_debug_read(struct file *file, char __user *buf,
948 				 size_t nbytes, loff_t *ppos)
949 {
950 	return simple_read_from_buffer(buf, nbytes, ppos, file->private_data,
951 				       i_size_read(file->f_mapping->host));
952 }
953 #else
954 static int o2hb_debug_open(struct inode *inode, struct file *file)
955 {
956 	return 0;
957 }
958 static int o2hb_debug_release(struct inode *inode, struct file *file)
959 {
960 	return 0;
961 }
962 static ssize_t o2hb_debug_read(struct file *file, char __user *buf,
963 			       size_t nbytes, loff_t *ppos)
964 {
965 	return 0;
966 }
967 #endif  /* CONFIG_DEBUG_FS */
968 
969 static struct file_operations o2hb_debug_fops = {
970 	.open =		o2hb_debug_open,
971 	.release =	o2hb_debug_release,
972 	.read =		o2hb_debug_read,
973 	.llseek =	generic_file_llseek,
974 };
975 
976 void o2hb_exit(void)
977 {
978 	if (o2hb_debug_livenodes)
979 		debugfs_remove(o2hb_debug_livenodes);
980 	if (o2hb_debug_dir)
981 		debugfs_remove(o2hb_debug_dir);
982 }
983 
984 int o2hb_init(void)
985 {
986 	int i;
987 
988 	for (i = 0; i < ARRAY_SIZE(o2hb_callbacks); i++)
989 		INIT_LIST_HEAD(&o2hb_callbacks[i].list);
990 
991 	for (i = 0; i < ARRAY_SIZE(o2hb_live_slots); i++)
992 		INIT_LIST_HEAD(&o2hb_live_slots[i]);
993 
994 	INIT_LIST_HEAD(&o2hb_node_events);
995 
996 	memset(o2hb_live_node_bitmap, 0, sizeof(o2hb_live_node_bitmap));
997 
998 	o2hb_debug_dir = debugfs_create_dir(O2HB_DEBUG_DIR, NULL);
999 	if (!o2hb_debug_dir) {
1000 		mlog_errno(-ENOMEM);
1001 		return -ENOMEM;
1002 	}
1003 
1004 	o2hb_debug_livenodes = debugfs_create_file(O2HB_DEBUG_LIVENODES,
1005 						   S_IFREG|S_IRUSR,
1006 						   o2hb_debug_dir, NULL,
1007 						   &o2hb_debug_fops);
1008 	if (!o2hb_debug_livenodes) {
1009 		mlog_errno(-ENOMEM);
1010 		debugfs_remove(o2hb_debug_dir);
1011 		return -ENOMEM;
1012 	}
1013 
1014 	return 0;
1015 }
1016 
1017 /* if we're already in a callback then we're already serialized by the sem */
1018 static void o2hb_fill_node_map_from_callback(unsigned long *map,
1019 					     unsigned bytes)
1020 {
1021 	BUG_ON(bytes < (BITS_TO_LONGS(O2NM_MAX_NODES) * sizeof(unsigned long)));
1022 
1023 	memcpy(map, &o2hb_live_node_bitmap, bytes);
1024 }
1025 
1026 /*
1027  * get a map of all nodes that are heartbeating in any regions
1028  */
1029 void o2hb_fill_node_map(unsigned long *map, unsigned bytes)
1030 {
1031 	/* callers want to serialize this map and callbacks so that they
1032 	 * can trust that they don't miss nodes coming to the party */
1033 	down_read(&o2hb_callback_sem);
1034 	spin_lock(&o2hb_live_lock);
1035 	o2hb_fill_node_map_from_callback(map, bytes);
1036 	spin_unlock(&o2hb_live_lock);
1037 	up_read(&o2hb_callback_sem);
1038 }
1039 EXPORT_SYMBOL_GPL(o2hb_fill_node_map);
1040 
1041 /*
1042  * heartbeat configfs bits.  The heartbeat set is a default set under
1043  * the cluster set in nodemanager.c.
