Commit | Line | Data |
---|---|---|
a7f6a5fb MF |
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 | ||
37 | #include "heartbeat.h" | |
38 | #include "tcp.h" | |
39 | #include "nodemanager.h" | |
40 | #include "quorum.h" | |
41 | ||
42 | #include "masklog.h" | |
43 | ||
44 | ||
45 | /* | |
46 | * The first heartbeat pass had one global thread that would serialize all hb | |
47 | * callback calls. This global serializing sem should only be removed once | |
48 | * we've made sure that all callees can deal with being called concurrently | |
49 | * from multiple hb region threads. | |
50 | */ | |
51 | static DECLARE_RWSEM(o2hb_callback_sem); | |
52 | ||
53 | /* | |
54 | * multiple hb threads are watching multiple regions. A node is live | |
55 | * whenever any of the threads sees activity from the node in its region. | |
56 | */ | |
57 | static spinlock_t o2hb_live_lock = SPIN_LOCK_UNLOCKED; | |
58 | static struct list_head o2hb_live_slots[O2NM_MAX_NODES]; | |
59 | static unsigned long o2hb_live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; | |
60 | static LIST_HEAD(o2hb_node_events); | |
61 | static DECLARE_WAIT_QUEUE_HEAD(o2hb_steady_queue); | |
62 | ||
63 | static LIST_HEAD(o2hb_all_regions); | |
64 | ||
65 | static struct o2hb_callback { | |
66 | struct list_head list; | |
67 | } o2hb_callbacks[O2HB_NUM_CB]; | |
68 | ||
69 | static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type); | |
70 | ||
71 | #define O2HB_DEFAULT_BLOCK_BITS 9 | |
72 | ||
73 | unsigned int o2hb_dead_threshold = O2HB_DEFAULT_DEAD_THRESHOLD; | |
74 | ||
75 | /* Only sets a new threshold if there are no active regions. | |
76 | * | |
77 | * No locking or otherwise interesting code is required for reading | |
78 | * o2hb_dead_threshold as it can't change once regions are active and | |
79 | * it's not interesting to anyone until then anyway. */ | |
80 | static void o2hb_dead_threshold_set(unsigned int threshold) | |
81 | { | |
82 | if (threshold > O2HB_MIN_DEAD_THRESHOLD) { | |
83 | spin_lock(&o2hb_live_lock); | |
84 | if (list_empty(&o2hb_all_regions)) | |
85 | o2hb_dead_threshold = threshold; | |
86 | spin_unlock(&o2hb_live_lock); | |
87 | } | |
88 | } | |
89 | ||
90 | struct o2hb_node_event { | |
91 | struct list_head hn_item; | |
92 | enum o2hb_callback_type hn_event_type; | |
93 | struct o2nm_node *hn_node; | |
94 | int hn_node_num; | |
95 | }; | |
96 | ||
97 | struct o2hb_disk_slot { | |
98 | struct o2hb_disk_heartbeat_block *ds_raw_block; | |
99 | u8 ds_node_num; | |
100 | u64 ds_last_time; | |
101 | u64 ds_last_generation; | |
102 | u16 ds_equal_samples; | |
103 | u16 ds_changed_samples; | |
104 | struct list_head ds_live_item; | |
105 | }; | |
106 | ||
107 | /* each thread owns a region.. when we're asked to tear down the region | |
108 | * we ask the thread to stop, who cleans up the region */ | |
109 | struct o2hb_region { | |
110 | struct config_item hr_item; | |
111 | ||
112 | struct list_head hr_all_item; | |
113 | unsigned hr_unclean_stop:1; | |
114 | ||
115 | /* protected by the hr_callback_sem */ | |
116 | struct task_struct *hr_task; | |
117 | ||
118 | unsigned int hr_blocks; | |
119 | unsigned long long hr_start_block; | |
120 | ||
121 | unsigned int hr_block_bits; | |
122 | unsigned int hr_block_bytes; | |
123 | ||
124 | unsigned int hr_slots_per_page; | |
125 | unsigned int hr_num_pages; | |
126 | ||
127 | struct page **hr_slot_data; | |
128 | struct block_device *hr_bdev; | |
129 | struct o2hb_disk_slot *hr_slots; | |
130 | ||
131 | /* let the person setting up hb wait for it to return until it | |
132 | * has reached a 'steady' state. This will be fixed when we have | |
133 | * a more complete api that doesn't lead to this sort of fragility. */ | |
134 | atomic_t hr_steady_iterations; | |
135 | ||
136 | char hr_dev_name[BDEVNAME_SIZE]; | |
137 | ||
138 | unsigned int hr_timeout_ms; | |
139 | ||
140 | /* randomized as the region goes up and down so that a node | |
141 | * recognizes a node going up and down in one iteration */ | |
142 | u64 hr_generation; | |
143 | ||
144 | struct work_struct hr_write_timeout_work; | |
145 | unsigned long hr_last_timeout_start; | |
146 | ||
147 | /* Used during o2hb_check_slot to hold a copy of the block | |
148 | * being checked because we temporarily have to zero out the | |
149 | * crc field. */ | |
150 | struct o2hb_disk_heartbeat_block *hr_tmp_block; | |
151 | }; | |
152 | ||
153 | struct o2hb_bio_wait_ctxt { | |
154 | atomic_t wc_num_reqs; | |
155 | struct completion wc_io_complete; | |
a9e2ae39 | 156 | int wc_error; |
a7f6a5fb MF |
157 | }; |
158 | ||
159 | static void o2hb_write_timeout(void *arg) | |
160 | { | |
161 | struct o2hb_region *reg = arg; | |
162 | ||
163 | mlog(ML_ERROR, "Heartbeat write timeout to device %s after %u " | |
164 | "milliseconds\n", reg->hr_dev_name, | |
165 | jiffies_to_msecs(jiffies - reg->hr_last_timeout_start)); | |
166 | o2quo_disk_timeout(); | |
167 | } | |
168 | ||
169 | static void o2hb_arm_write_timeout(struct o2hb_region *reg) | |
170 | { | |
171 | mlog(0, "Queue write timeout for %u ms\n", O2HB_MAX_WRITE_TIMEOUT_MS); | |
172 | ||
173 | cancel_delayed_work(®->hr_write_timeout_work); | |
174 | reg->hr_last_timeout_start = jiffies; | |
175 | schedule_delayed_work(®->hr_write_timeout_work, | |
176 | msecs_to_jiffies(O2HB_MAX_WRITE_TIMEOUT_MS)); | |
177 | } | |
178 | ||
179 | static void o2hb_disarm_write_timeout(struct o2hb_region *reg) | |
180 | { | |
181 | cancel_delayed_work(®->hr_write_timeout_work); | |
182 | flush_scheduled_work(); | |
183 | } | |
184 | ||
185 | static inline void o2hb_bio_wait_init(struct o2hb_bio_wait_ctxt *wc, | |
186 | unsigned int num_ios) | |
187 | { | |
188 | atomic_set(&wc->wc_num_reqs, num_ios); | |
189 | init_completion(&wc->wc_io_complete); | |
a9e2ae39 | 190 | wc->wc_error = 0; |
a7f6a5fb MF |
191 | } |
192 | ||