1044  */
1045 
1046 static struct o2hb_region *to_o2hb_region(struct config_item *item)
1047 {
1048 	return item ? container_of(item, struct o2hb_region, hr_item) : NULL;
1049 }
1050 
1051 /* drop_item only drops its ref after killing the thread, nothing should
1052  * be using the region anymore.  this has to clean up any state that
1053  * attributes might have built up. */
1054 static void o2hb_region_release(struct config_item *item)
1055 {
1056 	int i;
1057 	struct page *page;
1058 	struct o2hb_region *reg = to_o2hb_region(item);
1059 
1060 	if (reg->hr_tmp_block)
1061 		kfree(reg->hr_tmp_block);
1062 
1063 	if (reg->hr_slot_data) {
1064 		for (i = 0; i < reg->hr_num_pages; i++) {
1065 			page = reg->hr_slot_data[i];
1066 			if (page)
1067 				__free_page(page);
1068 		}
1069 		kfree(reg->hr_slot_data);
1070 	}
1071 
1072 	if (reg->hr_bdev)
1073 		blkdev_put(reg->hr_bdev, FMODE_READ|FMODE_WRITE);
1074 
1075 	if (reg->hr_slots)
1076 		kfree(reg->hr_slots);
1077 
1078 	spin_lock(&o2hb_live_lock);
1079 	list_del(&reg->hr_all_item);
1080 	spin_unlock(&o2hb_live_lock);
1081 
1082 	kfree(reg);
1083 }
1084 
1085 static int o2hb_read_block_input(struct o2hb_region *reg,
1086 				 const char *page,
1087 				 size_t count,
1088 				 unsigned long *ret_bytes,
1089 				 unsigned int *ret_bits)
1090 {
1091 	unsigned long bytes;
1092 	char *p = (char *)page;
1093 
1094 	bytes = simple_strtoul(p, &p, 0);
1095 	if (!p || (*p && (*p != '\n')))
1096 		return -EINVAL;
1097 
1098 	/* Heartbeat and fs min / max block sizes are the same. */
1099 	if (bytes > 4096 || bytes < 512)
1100 		return -ERANGE;
1101 	if (hweight16(bytes) != 1)
1102 		return -EINVAL;
1103 
1104 	if (ret_bytes)
1105 		*ret_bytes = bytes;
1106 	if (ret_bits)
1107 		*ret_bits = ffs(bytes) - 1;
1108 
1109 	return 0;
1110 }
1111 
1112 static ssize_t o2hb_region_block_bytes_read(struct o2hb_region *reg,
1113 					    char *page)
1114 {
1115 	return sprintf(page, "%u\n", reg->hr_block_bytes);
1116 }
1117 
1118 static ssize_t o2hb_region_block_bytes_write(struct o2hb_region *reg,
1119 					     const char *page,
1120 					     size_t count)
1121 {
1122 	int status;
1123 	unsigned long block_bytes;
1124 	unsigned int block_bits;
1125 
1126 	if (reg->hr_bdev)
1127 		return -EINVAL;
1128 
1129 	status = o2hb_read_block_input(reg, page, count,
1130 				       &block_bytes, &block_bits);
1131 	if (status)
1132 		return status;
1133 
1134 	reg->hr_block_bytes = (unsigned int)block_bytes;
1135 	reg->hr_block_bits = block_bits;
1136 
1137 	return count;
1138 }
1139 
1140 static ssize_t o2hb_region_start_block_read(struct o2hb_region *reg,
1141 					    char *page)
1142 {
1143 	return sprintf(page, "%llu\n", reg->hr_start_block);
1144 }
1145 
1146 static ssize_t o2hb_region_start_block_write(struct o2hb_region *reg,
1147 					     const char *page,
1148 					     size_t count)
1149 {
1150 	unsigned long long tmp;
1151 	char *p = (char *)page;
1152 
1153 	if (reg->hr_bdev)
1154 		return -EINVAL;
1155 
1156 	tmp = simple_strtoull(p, &p, 0);
1157 	if (!p || (*p && (*p != '\n')))
1158 		return -EINVAL;
1159 
1160 	reg->hr_start_block = tmp;
1161 
1162 	return count;
1163 }
1164 
1165 static ssize_t o2hb_region_blocks_read(struct o2hb_region *reg,
1166 				       char *page)
1167 {
1168 	return sprintf(page, "%d\n", reg->hr_blocks);
1169 }
1170 
1171 static ssize_t o2hb_region_blocks_write(struct o2hb_region *reg,
1172 					const char *page,
1173 					size_t count)
1174 {
1175 	unsigned long tmp;
1176 	char *p = (char *)page;
1177 
1178 	if (reg->hr_bdev)
1179 		return -EINVAL;
1180 
1181 	tmp = simple_strtoul(p, &p, 0);
1182 	if (!p || (*p && (*p != '\n')))
1183 		return -EINVAL;
1184 
1185 	if (tmp > O2NM_MAX_NODES || tmp == 0)
1186 		return -ERANGE;
1187 
1188 	reg->hr_blocks = (unsigned int)tmp;
1189 
1190 	return count;
1191 }