193 | /* Used in error paths too */ | |
194 | static inline void o2hb_bio_wait_dec(struct o2hb_bio_wait_ctxt *wc, | |
195 | unsigned int num) | |
196 | { | |
197 | /* sadly atomic_sub_and_test() isn't available on all platforms. The | |
198 | * good news is that the fast path only completes one at a time */ | |
199 | while(num--) { | |
200 | if (atomic_dec_and_test(&wc->wc_num_reqs)) { | |
201 | BUG_ON(num > 0); | |
202 | complete(&wc->wc_io_complete); | |
203 | } | |
204 | } | |
205 | } | |
206 | ||
207 | static void o2hb_wait_on_io(struct o2hb_region *reg, | |
208 | struct o2hb_bio_wait_ctxt *wc) | |
209 | { | |
210 | struct address_space *mapping = reg->hr_bdev->bd_inode->i_mapping; | |
211 | ||
212 | blk_run_address_space(mapping); | |
213 | ||
214 | wait_for_completion(&wc->wc_io_complete); | |
215 | } | |
216 | ||
217 | static int o2hb_bio_end_io(struct bio *bio, | |
218 | unsigned int bytes_done, | |
219 | int error) | |
220 | { | |
221 | struct o2hb_bio_wait_ctxt *wc = bio->bi_private; | |
222 | ||
a9e2ae39 | 223 | if (error) { |
a7f6a5fb | 224 | mlog(ML_ERROR, "IO Error %d\n", error); |
a9e2ae39 MF |
225 | wc->wc_error = error; |
226 | } | |
a7f6a5fb MF |
227 | |
228 | if (bio->bi_size) | |
229 | return 1; | |
230 | ||
231 | o2hb_bio_wait_dec(wc, 1); | |
232 | return 0; | |
233 | } | |
234 | ||
235 | /* Setup a Bio to cover I/O against num_slots slots starting at | |
236 | * start_slot. */ | |
237 | static struct bio *o2hb_setup_one_bio(struct o2hb_region *reg, | |
238 | struct o2hb_bio_wait_ctxt *wc, | |
239 | unsigned int start_slot, | |
240 | unsigned int num_slots) | |
241 | { | |
242 | int i, nr_vecs, len, first_page, last_page; | |
243 | unsigned int vec_len, vec_start; | |
244 | unsigned int bits = reg->hr_block_bits; | |
245 | unsigned int spp = reg->hr_slots_per_page; | |
246 | struct bio *bio; | |
247 | struct page *page; | |
248 | ||
249 | nr_vecs = (num_slots + spp - 1) / spp; | |
250 | ||
251 | /* Testing has shown this allocation to take long enough under | |
252 | * GFP_KERNEL that the local node can get fenced. It would be | |
253 | * nicest if we could pre-allocate these bios and avoid this | |
254 | * all together. */ | |
255 | bio = bio_alloc(GFP_ATOMIC, nr_vecs); | |
256 | if (!bio) { | |
257 | mlog(ML_ERROR, "Could not alloc slots BIO!\n"); | |
258 | bio = ERR_PTR(-ENOMEM); | |
259 | goto bail; | |
260 | } | |
261 | ||
262 | /* Must put everything in 512 byte sectors for the bio... */ | |
263 | bio->bi_sector = (reg->hr_start_block + start_slot) << (bits - 9); | |
264 | bio->bi_bdev = reg->hr_bdev; | |
265 | bio->bi_private = wc; | |
266 | bio->bi_end_io = o2hb_bio_end_io; | |
267 | ||
268 | first_page = start_slot / spp; | |
269 | last_page = first_page + nr_vecs; | |
270 | vec_start = (start_slot << bits) % PAGE_CACHE_SIZE; | |
271 | for(i = first_page; i < last_page; i++) { | |
272 | page = reg->hr_slot_data[i]; | |
273 | ||
274 | vec_len = PAGE_CACHE_SIZE; | |
275 | /* last page might be short */ | |
276 | if (((i + 1) * spp) > (start_slot + num_slots)) | |
277 | vec_len = ((num_slots + start_slot) % spp) << bits; | |
278 | vec_len -= vec_start; | |
279 | ||
280 | mlog(ML_HB_BIO, "page %d, vec_len = %u, vec_start = %u\n", | |
281 | i, vec_len, vec_start); | |
282 | ||
283 | len = bio_add_page(bio, page, vec_len, vec_start); | |
284 | if (len != vec_len) { | |
285 | bio_put(bio); | |
286 | bio = ERR_PTR(-EIO); | |
287 | ||
288 | mlog(ML_ERROR, "Error adding page to bio i = %d, " | |
289 | "vec_len = %u, len = %d\n, start = %u\n", | |
290 | i, vec_len, len, vec_start); | |
291 | goto bail; | |
292 | } | |
293 | ||
294 | vec_start = 0; | |
295 | } | |
296 | ||
297 | bail: | |
298 | return bio; | |
299 | } | |
300 | ||
301 | /* | |
302 | * Compute the maximum number of sectors the bdev can handle in one bio, | |
303 | * as a power of two. | |
304 | * | |
305 | * Stolen from oracleasm, thanks Joel! | |
306 | */ | |
307 | static int compute_max_sectors(struct block_device *bdev) | |
308 | { | |
309 | int max_pages, max_sectors, pow_two_sectors; | |
310 | ||
311 | struct request_queue *q; | |
312 | ||
313 | q = bdev_get_queue(bdev); | |
314 | max_pages = q->max_sectors >> (PAGE_SHIFT - 9); | |
315 | if (max_pages > BIO_MAX_PAGES) | |
316 | max_pages = BIO_MAX_PAGES; | |
317 | if (max_pages > q->max_phys_segments) | |
318 | max_pages = q->max_phys_segments; | |
319 | if (max_pages > q->max_hw_segments) | |
320 | max_pages = q->max_hw_segments; | |
321 | max_pages--; /* Handle I/Os that straddle a page */ | |
322 | ||
323 | max_sectors = max_pages << (PAGE_SHIFT - 9); | |
324 | ||
325 | /* Why is fls() 1-based???? */ | |
326 | pow_two_sectors = 1 << (fls(max_sectors) - 1); | |
327 | ||
328 | return pow_two_sectors; | |
329 | } | |
330 | ||
331 | static inline void o2hb_compute_request_limits(struct o2hb_region *reg, | |
332 | unsigned int num_slots, | |
333 | unsigned int *num_bios, | |
334 | unsigned int *slots_per_bio) | |
335 | { | |
336 | unsigned int max_sectors, io_sectors; | |
337 | ||
338 | max_sectors = compute_max_sectors(reg->hr_bdev); | |
339 | ||
340 | io_sectors = num_slots << (reg->hr_block_bits - 9); | |
341 | ||
342 | *num_bios = (io_sectors + max_sectors - 1) / max_sectors; | |
343 | *slots_per_bio = max_sectors >> (reg->hr_block_bits - 9); | |
344 | ||
345 | mlog(ML_HB_BIO, "My io size is %u sectors for %u slots. This " | |
346 | "device can handle %u sectors of I/O\n", io_sectors, num_slots, | |
347 | max_sectors); | |
348 | mlog(ML_HB_BIO, "Will need %u bios holding %u slots each\n", | |
349 | *num_bios, *slots_per_bio); | |
350 | } | |
351 | ||
352 | static int o2hb_read_slots(struct o2hb_region *reg, | |
353 | unsigned int max_slots) | |
354 | { | |
355 | unsigned int num_bios, slots_per_bio, start_slot, num_slots; | |
356 | int i, status; | |
357 | struct o2hb_bio_wait_ctxt wc; | |
358 | struct bio **bios; | |
359 | struct bio *bio; | |
360 | ||
361 | o2hb_compute_request_limits(reg, max_slots, &num_bios, &slots_per_bio); | |
362 | ||
363 | bios = kcalloc(num_bios, sizeof(struct bio *), GFP_KERNEL); | |
364 | if (!bios) { | |
365 | status = -ENOMEM; | |
366 | mlog_errno(status); | |
367 | return status; | |
368 | } | |
369 | ||
370 | o2hb_bio_wait_init(&wc, num_bios); | |
371 | ||
372 | num_slots = slots_per_bio; | |
373 | for(i = 0; i < num_bios; i++) { | |
374 | start_slot = i * slots_per_bio; | |
375 | ||
376 | /* adjust num_slots at last bio */ | |
377 | if (max_slots < (start_slot + num_slots)) | |
378 | num_slots = max_slots - start_slot; | |
379 | ||
380 | bio = o2hb_setup_one_bio(reg, &wc, start_slot, num_slots); | |
381 | if (IS_ERR(bio)) { | |
382 | o2hb_bio_wait_dec(&wc, num_bios - i); | |
383 | ||
384 | status = PTR_ERR(bio); | |
385 | mlog_errno(status); | |
386 | goto bail_and_wait; | |
387 | } | |
388 | bios[i] = bio; | |
389 | ||
390 | submit_bio(READ, bio); | |
391 | } | |
392 | ||
393 | status = 0; | |
394 | ||
395 | bail_and_wait: | |
396 | o2hb_wait_on_io(reg, &wc); | |
a9e2ae39 MF |
397 | if (wc.wc_error && !status) |
398 | status = wc.wc_error; | |
a7f6a5fb MF |
399 | |
400 | if (bios) { | |
401 | for(i = 0; i < num_bios; i++) | |
402 | if (bios[i]) | |
403 | bio_put(bios[i]); | |
404 | kfree(bios); | |
405 | } | |
406 | ||
407 | return status; | |
408 | } | |
409 | ||
410 | static int o2hb_issue_node_write(struct o2hb_region *reg, | |
411 | struct bio **write_bio, | |
412 | struct o2hb_bio_wait_ctxt *write_wc) | |
413 | { | |
414 | int status; | |
415 | unsigned int slot; | |
416 | struct bio *bio; | |
417 | ||
418 | o2hb_bio_wait_init(write_wc, 1); | |
419 | ||
420 | slot = o2nm_this_node(); | |
421 | ||
422 | bio = o2hb_setup_one_bio(reg, write_wc, slot, 1); | |
423 | if (IS_ERR(bio)) { | |
424 | status = PTR_ERR(bio); | |
425 | mlog_errno(status); | |
426 | goto bail; | |
427 | } | |
428 | ||
429 | submit_bio(WRITE, bio); | |
430 | ||
431 | *write_bio = bio; | |
432 | status = 0; | |
433 | bail: | |
434 | return status; | |
435 | } | |
436 | ||
437 | static u32 o2hb_compute_block_crc_le(struct o2hb_region *reg, | |
438 | struct o2hb_disk_heartbeat_block *hb_block) | |
439 | { | |
440 | __le32 old_cksum; | |
441 | u32 ret; | |
442 | ||
443 | /* We want to compute the block crc with a 0 value in the | |
444 | * hb_cksum field. Save it off here and replace after the | |
445 | * crc. */ | |
446 | old_cksum = hb_block->hb_cksum; | |
447 | hb_block->hb_cksum = 0; | |
448 | ||
449 | ret = crc32_le(0, (unsigned char *) hb_block, reg->hr_block_bytes); | |
450 | ||
451 | hb_block->hb_cksum = old_cksum; | |
452 | ||
453 | return ret; | |
454 | } | |
455 | ||
456 | static void o2hb_dump_slot(struct o2hb_disk_heartbeat_block *hb_block) | |
457 | { | |
70bacbdb MF |
458 | mlog(ML_ERROR, "Dump slot information: seq = 0x%llx, node = %u, " |
459 | "cksum = 0x%x, generation 0x%llx\n", | |
460 | (long long)le64_to_cpu(hb_block->hb_seq), | |
461 | hb_block->hb_node, le32_to_cpu(hb_block->hb_cksum), | |
462 | (long long)le64_to_cpu(hb_block->hb_generation)); | |
a7f6a5fb MF |
463 | } |
464 | ||
465 | static int o2hb_verify_crc(struct o2hb_region *reg, | |
466 | struct o2hb_disk_heartbeat_block *hb_block) | |
467 | { | |
468 | u32 read, computed; | |
469 | ||
470 | read = le32_to_cpu(hb_block->hb_cksum); | |
471 | computed = o2hb_compute_block_crc_le(reg, hb_block); | |
472 | ||
473 | return read == computed; | |
474 | } | |
475 | ||
476 | /* We want to make sure that nobody is heartbeating on top of us -- | |
477 | * this will help detect an invalid configuration. */ | |
478 | static int o2hb_check_last_timestamp(struct o2hb_region *reg) | |
479 | { | |
480 | int node_num, ret; | |
481 | struct o2hb_disk_slot *slot; | |
482 | struct o2hb_disk_heartbeat_block *hb_block; | |
483 | ||
484 | node_num = o2nm_this_node(); | |
485 | ||
486 | ret = 1; | |
487 | slot = ®->hr_slots[node_num]; | |
488 | /* Don't check on our 1st timestamp */ | |
489 | if (slot->ds_last_time) { | |
490 | hb_block = slot->ds_raw_block; | |
491 | ||
492 | if (le64_to_cpu(hb_block->hb_seq) != slot->ds_last_time) | |
493 | ret = 0; | |
494 | } | |
495 | ||
496 | return ret; | |
497 | } | |
498 | ||
499 | static inline void o2hb_prepare_block(struct o2hb_region *reg, | |
500 | u64 generation) | |
501 | { | |
502 | int node_num; | |
503 | u64 cputime; | |
504 | struct o2hb_disk_slot *slot; | |
505 | struct o2hb_disk_heartbeat_block *hb_block; | |
506 | ||
507 | node_num = o2nm_this_node(); | |
508 | slot = ®->hr_slots[node_num]; | |
509 | ||
510 | hb_block = (struct o2hb_disk_heartbeat_block *)slot->ds_raw_block; | |
511 | memset(hb_block, 0, reg->hr_block_bytes); | |
512 | /* TODO: time stuff */ | |
513 | cputime = CURRENT_TIME.tv_sec; | |
514 | if (!cputime) | |
515 | cputime = 1; | |
516 | ||
517 | hb_block->hb_seq = cpu_to_le64(cputime); | |
518 | hb_block->hb_node = node_num; | |
519 | hb_block->hb_generation = cpu_to_le64(generation); | |
520 | ||
521 | /* This step must always happen last! */ | |
522 | hb_block->hb_cksum = cpu_to_le32(o2hb_compute_block_crc_le(reg, | |
523 | hb_block)); | |
524 | ||
70bacbdb MF |
525 | mlog(ML_HB_BIO, "our node generation = 0x%llx, cksum = 0x%x\n", |
526 | (long long)cpu_to_le64(generation), | |
527 | le32_to_cpu(hb_block->hb_cksum)); | |
a7f6a5fb MF |
528 | } |
529 | ||
530 | static void o2hb_fire_callbacks(struct o2hb_callback *hbcall, | |
531 | struct o2nm_node *node, | |
532 | int idx) | |
533 | { | |
534 | struct list_head *iter; | |
535 | struct o2hb_callback_func *f; | |
536 | ||
537 | list_for_each(iter, &hbcall->list) { | |
538 | f = list_entry(iter, struct o2hb_callback_func, hc_item); | |
539 | mlog(ML_HEARTBEAT, "calling funcs %p\n", f); | |
540 | (f->hc_func)(node, idx, f->hc_data); | |
541 | } | |
542 | } | |
543 | ||