1192 
1193 static ssize_t o2hb_region_dev_read(struct o2hb_region *reg,
1194 				    char *page)
1195 {
1196 	unsigned int ret = 0;
1197 
1198 	if (reg->hr_bdev)
1199 		ret = sprintf(page, "%s\n", reg->hr_dev_name);
1200 
1201 	return ret;
1202 }
1203 
1204 static void o2hb_init_region_params(struct o2hb_region *reg)
1205 {
1206 	reg->hr_slots_per_page = PAGE_CACHE_SIZE >> reg->hr_block_bits;
1207 	reg->hr_timeout_ms = O2HB_REGION_TIMEOUT_MS;
1208 
1209 	mlog(ML_HEARTBEAT, "hr_start_block = %llu, hr_blocks = %u\n",
1210 	     reg->hr_start_block, reg->hr_blocks);
1211 	mlog(ML_HEARTBEAT, "hr_block_bytes = %u, hr_block_bits = %u\n",
1212 	     reg->hr_block_bytes, reg->hr_block_bits);
1213 	mlog(ML_HEARTBEAT, "hr_timeout_ms = %u\n", reg->hr_timeout_ms);
1214 	mlog(ML_HEARTBEAT, "dead threshold = %u\n", o2hb_dead_threshold);
1215 }
1216 
1217 static int o2hb_map_slot_data(struct o2hb_region *reg)
1218 {
1219 	int i, j;
1220 	unsigned int last_slot;
1221 	unsigned int spp = reg->hr_slots_per_page;
1222 	struct page *page;
1223 	char *raw;
1224 	struct o2hb_disk_slot *slot;
1225 
1226 	reg->hr_tmp_block = kmalloc(reg->hr_block_bytes, GFP_KERNEL);
1227 	if (reg->hr_tmp_block == NULL) {
1228 		mlog_errno(-ENOMEM);
1229 		return -ENOMEM;
1230 	}
1231 
1232 	reg->hr_slots = kcalloc(reg->hr_blocks,
1233 				sizeof(struct o2hb_disk_slot), GFP_KERNEL);
1234 	if (reg->hr_slots == NULL) {
1235 		mlog_errno(-ENOMEM);
1236 		return -ENOMEM;
1237 	}
1238 
1239 	for(i = 0; i < reg->hr_blocks; i++) {
1240 		slot = &reg->hr_slots[i];
1241 		slot->ds_node_num = i;
1242 		INIT_LIST_HEAD(&slot->ds_live_item);
1243 		slot->ds_raw_block = NULL;
1244 	}
1245 
1246 	reg->hr_num_pages = (reg->hr_blocks + spp - 1) / spp;
1247 	mlog(ML_HEARTBEAT, "Going to require %u pages to cover %u blocks "
1248 			   "at %u blocks per page\n",
1249 	     reg->hr_num_pages, reg->hr_blocks, spp);
1250 
1251 	reg->hr_slot_data = kcalloc(reg->hr_num_pages, sizeof(struct page *),
1252 				    GFP_KERNEL);
1253 	if (!reg->hr_slot_data) {
1254 		mlog_errno(-ENOMEM);
1255 		return -ENOMEM;
1256 	}
1257 
1258 	for(i = 0; i < reg->hr_num_pages; i++) {
1259 		page = alloc_page(GFP_KERNEL);
1260 		if (!page) {
1261 			mlog_errno(-ENOMEM);
1262 			return -ENOMEM;
1263 		}
1264 
1265 		reg->hr_slot_data[i] = page;
1266 
1267 		last_slot = i * spp;
1268 		raw = page_address(page);
1269 		for (j = 0;
1270 		     (j < spp) && ((j + last_slot) < reg->hr_blocks);
1271 		     j++) {
1272 			BUG_ON((j + last_slot) >= reg->hr_blocks);
1273 
1274 			slot = &reg->hr_slots[j + last_slot];
1275 			slot->ds_raw_block =
1276 				(struct o2hb_disk_heartbeat_block *) raw;
1277 
1278 			raw += reg->hr_block_bytes;
1279 		}
1280 	}
1281 
1282 	return 0;
1283 }
1284 
1285 /* Read in all the slots available and populate the tracking
1286  * structures so that we can start with a baseline idea of what's
1287  * there. */
1288 static int o2hb_populate_slot_data(struct o2hb_region *reg)
1289 {
1290 	int ret, i;
1291 	struct o2hb_disk_slot *slot;
1292 	struct o2hb_disk_heartbeat_block *hb_block;
1293 
1294 	mlog_entry_void();
1295 
1296 	ret = o2hb_read_slots(reg, reg->hr_blocks);
1297 	if (ret) {
1298 		mlog_errno(ret);
1299 		goto out;
1300 	}
1301 
1302 	/* We only want to get an idea of the values initially in each
1303 	 * slot, so we do no verification - o2hb_check_slot will
1304 	 * actually determine if each configured slot is valid and
1305 	 * whether any values have changed. */
1306 	for(i = 0; i < reg->hr_blocks; i++) {
1307 		slot = &reg->hr_slots[i];
1308 		hb_block = (struct o2hb_disk_heartbeat_block *) slot->ds_raw_block;
1309 
1310 		/* Only fill the values that o2hb_check_slot uses to
1311 		 * determine changing slots */
1312 		slot->ds_last_time = le64_to_cpu(hb_block->hb_seq);
1313 		slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation);