544 | /* Will run the list in order until we process the passed event */ | |
545 | static void o2hb_run_event_list(struct o2hb_node_event *queued_event) | |
546 | { | |
547 | int empty; | |
548 | struct o2hb_callback *hbcall; | |
549 | struct o2hb_node_event *event; | |
550 | ||
551 | spin_lock(&o2hb_live_lock); | |
552 | empty = list_empty(&queued_event->hn_item); | |
553 | spin_unlock(&o2hb_live_lock); | |
554 | if (empty) | |
555 | return; | |
556 | ||
557 | /* Holding callback sem assures we don't alter the callback | |
558 | * lists when doing this, and serializes ourselves with other | |
559 | * processes wanting callbacks. */ | |
560 | down_write(&o2hb_callback_sem); | |
561 | ||
562 | spin_lock(&o2hb_live_lock); | |
563 | while (!list_empty(&o2hb_node_events) | |
564 | && !list_empty(&queued_event->hn_item)) { | |
565 | event = list_entry(o2hb_node_events.next, | |
566 | struct o2hb_node_event, | |
567 | hn_item); | |
568 | list_del_init(&event->hn_item); | |
569 | spin_unlock(&o2hb_live_lock); | |
570 | ||
571 | mlog(ML_HEARTBEAT, "Node %s event for %d\n", | |
572 | event->hn_event_type == O2HB_NODE_UP_CB ? "UP" : "DOWN", | |
573 | event->hn_node_num); | |
574 | ||
575 | hbcall = hbcall_from_type(event->hn_event_type); | |
576 | ||
577 | /* We should *never* have gotten on to the list with a | |
578 | * bad type... This isn't something that we should try | |
579 | * to recover from. */ | |
580 | BUG_ON(IS_ERR(hbcall)); | |
581 | ||
582 | o2hb_fire_callbacks(hbcall, event->hn_node, event->hn_node_num); | |
583 | ||
584 | spin_lock(&o2hb_live_lock); | |
585 | } | |
586 | spin_unlock(&o2hb_live_lock); | |
587 | ||
588 | up_write(&o2hb_callback_sem); | |
589 | } | |
590 | ||
591 | static void o2hb_queue_node_event(struct o2hb_node_event *event, | |
592 | enum o2hb_callback_type type, | |
593 | struct o2nm_node *node, | |
594 | int node_num) | |
595 | { | |
596 | assert_spin_locked(&o2hb_live_lock); | |
597 | ||
598 | event->hn_event_type = type; | |
599 | event->hn_node = node; | |
600 | event->hn_node_num = node_num; | |
601 | ||
602 | mlog(ML_HEARTBEAT, "Queue node %s event for node %d\n", | |
603 | type == O2HB_NODE_UP_CB ? "UP" : "DOWN", node_num); | |
604 | ||
605 | list_add_tail(&event->hn_item, &o2hb_node_events); | |
606 | } | |
607 | ||
608 | static void o2hb_shutdown_slot(struct o2hb_disk_slot *slot) | |
609 | { | |
610 | struct o2hb_node_event event = | |
611 | { .hn_item = LIST_HEAD_INIT(event.hn_item), }; | |
612 | struct o2nm_node *node; | |
613 | ||
614 | node = o2nm_get_node_by_num(slot->ds_node_num); | |
615 | if (!node) | |
616 | return; | |
617 | ||
618 | spin_lock(&o2hb_live_lock); | |
619 | if (!list_empty(&slot->ds_live_item)) { | |
620 | mlog(ML_HEARTBEAT, "Shutdown, node %d leaves region\n", | |
621 | slot->ds_node_num); | |
622 | ||
623 | list_del_init(&slot->ds_live_item); | |
624 | ||
625 | if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { | |
626 | clear_bit(slot->ds_node_num, o2hb_live_node_bitmap); | |
627 | ||
628 | o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node, | |
629 | slot->ds_node_num); | |
630 | } | |
631 | } | |
632 | spin_unlock(&o2hb_live_lock); | |
633 | ||
634 | o2hb_run_event_list(&event); | |
635 | ||
636 | o2nm_node_put(node); | |
637 | } | |
638 | ||
639 | static int o2hb_check_slot(struct o2hb_region *reg, | |
640 | struct o2hb_disk_slot *slot) | |
641 | { | |
642 | int changed = 0, gen_changed = 0; | |
643 | struct o2hb_node_event event = | |
644 | { .hn_item = LIST_HEAD_INIT(event.hn_item), }; | |
645 | struct o2nm_node *node; | |
646 | struct o2hb_disk_heartbeat_block *hb_block = reg->hr_tmp_block; | |
647 | u64 cputime; | |
648 | ||
649 | memcpy(hb_block, slot->ds_raw_block, reg->hr_block_bytes); | |
650 | ||
651 | /* Is this correct? Do we assume that the node doesn't exist | |
652 | * if we're not configured for him? */ | |
653 | node = o2nm_get_node_by_num(slot->ds_node_num); | |
654 | if (!node) | |
655 | return 0; | |
656 | ||
657 | if (!o2hb_verify_crc(reg, hb_block)) { | |
658 | /* all paths from here will drop o2hb_live_lock for | |
659 | * us. */ | |
660 | spin_lock(&o2hb_live_lock); | |
661 | ||
662 | /* Don't print an error on the console in this case - | |
663 | * a freshly formatted heartbeat area will not have a | |
664 | * crc set on it. */ | |
665 | if (list_empty(&slot->ds_live_item)) | |
666 | goto out; | |
667 | ||
668 | /* The node is live but pushed out a bad crc. We | |
669 | * consider it a transient miss but don't populate any | |
670 | * other values as they may be junk. */ | |
671 | mlog(ML_ERROR, "Node %d has written a bad crc to %s\n", | |
672 | slot->ds_node_num, reg->hr_dev_name); | |
673 | o2hb_dump_slot(hb_block); | |
674 | ||
675 | slot->ds_equal_samples++; | |
676 | goto fire_callbacks; | |
677 | } | |
678 | ||
679 | /* we don't care if these wrap.. the state transitions below | |
680 | * clear at the right places */ | |
681 | cputime = le64_to_cpu(hb_block->hb_seq); | |
682 | if (slot->ds_last_time != cputime) | |
683 | slot->ds_changed_samples++; | |
684 | else | |
685 | slot->ds_equal_samples++; | |
686 | slot->ds_last_time = cputime; | |
687 | ||
688 | /* The node changed heartbeat generations. We assume this to | |
689 | * mean it dropped off but came back before we timed out. We | |
690 | * want to consider it down for the time being but don't want | |
691 | * to lose any changed_samples state we might build up to | |
692 | * considering it live again. */ | |
693 | if (slot->ds_last_generation != le64_to_cpu(hb_block->hb_generation)) { | |
694 | gen_changed = 1; | |
695 | slot->ds_equal_samples = 0; | |
70bacbdb MF |
696 | mlog(ML_HEARTBEAT, "Node %d changed generation (0x%llx " |
697 | "to 0x%llx)\n", slot->ds_node_num, | |
698 | (long long)slot->ds_last_generation, | |
699 | (long long)le64_to_cpu(hb_block->hb_generation)); | |
a7f6a5fb MF |
700 | } |
701 | ||
702 | slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation); | |
703 | ||
70bacbdb MF |
704 | mlog(ML_HEARTBEAT, "Slot %d gen 0x%llx cksum 0x%x " |
705 | "seq %llu last %llu changed %u equal %u\n", | |
706 | slot->ds_node_num, (long long)slot->ds_last_generation, | |
707 | le32_to_cpu(hb_block->hb_cksum), | |
708 | (unsigned long long)le64_to_cpu(hb_block->hb_seq), | |
709 | (unsigned long long)slot->ds_last_time, slot->ds_changed_samples, | |
a7f6a5fb MF |
710 | slot->ds_equal_samples); |
711 | ||
712 | spin_lock(&o2hb_live_lock); | |
713 | ||
714 | fire_callbacks: | |
715 | /* dead nodes only come to life after some number of | |
716 | * changes at any time during their dead time */ | |
717 | if (list_empty(&slot->ds_live_item) && | |
718 | slot->ds_changed_samples >= O2HB_LIVE_THRESHOLD) { | |
70bacbdb MF |
719 | mlog(ML_HEARTBEAT, "Node %d (id 0x%llx) joined my region\n", |
720 | slot->ds_node_num, (long long)slot->ds_last_generation); | |
a7f6a5fb MF |
721 | |
722 | /* first on the list generates a callback */ | |
723 | if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { | |
724 | set_bit(slot->ds_node_num, o2hb_live_node_bitmap); | |
725 | ||
726 | o2hb_queue_node_event(&event, O2HB_NODE_UP_CB, node, | |
727 | slot->ds_node_num); | |
728 | ||
729 | changed = 1; | |
730 | } | |
731 | ||
732 | list_add_tail(&slot->ds_live_item, | |
733 | &o2hb_live_slots[slot->ds_node_num]); | |
734 | ||
735 | slot->ds_equal_samples = 0; | |
736 | goto out; | |
737 | } | |
738 | ||
739 | /* if the list is dead, we're done.. */ | |
740 | if (list_empty(&slot->ds_live_item)) | |
741 | goto out; | |
742 | ||
743 | /* live nodes only go dead after enough consequtive missed | |
744 | * samples.. reset the missed counter whenever we see | |
745 | * activity */ | |
746 | if (slot->ds_equal_samples >= o2hb_dead_threshold || gen_changed) { | |
747 | mlog(ML_HEARTBEAT, "Node %d left my region\n", | |
748 | slot->ds_node_num); | |
749 | ||
750 | /* last off the live_slot generates a callback */ | |
751 | list_del_init(&slot->ds_live_item); | |
752 | if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { | |
753 | clear_bit(slot->ds_node_num, o2hb_live_node_bitmap); | |
754 | ||
755 | o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node, | |
756 | slot->ds_node_num); | |
757 | ||
758 | changed = 1; | |
759 | } | |
760 | ||
761 | /* We don't clear this because the node is still | |
762 | * actually writing new blocks. */ | |
763 | if (!gen_changed) | |
764 | slot->ds_changed_samples = 0; | |
765 | goto out; | |
766 | } | |
767 | if (slot->ds_changed_samples) { | |
768 | slot->ds_changed_samples = 0; | |
769 | slot->ds_equal_samples = 0; | |
770 | } | |
771 | out: | |
772 | spin_unlock(&o2hb_live_lock); | |
773 | ||
774 | o2hb_run_event_list(&event); | |
775 | ||
776 | o2nm_node_put(node); | |
777 | return changed; | |
778 | } | |
779 | ||
780 | /* This could be faster if we just implmented a find_last_bit, but I | |
781 | * don't think the circumstances warrant it. */ | |
782 | static int o2hb_highest_node(unsigned long *nodes, | |
783 | int numbits) | |
784 | { | |
785 | int highest, node; | |
786 | ||
787 | highest = numbits; | |
788 | node = -1; | |
789 | while ((node = find_next_bit(nodes, numbits, node + 1)) != -1) { | |
790 | if (node >= numbits) | |
791 | break; | |
792 | ||
793 | highest = node; | |
794 | } | |
795 | ||
796 | return highest; | |
797 | } | |
798 | ||
a9e2ae39 | 799 | static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) |
a7f6a5fb MF |
800 | { |
801 | int i, ret, highest_node, change = 0; | |
802 | unsigned long configured_nodes[BITS_TO_LONGS(O2NM_MAX_NODES)]; | |
803 | struct bio *write_bio; | |
804 | struct o2hb_bio_wait_ctxt write_wc; | |
805 | ||
a9e2ae39 MF |
806 | ret = o2nm_configured_node_map(configured_nodes, |
807 | sizeof(configured_nodes)); | |
808 | if (ret) { | |
809 | mlog_errno(ret); | |
810 | return ret; | |
811 | } | |
a7f6a5fb MF |
812 | |
813 | highest_node = o2hb_highest_node(configured_nodes, O2NM_MAX_NODES); | |
814 | if (highest_node >= O2NM_MAX_NODES) { | |
815 | mlog(ML_NOTICE, "ocfs2_heartbeat: no configured nodes found!\n"); | |
a9e2ae39 | 816 | return -EINVAL; |
a7f6a5fb MF |
817 | } |
818 | ||
819 | /* No sense in reading the slots of nodes that don't exist | |
820 | * yet. Of course, if the node definitions have holes in them | |
821 | * then we're reading an empty slot anyway... Consider this | |
822 | * best-effort. */ | |
823 | ret = o2hb_read_slots(reg, highest_node + 1); | |
824 | if (ret < 0) { | |
825 | mlog_errno(ret); | |
a9e2ae39 | 826 | return ret; |
a7f6a5fb MF |
827 | } |
828 | ||
829 | /* With an up to date view of the slots, we can check that no | |
830 | * other node has been improperly configured to heartbeat in | |
831 | * our slot. */ | |
832 | if (!o2hb_check_last_timestamp(reg)) | |
833 | mlog(ML_ERROR, "Device \"%s\": another node is heartbeating " | |
834 | "in our slot!\n", reg->hr_dev_name); | |
835 | ||
836 | /* fill in the proper info for our next heartbeat */ | |
837 | o2hb_prepare_block(reg, reg->hr_generation); | |
838 | ||
839 | /* And fire off the write. Note that we don't wait on this I/O | |
840 | * until later. */ | |
841 | ret = o2hb_issue_node_write(reg, &write_bio, &write_wc); | |
842 | if (ret < 0) { | |
843 | mlog_errno(ret); | |
a9e2ae39 | 844 | return ret; |
a7f6a5fb MF |
845 | } |
846 | ||
847 | i = -1; | |
848 | while((i = find_next_bit(configured_nodes, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { | |
849 | ||
850 | change |= o2hb_check_slot(reg, ®->hr_slots[i]); | |
851 | } | |
852 | ||
853 | /* | |
854 | * We have to be sure we've advertised ourselves on disk | |
855 | * before we can go to steady state. This ensures that | |
856 | * people we find in our steady state have seen us. | |
857 | */ | |