1314 	}
1315 
1316 out:
1317 	mlog_exit(ret);
1318 	return ret;
1319 }
1320 
1321 /* this is acting as commit; we set up all of hr_bdev and hr_task or nothing */
1322 static ssize_t o2hb_region_dev_write(struct o2hb_region *reg,
1323 				     const char *page,
1324 				     size_t count)
1325 {
1326 	struct task_struct *hb_task;
1327 	long fd;
1328 	int sectsize;
1329 	char *p = (char *)page;
1330 	struct file *filp = NULL;
1331 	struct inode *inode = NULL;
1332 	ssize_t ret = -EINVAL;
1333 
1334 	if (reg->hr_bdev)
1335 		goto out;
1336 
1337 	/* We can't heartbeat without having had our node number
1338 	 * configured yet. */
1339 	if (o2nm_this_node() == O2NM_MAX_NODES)
1340 		goto out;
1341 
1342 	fd = simple_strtol(p, &p, 0);
1343 	if (!p || (*p && (*p != '\n')))
1344 		goto out;
1345 
1346 	if (fd < 0 || fd >= INT_MAX)
1347 		goto out;
1348 
1349 	filp = fget(fd);
1350 	if (filp == NULL)
1351 		goto out;
1352 
1353 	if (reg->hr_blocks == 0 || reg->hr_start_block == 0 ||
1354 	    reg->hr_block_bytes == 0)
1355 		goto out;
1356 
1357 	inode = igrab(filp->f_mapping->host);
1358 	if (inode == NULL)
1359 		goto out;
1360 
1361 	if (!S_ISBLK(inode->i_mode))
1362 		goto out;
1363 
1364 	reg->hr_bdev = I_BDEV(filp->f_mapping->host);
1365 	ret = blkdev_get(reg->hr_bdev, FMODE_WRITE | FMODE_READ);
1366 	if (ret) {
1367 		reg->hr_bdev = NULL;
1368 		goto out;
1369 	}
1370 	inode = NULL;
1371 
1372 	bdevname(reg->hr_bdev, reg->hr_dev_name);
1373 
1374 	sectsize = bdev_hardsect_size(reg->hr_bdev);
1375 	if (sectsize != reg->hr_block_bytes) {
1376 		mlog(ML_ERROR,
1377 		     "blocksize %u incorrect for device, expected %d",
1378 		     reg->hr_block_bytes, sectsize);
1379 		ret = -EINVAL;
1380 		goto out;
1381 	}
1382 
1383 	o2hb_init_region_params(reg);
1384 
1385 	/* Generation of zero is invalid */
1386 	do {
1387 		get_random_bytes(&reg->hr_generation,
1388 				 sizeof(reg->hr_generation));
1389 	} while (reg->hr_generation == 0);
1390 
1391 	ret = o2hb_map_slot_data(reg);
1392 	if (ret) {
1393 		mlog_errno(ret);
1394 		goto out;
1395 	}
1396 
1397 	ret = o2hb_populate_slot_data(reg);
1398 	if (ret) {
1399 		mlog_errno(ret);
1400 		goto out;
1401 	}
1402 
1403 	INIT_DELAYED_WORK(&reg->hr_write_timeout_work, o2hb_write_timeout);
1404 
1405 	/*
1406 	 * A node is considered live after it has beat LIVE_THRESHOLD
1407 	 * times.  We're not steady until we've given them a chance
1408 	 * _after_ our first read.
1409 	 */
1410 	atomic_set(&reg->hr_steady_iterations, O2HB_LIVE_THRESHOLD + 1);
1411 
1412 	hb_task = kthread_run(o2hb_thread, reg, "o2hb-%s",
1413 			      reg->hr_item.ci_name);
1414 	if (IS_ERR(hb_task)) {
1415 		ret = PTR_ERR(hb_task);
1416 		mlog_errno(ret);
1417 		goto out;
1418 	}
1419 
1420 	spin_lock(&o2hb_live_lock);
1421 	reg->hr_task = hb_task;
1422 	spin_unlock(&o2hb_live_lock);
1423 
1424 	ret = wait_event_interruptible(o2hb_steady_queue,
1425 				atomic_read(&reg->hr_steady_iterations) == 0);
1426 	if (ret) {
1427 		/* We got interrupted (hello ptrace!).  Clean up */
1428 		spin_lock(&o2hb_live_lock);
1429 		hb_task = reg->hr_task;
1430 		reg->hr_task = NULL;
1431 		spin_unlock(&o2hb_live_lock);
1432 
1433 		if (hb_task)
1434 			kthread_stop(hb_task);
1435 		goto out;
1436 	}
1437 
1438 	/* Ok, we were woken.  Make sure it wasn't by drop_item() */
1439 	spin_lock(&o2hb_live_lock);
1440 	hb_task = reg->hr_task;
1441 	spin_unlock(&o2hb_live_lock);
1442 
1443 	if (hb_task)
1444 		ret = count;
1445 	else
1446 		ret = -EIO;
1447 
1448 out:
1449 	if (filp)
1450 		fput(filp);
1451 	if (inode)
1452 		iput(inode);
1453 	if (ret < 0) {
1454 		if (reg->hr_bdev) {
1455 			blkdev_put(reg->hr_bdev, FMODE_READ|FMODE_WRITE);
1456 			reg->hr_bdev = NULL;
1457 		}
1458 	}
1459 	return ret;
1460 }
1461 
1462 static ssize_t o2hb_region_pid_read(struct o2hb_region *reg,