858 | o2hb_wait_on_io(reg, &write_wc); | |
859 | bio_put(write_bio); | |
a9e2ae39 MF |
860 | if (write_wc.wc_error) { |
861 | /* Do not re-arm the write timeout on I/O error - we | |
862 | * can't be sure that the new block ever made it to | |
863 | * disk */ | |
864 | mlog(ML_ERROR, "Write error %d on device \"%s\"\n", | |
865 | write_wc.wc_error, reg->hr_dev_name); | |
866 | return write_wc.wc_error; | |
867 | } | |
868 | ||
a7f6a5fb MF |
869 | o2hb_arm_write_timeout(reg); |
870 | ||
871 | /* let the person who launched us know when things are steady */ | |
872 | if (!change && (atomic_read(®->hr_steady_iterations) != 0)) { | |
873 | if (atomic_dec_and_test(®->hr_steady_iterations)) | |
874 | wake_up(&o2hb_steady_queue); | |
875 | } | |
a9e2ae39 MF |
876 | |
877 | return 0; | |
a7f6a5fb MF |
878 | } |
879 | ||
880 | /* Subtract b from a, storing the result in a. a *must* have a larger | |
881 | * value than b. */ | |
882 | static void o2hb_tv_subtract(struct timeval *a, | |
883 | struct timeval *b) | |
884 | { | |
885 | /* just return 0 when a is after b */ | |
886 | if (a->tv_sec < b->tv_sec || | |
887 | (a->tv_sec == b->tv_sec && a->tv_usec < b->tv_usec)) { | |
888 | a->tv_sec = 0; | |
889 | a->tv_usec = 0; | |
890 | return; | |
891 | } | |
892 | ||
893 | a->tv_sec -= b->tv_sec; | |
894 | a->tv_usec -= b->tv_usec; | |
895 | while ( a->tv_usec < 0 ) { | |
896 | a->tv_sec--; | |
897 | a->tv_usec += 1000000; | |
898 | } | |
899 | } | |
900 | ||
901 | static unsigned int o2hb_elapsed_msecs(struct timeval *start, | |
902 | struct timeval *end) | |
903 | { | |
904 | struct timeval res = *end; | |
905 | ||
906 | o2hb_tv_subtract(&res, start); | |
907 | ||
908 | return res.tv_sec * 1000 + res.tv_usec / 1000; | |
909 | } | |
910 | ||
911 | /* | |
912 | * we ride the region ref that the region dir holds. before the region | |
913 | * dir is removed and drops it ref it will wait to tear down this | |
914 | * thread. | |
915 | */ | |
916 | static int o2hb_thread(void *data) | |
917 | { | |
918 | int i, ret; | |
919 | struct o2hb_region *reg = data; | |
920 | struct bio *write_bio; | |
921 | struct o2hb_bio_wait_ctxt write_wc; | |
922 | struct timeval before_hb, after_hb; | |
923 | unsigned int elapsed_msec; | |
924 | ||
925 | mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread running\n"); | |
926 | ||
927 | set_user_nice(current, -20); | |
928 | ||
929 | while (!kthread_should_stop() && !reg->hr_unclean_stop) { | |
930 | /* We track the time spent inside | |
931 | * o2hb_do_disk_heartbeat so that we avoid more then | |
932 | * hr_timeout_ms between disk writes. On busy systems | |
933 | * this should result in a heartbeat which is less | |
934 | * likely to time itself out. */ | |
935 | do_gettimeofday(&before_hb); | |
936 | ||
a9e2ae39 MF |
937 | i = 0; |
938 | do { | |
939 | ret = o2hb_do_disk_heartbeat(reg); | |
940 | } while (ret && ++i < 2); | |
a7f6a5fb MF |
941 | |
942 | do_gettimeofday(&after_hb); | |
943 | elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb); | |
944 | ||
945 | mlog(0, "start = %lu.%lu, end = %lu.%lu, msec = %u\n", | |
215c7f9f MF |
946 | before_hb.tv_sec, (unsigned long) before_hb.tv_usec, |
947 | after_hb.tv_sec, (unsigned long) after_hb.tv_usec, | |
948 | elapsed_msec); | |
a7f6a5fb MF |
949 | |
950 | if (elapsed_msec < reg->hr_timeout_ms) { | |
951 | /* the kthread api has blocked signals for us so no | |
952 | * need to record the return value. */ | |
953 | msleep_interruptible(reg->hr_timeout_ms - elapsed_msec); | |
954 | } | |
955 | } | |
956 | ||
957 | o2hb_disarm_write_timeout(reg); | |
958 | ||
959 | /* unclean stop is only used in very bad situation */ | |
960 | for(i = 0; !reg->hr_unclean_stop && i < reg->hr_blocks; i++) | |
961 | o2hb_shutdown_slot(®->hr_slots[i]); | |
962 | ||
963 | /* Explicit down notification - avoid forcing the other nodes | |
964 | * to timeout on this region when we could just as easily | |
965 | * write a clear generation - thus indicating to them that | |
966 | * this node has left this region. | |
967 | * | |
968 | * XXX: Should we skip this on unclean_stop? */ | |
969 | o2hb_prepare_block(reg, 0); | |
970 | ret = o2hb_issue_node_write(reg, &write_bio, &write_wc); | |
971 | if (ret == 0) { | |
972 | o2hb_wait_on_io(reg, &write_wc); | |
973 | bio_put(write_bio); | |
974 | } else { | |
975 | mlog_errno(ret); | |
976 | } | |
977 | ||
978 | mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread exiting\n"); | |
979 | ||
980 | return 0; | |
981 | } | |
982 | ||
983 | void o2hb_init(void) | |
984 | { | |
985 | int i; | |
986 | ||
987 | for (i = 0; i < ARRAY_SIZE(o2hb_callbacks); i++) | |
988 | INIT_LIST_HEAD(&o2hb_callbacks[i].list); | |
989 | ||
990 | for (i = 0; i < ARRAY_SIZE(o2hb_live_slots); i++) | |
991 | INIT_LIST_HEAD(&o2hb_live_slots[i]); | |
992 | ||
993 | INIT_LIST_HEAD(&o2hb_node_events); | |
994 | ||
995 | memset(o2hb_live_node_bitmap, 0, sizeof(o2hb_live_node_bitmap)); | |
996 | } | |
997 | ||
998 | /* if we're already in a callback then we're already serialized by the sem */ | |
999 | static void o2hb_fill_node_map_from_callback(unsigned long *map, | |
1000 | unsigned bytes) | |
1001 | { | |
1002 | BUG_ON(bytes < (BITS_TO_LONGS(O2NM_MAX_NODES) * sizeof(unsigned long))); | |
1003 | ||
1004 | memcpy(map, &o2hb_live_node_bitmap, bytes); | |
1005 | } | |
1006 | ||
1007 | /* | |
1008 | * get a map of all nodes that are heartbeating in any regions | |
1009 | */ | |
1010 | void o2hb_fill_node_map(unsigned long *map, unsigned bytes) | |
1011 | { | |
1012 | /* callers want to serialize this map and callbacks so that they | |
1013 | * can trust that they don't miss nodes coming to the party */ | |
1014 | down_read(&o2hb_callback_sem); | |
1015 | spin_lock(&o2hb_live_lock); | |
1016 | o2hb_fill_node_map_from_callback(map, bytes); | |
1017 | spin_unlock(&o2hb_live_lock); | |