1463                                       char *page)
1464 {
1465 	pid_t pid = 0;
1466 
1467 	spin_lock(&o2hb_live_lock);
1468 	if (reg->hr_task)
1469 		pid = task_pid_nr(reg->hr_task);
1470 	spin_unlock(&o2hb_live_lock);
1471 
1472 	if (!pid)
1473 		return 0;
1474 
1475 	return sprintf(page, "%u\n", pid);
1476 }
1477 
1478 struct o2hb_region_attribute {
1479 	struct configfs_attribute attr;
1480 	ssize_t (*show)(struct o2hb_region *, char *);
1481 	ssize_t (*store)(struct o2hb_region *, const char *, size_t);
1482 };
1483 
1484 static struct o2hb_region_attribute o2hb_region_attr_block_bytes = {
1485 	.attr	= { .ca_owner = THIS_MODULE,
1486 		    .ca_name = "block_bytes",
1487 		    .ca_mode = S_IRUGO | S_IWUSR },
1488 	.show	= o2hb_region_block_bytes_read,
1489 	.store	= o2hb_region_block_bytes_write,
1490 };
1491 
1492 static struct o2hb_region_attribute o2hb_region_attr_start_block = {
1493 	.attr	= { .ca_owner = THIS_MODULE,
1494 		    .ca_name = "start_block",
1495 		    .ca_mode = S_IRUGO | S_IWUSR },
1496 	.show	= o2hb_region_start_block_read,
1497 	.store	= o2hb_region_start_block_write,
1498 };
1499 
1500 static struct o2hb_region_attribute o2hb_region_attr_blocks = {
1501 	.attr	= { .ca_owner = THIS_MODULE,
1502 		    .ca_name = "blocks",
1503 		    .ca_mode = S_IRUGO | S_IWUSR },
1504 	.show	= o2hb_region_blocks_read,
1505 	.store	= o2hb_region_blocks_write,
1506 };
1507 
1508 static struct o2hb_region_attribute o2hb_region_attr_dev = {
1509 	.attr	= { .ca_owner = THIS_MODULE,
1510 		    .ca_name = "dev",
1511 		    .ca_mode = S_IRUGO | S_IWUSR },
1512 	.show	= o2hb_region_dev_read,
1513 	.store	= o2hb_region_dev_write,
1514 };
1515 
1516 static struct o2hb_region_attribute o2hb_region_attr_pid = {
1517        .attr   = { .ca_owner = THIS_MODULE,
1518                    .ca_name = "pid",
1519                    .ca_mode = S_IRUGO | S_IRUSR },
1520        .show   = o2hb_region_pid_read,
1521 };
1522 
1523 static struct configfs_attribute *o2hb_region_attrs[] = {
1524 	&o2hb_region_attr_block_bytes.attr,
1525 	&o2hb_region_attr_start_block.attr,
1526 	&o2hb_region_attr_blocks.attr,
1527 	&o2hb_region_attr_dev.attr,
1528 	&o2hb_region_attr_pid.attr,
1529 	NULL,
1530 };
1531 
1532 static ssize_t o2hb_region_show(struct config_item *item,
1533 				struct configfs_attribute *attr,
1534 				char *page)
1535 {
1536 	struct o2hb_region *reg = to_o2hb_region(item);
1537 	struct o2hb_region_attribute *o2hb_region_attr =
1538 		container_of(attr, struct o2hb_region_attribute, attr);
1539 	ssize_t ret = 0;
1540 
1541 	if (o2hb_region_attr->show)
1542 		ret = o2hb_region_attr->show(reg, page);
1543 	return ret;
1544 }
1545 
1546 static ssize_t o2hb_region_store(struct config_item *item,
1547 				 struct configfs_attribute *attr,
1548 				 const char *page, size_t count)
1549 {
1550 	struct o2hb_region *reg = to_o2hb_region(item);
1551 	struct o2hb_region_attribute *o2hb_region_attr =
1552 		container_of(attr, struct o2hb_region_attribute, attr);
1553 	ssize_t ret = -EINVAL;
1554 
1555 	if (o2hb_region_attr->store)
1556 		ret = o2hb_region_attr->store(reg, page, count);
1557 	return ret;
1558 }
1559 
1560 static struct configfs_item_operations o2hb_region_item_ops = {
1561 	.release		= o2hb_region_release,
1562 	.show_attribute		= o2hb_region_show,
1563 	.store_attribute	= o2hb_region_store,
1564 };
1565 
1566 static struct config_item_type o2hb_region_type = {
1567 	.ct_item_ops	= &o2hb_region_item_ops,
1568 	.ct_attrs	= o2hb_region_attrs,
1569 	.ct_owner	= THIS_MODULE,
1570 };
1571 
1572 /* heartbeat set */
1573 
1574 struct o2hb_heartbeat_group {
1575 	struct config_group hs_group;
1576 	/* some stuff? */
1577 };
1578 
1579 static struct o2hb_heartbeat_group *to_o2hb_heartbeat_group(struct config_group *group)
1580 {
1581 	return group ?