1018 | up_read(&o2hb_callback_sem); | |
1019 | } | |
1020 | EXPORT_SYMBOL_GPL(o2hb_fill_node_map); | |
1021 | ||
1022 | /* | |
1023 | * heartbeat configfs bits. The heartbeat set is a default set under | |
1024 | * the cluster set in nodemanager.c. | |
1025 | */ | |
1026 | ||
1027 | static struct o2hb_region *to_o2hb_region(struct config_item *item) | |
1028 | { | |
1029 | return item ? container_of(item, struct o2hb_region, hr_item) : NULL; | |
1030 | } | |
1031 | ||
1032 | /* drop_item only drops its ref after killing the thread, nothing should | |
1033 | * be using the region anymore. this has to clean up any state that | |
1034 | * attributes might have built up. */ | |
1035 | static void o2hb_region_release(struct config_item *item) | |
1036 | { | |
1037 | int i; | |
1038 | struct page *page; | |
1039 | struct o2hb_region *reg = to_o2hb_region(item); | |
1040 | ||
1041 | if (reg->hr_tmp_block) | |
1042 | kfree(reg->hr_tmp_block); | |
1043 | ||
1044 | if (reg->hr_slot_data) { | |
1045 | for (i = 0; i < reg->hr_num_pages; i++) { | |
1046 | page = reg->hr_slot_data[i]; | |
1047 | if (page) | |
1048 | __free_page(page); | |
1049 | } | |
1050 | kfree(reg->hr_slot_data); | |
1051 | } | |
1052 | ||
1053 | if (reg->hr_bdev) | |
1054 | blkdev_put(reg->hr_bdev); | |
1055 | ||
1056 | if (reg->hr_slots) | |
1057 | kfree(reg->hr_slots); | |
1058 | ||
1059 | spin_lock(&o2hb_live_lock); | |
1060 | list_del(®->hr_all_item); | |
1061 | spin_unlock(&o2hb_live_lock); | |
1062 | ||
1063 | kfree(reg); | |
1064 | } | |
1065 | ||
1066 | static int o2hb_read_block_input(struct o2hb_region *reg, | |
1067 | const char *page, | |
1068 | size_t count, | |
1069 | unsigned long *ret_bytes, | |
1070 | unsigned int *ret_bits) | |
1071 | { | |
1072 | unsigned long bytes; | |
1073 | char *p = (char *)page; | |
1074 | ||
1075 | bytes = simple_strtoul(p, &p, 0); | |
1076 | if (!p || (*p && (*p != '\n'))) | |
1077 | return -EINVAL; | |
1078 | ||
1079 | /* Heartbeat and fs min / max block sizes are the same. */ | |
1080 | if (bytes > 4096 || bytes < 512) | |
1081 | return -ERANGE; | |
1082 | if (hweight16(bytes) != 1) | |
1083 | return -EINVAL; | |
1084 | ||
1085 | if (ret_bytes) | |
1086 | *ret_bytes = bytes; | |
1087 | if (ret_bits) | |
1088 | *ret_bits = ffs(bytes) - 1; | |
1089 | ||
1090 | return 0; | |
1091 | } | |
1092 | ||
1093 | static ssize_t o2hb_region_block_bytes_read(struct o2hb_region *reg, | |
1094 | char *page) | |
1095 | { | |
1096 | return sprintf(page, "%u\n", reg->hr_block_bytes); | |
1097 | } | |
1098 | ||
1099 | static ssize_t o2hb_region_block_bytes_write(struct o2hb_region *reg, | |
1100 | const char *page, | |
1101 | size_t count) | |
1102 | { | |
1103 | int status; | |
1104 | unsigned long block_bytes; | |
1105 | unsigned int block_bits; | |
1106 | ||
1107 | if (reg->hr_bdev) | |
1108 | return -EINVAL; | |
1109 | ||
1110 | status = o2hb_read_block_input(reg, page, count, | |
1111 | &block_bytes, &block_bits); | |
1112 | if (status) | |
1113 | return status; | |
1114 | ||
1115 | reg->hr_block_bytes = (unsigned int)block_bytes; | |
1116 | reg->hr_block_bits = block_bits; | |
1117 | ||
1118 | return count; | |
1119 | } | |
1120 | ||
1121 | static ssize_t o2hb_region_start_block_read(struct o2hb_region *reg, | |
1122 | char *page) | |
1123 | { | |
1124 | return sprintf(page, "%llu\n", reg->hr_start_block); | |
1125 | } | |
1126 | ||
1127 | static ssize_t o2hb_region_start_block_write(struct o2hb_region *reg, | |
1128 | const char *page, | |
1129 | size_t count) | |
1130 | { | |
1131 | unsigned long long tmp; | |
1132 | char *p = (char *)page; | |
1133 | ||
1134 | if (reg->hr_bdev) | |
1135 | return -EINVAL; | |
1136 | ||
1137 | tmp = simple_strtoull(p, &p, 0); | |
1138 | if (!p || (*p && (*p != '\n'))) | |
1139 | return -EINVAL; | |
1140 | ||
1141 | reg->hr_start_block = tmp; | |
1142 | ||
1143 | return count; | |
1144 | } | |
1145 | ||
1146 | static ssize_t o2hb_region_blocks_read(struct o2hb_region *reg, | |
1147 | char *page) | |
1148 | { | |
1149 | return sprintf(page, "%d\n", reg->hr_blocks); | |
1150 | } | |
1151 | ||
1152 | static ssize_t o2hb_region_blocks_write(struct o2hb_region *reg, | |
1153 | const char *page, | |
1154 | size_t count) | |
1155 | { | |
1156 | unsigned long tmp; | |
1157 | char *p = (char *)page; | |
1158 | ||
1159 | if (reg->hr_bdev) | |
1160 | return -EINVAL; | |
1161 | ||
1162 | tmp = simple_strtoul(p, &p, 0); | |
1163 | if (!p || (*p && (*p != '\n'))) | |
1164 | return -EINVAL; | |
1165 | ||
1166 | if (tmp > O2NM_MAX_NODES || tmp == 0) | |
1167 | return -ERANGE; | |
1168 | ||
1169 | reg->hr_blocks = (unsigned int)tmp; | |
1170 | ||
1171 | return count; | |
1172 | } | |
1173 | ||
1174 | static ssize_t o2hb_region_dev_read(struct o2hb_region *reg, | |
1175 | char *page) | |
1176 | { | |
1177 | unsigned int ret = 0; | |
1178 | ||
1179 | if (reg->hr_bdev) | |
1180 | ret = sprintf(page, "%s\n", reg->hr_dev_name); | |
1181 | ||
1182 | return ret; | |
1183 | } | |
1184 | ||
1185 | static void o2hb_init_region_params(struct o2hb_region *reg) | |
1186 | { | |
1187 | reg->hr_slots_per_page = PAGE_CACHE_SIZE >> reg->hr_block_bits; | |
1188 | reg->hr_timeout_ms = O2HB_REGION_TIMEOUT_MS; | |
1189 | ||
1190 | mlog(ML_HEARTBEAT, "hr_start_block = %llu, hr_blocks = %u\n", | |
1191 | reg->hr_start_block, reg->hr_blocks); | |
1192 | mlog(ML_HEARTBEAT, "hr_block_bytes = %u, hr_block_bits = %u\n", | |
1193 | reg->hr_block_bytes, reg->hr_block_bits); | |
1194 | mlog(ML_HEARTBEAT, "hr_timeout_ms = %u\n", reg->hr_timeout_ms); | |
1195 | mlog(ML_HEARTBEAT, "dead threshold = %u\n", o2hb_dead_threshold); | |
1196 | } | |
1197 | ||
1198 | static int o2hb_map_slot_data(struct o2hb_region *reg) | |
1199 | { | |
1200 | int i, j; | |
1201 | unsigned int last_slot; | |
1202 | unsigned int spp = reg->hr_slots_per_page; | |
1203 | struct page *page; | |