1582 		container_of(group, struct o2hb_heartbeat_group, hs_group)
1583 		: NULL;
1584 }
1585 
1586 static struct config_item *o2hb_heartbeat_group_make_item(struct config_group *group,
1587 							  const char *name)
1588 {
1589 	struct o2hb_region *reg = NULL;
1590 
1591 	reg = kzalloc(sizeof(struct o2hb_region), GFP_KERNEL);
1592 	if (reg == NULL)
1593 		return ERR_PTR(-ENOMEM);
1594 
1595 	config_item_init_type_name(&reg->hr_item, name, &o2hb_region_type);
1596 
1597 	spin_lock(&o2hb_live_lock);
1598 	list_add_tail(&reg->hr_all_item, &o2hb_all_regions);
1599 	spin_unlock(&o2hb_live_lock);
1600 
1601 	return &reg->hr_item;
1602 }
1603 
1604 static void o2hb_heartbeat_group_drop_item(struct config_group *group,
1605 					   struct config_item *item)
1606 {
1607 	struct task_struct *hb_task;
1608 	struct o2hb_region *reg = to_o2hb_region(item);
1609 
1610 	/* stop the thread when the user removes the region dir */
1611 	spin_lock(&o2hb_live_lock);
1612 	hb_task = reg->hr_task;
1613 	reg->hr_task = NULL;
1614 	spin_unlock(&o2hb_live_lock);
1615 
1616 	if (hb_task)
1617 		kthread_stop(hb_task);
1618 
1619 	/*
1620 	 * If we're racing a dev_write(), we need to wake them.  They will
1621 	 * check reg->hr_task
1622 	 */
1623 	if (atomic_read(&reg->hr_steady_iterations) != 0) {
1624 		atomic_set(&reg->hr_steady_iterations, 0);
1625 		wake_up(&o2hb_steady_queue);
1626 	}
1627 
1628 	config_item_put(item);
1629 }
1630 
1631 struct o2hb_heartbeat_group_attribute {
1632 	struct configfs_attribute attr;
1633 	ssize_t (*show)(struct o2hb_heartbeat_group *, char *);
1634 	ssize_t (*store)(struct o2hb_heartbeat_group *, const char *, size_t);
1635 };
1636 
1637 static ssize_t o2hb_heartbeat_group_show(struct config_item *item,
1638 					 struct configfs_attribute *attr,
1639 					 char *page)
1640 {
1641 	struct o2hb_heartbeat_group *reg = to_o2hb_heartbeat_group(to_config_group(item));
1642 	struct o2hb_heartbeat_group_attribute *o2hb_heartbeat_group_attr =
1643 		container_of(attr, struct o2hb_heartbeat_group_attribute, attr);
1644 	ssize_t ret = 0;
1645 
1646 	if (o2hb_heartbeat_group_attr->show)
1647 		ret = o2hb_heartbeat_group_attr->show(reg, page);
1648 	return ret;
1649 }
1650 
1651 static ssize_t o2hb_heartbeat_group_store(struct config_item *item,
1652 					  struct configfs_attribute *attr,
1653 					  const char *page, size_t count)
1654 {
1655 	struct o2hb_heartbeat_group *reg = to_o2hb_heartbeat_group(to_config_group(item));
1656 	struct o2hb_heartbeat_group_attribute *o2hb_heartbeat_group_attr =
1657 		container_of(attr, struct o2hb_heartbeat_group_attribute, attr);
1658 	ssize_t ret = -EINVAL;
1659 
1660 	if (o2hb_heartbeat_group_attr->store)
1661 		ret = o2hb_heartbeat_group_attr->store(reg, page, count);
1662 	return ret;
1663 }
1664 
1665 static ssize_t o2hb_heartbeat_group_threshold_show(struct o2hb_heartbeat_group *group,
1666 						     char *page)
1667 {
1668 	return sprintf(page, "%u\n", o2hb_dead_threshold);
1669 }
1670 
1671 static ssize_t o2hb_heartbeat_group_threshold_store(struct o2hb_heartbeat_group *group,
1672 						    const char *page,
1673 						    size_t count)
1674 {
1675 	unsigned long tmp;
1676 	char *p = (char *)page;
1677 
1678 	tmp = simple_strtoul(p, &p, 10);
1679 	if (!p || (*p && (*p != '\n')))
1680                 return -EINVAL;
1681 
1682 	/* this will validate ranges for us. */
1683 	o2hb_dead_threshold_set((unsigned int) tmp);
1684 
1685 	return count;
1686 }
1687 
1688 static struct o2hb_heartbeat_group_attribute o2hb_heartbeat_group_attr_threshold = {
1689 	.attr	= { .ca_owner = THIS_MODULE,
1690 		    .ca_name = "dead_threshold",
1691 		    .ca_mode = S_IRUGO | S_IWUSR },
1692 	.show	= o2hb_heartbeat_group_threshold_show,
1693 	.store	= o2hb_heartbeat_group_threshold_store,
1694 };
1695 
1696 static struct configfs_attribute *o2hb_heartbeat_group_attrs[] = {
1697 	&o2hb_heartbeat_group_attr_threshold.attr,
1698 	NULL,
1699 };
1700 
1701 static struct configfs_item_operations o2hb_hearbeat_group_item_ops = {
1702 	.show_attribute		= o2hb_heartbeat_group_show,
1703 	.store_attribute	= o2hb_heartbeat_group_store,
1704 };
1705 
1706 static struct configfs_group_operations o2hb_heartbeat_group_group_ops = {
1707 	.make_item	= o2hb_heartbeat_group_make_item,
1708 	.drop_item	= o2hb_heartbeat_group_drop_item,
1709 };
1710 
1711 static struct config_item_type o2hb_heartbeat_group_type = {
1712 	.ct_group_ops	= &o2hb_heartbeat_group_group_ops,
1713 	.ct_item_ops	= &o2hb_hearbeat_group_item_ops,
1714 	.ct_attrs	= o2hb_heartbeat_group_attrs,
1715 	.ct_owner	= THIS_MODULE,
1716 };
1717 
1718 /* this is just here to avoid touching group in heartbeat.h which the
1719  * entire damn world #includes */
1720 struct config_group *o2hb_alloc_hb_set(void)
1721 {
1722 	struct o2hb_heartbeat_group *hs = NULL;
1723 	struct config_group *ret = NULL;
1724 
1725 	hs = kzalloc(sizeof(struct o2hb_heartbeat_group), GFP_KERNEL);
1726 	if (hs == NULL)
1727 		goto out;
1728 
1729 	config_group_init_type_name(&hs->hs_group, "heartbeat",
1730 				    &o2hb_heartbeat_group_type);
1731 
1732 	ret = &hs->hs_group;
1733 out:
1734 	if (ret == NULL)
1735 		kfree(hs);
1736 	return ret;
1737 }
1738 
1739 void o2hb_free_hb_set(struct config_group *group)
1740 {
1741 	struct o2hb_heartbeat_group *hs = to_o2hb_heartbeat_group(group);
1742 	kfree(hs);
1743 }
1744 
1745 /* hb callback registration and issueing */
1746 
1747 static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type)
1748 {
1749 	if (type == O2HB_NUM_CB)
1750 		return ERR_PTR(-EINVAL);
1751 
1752 	return &o2hb_callbacks[type];
1753 }
1754 
1755 void o2hb_setup_callback(struct o2hb_callback_func *hc,
1756 			 enum o2hb_callback_type type,
1757 			 o2hb_cb_func *func,
1758 			 void *data,
1759 			 int priority)
1760 {
1761 	INIT_LIST_HEAD(&hc->hc_item);
1762 	hc->hc_func = func;
1763 	hc->hc_data = data;
1764 	hc->hc_priority = priority;
1765 	hc->hc_type = type;
1766 	hc->hc_magic = O2HB_CB_MAGIC;
1767 }
1768 EXPORT_SYMBOL_GPL(o2hb_setup_callback);
1769 
1770 static struct o2hb_region *o2hb_find_region(const char *region_uuid)
1771 {
1772 	struct o2hb_region *p, *reg = NULL;
1773 
1774 	assert_spin_locked(&o2hb_live_lock);
1775 
1776 	list_for_each_entry(p, &o2hb_all_regions, hr_all_item) {
1777 		if (!strcmp(region_uuid, config_item_name(&p->hr_item))) {
1778 			reg = p;
1779 			break;
1780 		}
1781 	}
1782 
1783 	return reg;
1784 }
1785 
1786 static int o2hb_region_get(const char *region_uuid)
1787 {
1788 	int ret = 0;
1789 	struct o2hb_region *reg;
1790 
1791 	spin_lock(&o2hb_live_lock);
1792 
1793 	reg = o2hb_find_region(region_uuid);
1794 	if (!reg)
1795 		ret = -ENOENT;
1796 	spin_unlock(&o2hb_live_lock);
1797 
1798 	if (ret)
1799 		goto out;
1800 
1801 	ret = o2nm_depend_this_node();
1802 	if (ret)
1803 		goto out;
1804 
1805 	ret = o2nm_depend_item(&reg->hr_item);
1806 	if (ret)
1807 		o2nm_undepend_this_node();
1808 
1809 out:
1810 	return ret;
1811 }
1812 
1813 static void o2hb_region_put(const char *region_uuid)
1814 {
1815 	struct o2hb_region *reg;
1816 
1817 	spin_lock(&o2hb_live_lock);
1818 
1819 	reg = o2hb_find_region(region_uuid);
1820 
1821 	spin_unlock(&o2hb_live_lock);
1822 
1823 	if (reg) {
1824 		o2nm_undepend_item(&reg->hr_item);
1825 		o2nm_undepend_this_node();
1826 	}
1827 }
1828 
1829 int o2hb_register_callback(const char *region_uuid,
1830 			   struct o2hb_callback_func *hc)
1831 {
1832 	struct o2hb_callback_func *tmp;
1833 	struct list_head *iter;
1834 	struct o2hb_callback *hbcall;
1835 	int ret;
1836 
1837 	BUG_ON(hc->hc_magic != O2HB_CB_MAGIC);
1838 	BUG_ON(!list_empty(&hc->hc_item));
1839 
1840 	hbcall = hbcall_from_type(hc->hc_type);
1841 	if (IS_ERR(hbcall)) {
1842 		ret = PTR_ERR(hbcall);
1843 		goto out;
1844 	}
1845 
1846 	if (region_uuid) {
1847 		ret = o2hb_region_get(region_uuid);
1848 		if (ret)
1849 			goto out;
1850 	}
1851 
1852 	down_write(&o2hb_callback_sem);
1853 
1854 	list_for_each(iter, &hbcall->list) {
1855 		tmp = list_entry(iter, struct o2hb_callback_func, hc_item);
1856 		if (hc->hc_priority < tmp->hc_priority) {
1857 			list_add_tail(&hc->hc_item, iter);
1858 			break;
1859 		}
1860 	}
1861 	if (list_empty(&hc->hc_item))
1862 		list_add_tail(&hc->hc_item, &hbcall->list);
1863 
1864 	up_write(&o2hb_callback_sem);
1865 	ret = 0;
1866 out:
1867 	mlog(ML_HEARTBEAT, "returning %d on behalf of %p for funcs %p\n",
1868 	     ret, __builtin_return_address(0), hc);
1869 	return ret;
1870 }
1871 EXPORT_SYMBOL_GPL(o2hb_register_callback);
1872 
1873 void o2hb_unregister_callback(const char *region_uuid,
1874 			      struct o2hb_callback_func *hc)
1875 {
1876 	BUG_ON(hc->hc_magic != O2HB_CB_MAGIC);
1877 
1878 	mlog(ML_HEARTBEAT, "on behalf of %p for funcs %p\n",
1879 	     __builtin_return_address(0), hc);
1880 
1881 	/* XXX Can this happen _with_ a region reference? */
1882 	if (list_empty(&hc->hc_item))
1883 		return;
1884 
1885 	if (region_uuid)
1886 		o2hb_region_put(region_uuid);
1887 
1888 	down_write(&o2hb_callback_sem);
1889 
1890 	list_del_init(&hc->hc_item);
1891 
1892 	up_write(&o2hb_callback_sem);
1893 }
1894 EXPORT_SYMBOL_GPL(o2hb_unregister_callback);
1895 
1896 int o2hb_check_node_heartbeating(u8 node_num)
1897 {
1898 	unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)];
1899 
1900 	o2hb_fill_node_map(testing_map, sizeof(testing_map));
1901 	if (!test_bit(node_num, testing_map)) {
1902 		mlog(ML_HEARTBEAT,
1903 		     "node (%u) does not have heartbeating enabled.\n",
1904 		     node_num);
1905 		return 0;
1906 	}
1907 
1908 	return 1;
1909 }
1910 EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating);
1911 
1912 int o2hb_check_node_heartbeating_from_callback(u8 node_num)
1913 {
1914 	unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)];
1915 
1916 	o2hb_fill_node_map_from_callback(testing_map, sizeof(testing_map));
1917 	if (!test_bit(node_num, testing_map)) {
1918 		mlog(ML_HEARTBEAT,
1919 		     "node (%u) does not have heartbeating enabled.\n",
1920 		     node_num);
1921 		return 0;
1922 	}
1923 
1924 	return 1;
1925 }
1926 EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating_from_callback);
1927 
1928 /* Makes sure our local node is configured with a node number, and is
1929  * heartbeating. */
1930 int o2hb_check_local_node_heartbeating(void)
1931 {
1932 	u8 node_num;
1933 
1934 	/* if this node was set then we have networking */
1935 	node_num = o2nm_this_node();
1936 	if (node_num == O2NM_MAX_NODES) {
1937 		mlog(ML_HEARTBEAT, "this node has not been configured.\n");
1938 		return 0;
1939 	}
1940 
1941 	return o2hb_check_node_heartbeating(node_num);
1942 }
1943 EXPORT_SYMBOL_GPL(o2hb_check_local_node_heartbeating);
1944 
1945 /*
1946  * this is just a hack until we get the plumbing which flips file systems
1947  * read only and drops the hb ref instead of killing the node dead.
1948  */
1949 void o2hb_stop_all_regions(void)
1950 {
1951 	struct o2hb_region *reg;
1952 
1953 	mlog(ML_ERROR, "stopping heartbeat on all active regions.\n");
1954 
1955 	spin_lock(&o2hb_live_lock);
1956 
1957 	list_for_each_entry(reg, &o2hb_all_regions, hr_all_item)
1958 		reg->hr_unclean_stop = 1;
1959 
1960 	spin_unlock(&o2hb_live_lock);
1961 }
1962 EXPORT_SYMBOL_GPL(o2hb_stop_all_regions);
1963