1204 | char *raw; | |
1205 | struct o2hb_disk_slot *slot; | |
1206 | ||
1207 | reg->hr_tmp_block = kmalloc(reg->hr_block_bytes, GFP_KERNEL); | |
1208 | if (reg->hr_tmp_block == NULL) { | |
1209 | mlog_errno(-ENOMEM); | |
1210 | return -ENOMEM; | |
1211 | } | |
1212 | ||
1213 | reg->hr_slots = kcalloc(reg->hr_blocks, | |
1214 | sizeof(struct o2hb_disk_slot), GFP_KERNEL); | |
1215 | if (reg->hr_slots == NULL) { | |
1216 | mlog_errno(-ENOMEM); | |
1217 | return -ENOMEM; | |
1218 | } | |
1219 | ||
1220 | for(i = 0; i < reg->hr_blocks; i++) { | |
1221 | slot = ®->hr_slots[i]; | |
1222 | slot->ds_node_num = i; | |
1223 | INIT_LIST_HEAD(&slot->ds_live_item); | |
1224 | slot->ds_raw_block = NULL; | |
1225 | } | |
1226 | ||
1227 | reg->hr_num_pages = (reg->hr_blocks + spp - 1) / spp; | |
1228 | mlog(ML_HEARTBEAT, "Going to require %u pages to cover %u blocks " | |
1229 | "at %u blocks per page\n", | |
1230 | reg->hr_num_pages, reg->hr_blocks, spp); | |
1231 | ||
1232 | reg->hr_slot_data = kcalloc(reg->hr_num_pages, sizeof(struct page *), | |
1233 | GFP_KERNEL); | |
1234 | if (!reg->hr_slot_data) { | |
1235 | mlog_errno(-ENOMEM); | |
1236 | return -ENOMEM; | |
1237 | } | |
1238 | ||
1239 | for(i = 0; i < reg->hr_num_pages; i++) { | |
1240 | page = alloc_page(GFP_KERNEL); | |
1241 | if (!page) { | |
1242 | mlog_errno(-ENOMEM); | |
1243 | return -ENOMEM; | |
1244 | } | |
1245 | ||
1246 | reg->hr_slot_data[i] = page; | |
1247 | ||
1248 | last_slot = i * spp; | |
1249 | raw = page_address(page); | |
1250 | for (j = 0; | |
1251 | (j < spp) && ((j + last_slot) < reg->hr_blocks); | |
1252 | j++) { | |
1253 | BUG_ON((j + last_slot) >= reg->hr_blocks); | |
1254 | ||
1255 | slot = ®->hr_slots[j + last_slot]; | |
1256 | slot->ds_raw_block = | |
1257 | (struct o2hb_disk_heartbeat_block *) raw; | |
1258 | ||
1259 | raw += reg->hr_block_bytes; | |
1260 | } | |
1261 | } | |
1262 | ||
1263 | return 0; | |
1264 | } | |
1265 | ||
1266 | /* Read in all the slots available and populate the tracking | |
1267 | * structures so that we can start with a baseline idea of what's | |
1268 | * there. */ | |
1269 | static int o2hb_populate_slot_data(struct o2hb_region *reg) | |
1270 | { | |
1271 | int ret, i; | |
1272 | struct o2hb_disk_slot *slot; | |
1273 | struct o2hb_disk_heartbeat_block *hb_block; | |
1274 | ||
1275 | mlog_entry_void(); | |
1276 | ||
1277 | ret = o2hb_read_slots(reg, reg->hr_blocks); | |
1278 | if (ret) { | |
1279 | mlog_errno(ret); | |
1280 | goto out; | |
1281 | } | |
1282 | ||
1283 | /* We only want to get an idea of the values initially in each | |
1284 | * slot, so we do no verification - o2hb_check_slot will | |
1285 | * actually determine if each configured slot is valid and | |
1286 | * whether any values have changed. */ | |
1287 | for(i = 0; i < reg->hr_blocks; i++) { | |
1288 | slot = ®->hr_slots[i]; | |
1289 | hb_block = (struct o2hb_disk_heartbeat_block *) slot->ds_raw_block; | |
1290 | ||
1291 | /* Only fill the values that o2hb_check_slot uses to | |
1292 | * determine changing slots */ | |
1293 | slot->ds_last_time = le64_to_cpu(hb_block->hb_seq); | |
1294 | slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation); | |
1295 | } | |
1296 | ||
1297 | out: | |
1298 | mlog_exit(ret); | |
1299 | return ret; | |
1300 | } | |
1301 | ||
1302 | /* this is acting as commit; we set up all of hr_bdev and hr_task or nothing */ | |
1303 | static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, | |
1304 | const char *page, | |
1305 | size_t count) | |
1306 | { | |
1307 | long fd; | |
1308 | int sectsize; | |
1309 | char *p = (char *)page; | |
1310 | struct file *filp = NULL; | |
1311 | struct inode *inode = NULL; | |
1312 | ssize_t ret = -EINVAL; | |
1313 | ||
1314 | if (reg->hr_bdev) | |
1315 | goto out; | |
1316 | ||
1317 | /* We can't heartbeat without having had our node number | |
1318 | * configured yet. */ | |
1319 | if (o2nm_this_node() == O2NM_MAX_NODES) | |
1320 | goto out; | |
1321 | ||
1322 | fd = simple_strtol(p, &p, 0); | |
1323 | if (!p || (*p && (*p != '\n'))) | |
1324 | goto out; | |
1325 | ||
1326 | if (fd < 0 || fd >= INT_MAX) | |
1327 | goto out; | |
1328 | ||
1329 | filp = fget(fd); | |
1330 | if (filp == NULL) | |
1331 | goto out; | |
1332 | ||
1333 | if (reg->hr_blocks == 0 || reg->hr_start_block == 0 || | |
1334 | reg->hr_block_bytes == 0) | |
1335 | goto out; | |
1336 | ||
1337 | inode = igrab(filp->f_mapping->host); | |
1338 | if (inode == NULL) | |
1339 | goto out; | |
1340 | ||
1341 | if (!S_ISBLK(inode->i_mode)) | |
1342 | goto out; | |
1343 | ||
1344 | reg->hr_bdev = I_BDEV(filp->f_mapping->host); | |
1345 | ret = blkdev_get(reg->hr_bdev, FMODE_WRITE | FMODE_READ, 0); | |
1346 | if (ret) { | |
1347 | reg->hr_bdev = NULL; | |
1348 | goto out; | |
1349 | } | |
1350 | inode = NULL; | |
1351 | ||
1352 | bdevname(reg->hr_bdev, reg->hr_dev_name); | |
1353 | ||
1354 | sectsize = bdev_hardsect_size(reg->hr_bdev); | |
1355 | if (sectsize != reg->hr_block_bytes) { | |
1356 | mlog(ML_ERROR, | |
1357 | "blocksize %u incorrect for device, expected %d", | |
1358 | reg->hr_block_bytes, sectsize); | |
1359 | ret = -EINVAL; | |
1360 | goto out; | |
1361 | } | |
1362 | ||
1363 | o2hb_init_region_params(reg); | |
1364 | ||
1365 | /* Generation of zero is invalid */ | |
1366 | do { | |
1367 | get_random_bytes(®->hr_generation, | |
1368 | sizeof(reg->hr_generation)); | |
1369 | } while (reg->hr_generation == 0); | |
1370 | ||
1371 | ret = o2hb_map_slot_data(reg); | |
1372 | if (ret) { | |
1373 | mlog_errno(ret); | |
1374 | goto out; | |
1375 | } | |
1376 | ||
1377 | ret = o2hb_populate_slot_data(reg); | |
1378 | if (ret) { | |
1379 | mlog_errno(ret); | |
1380 | goto out; | |
1381 | } | |
1382 | ||
1383 | INIT_WORK(®->hr_write_timeout_work, o2hb_write_timeout, reg); | |
1384 | ||
1385 | /* | |
1386 | * A node is considered live after it has beat LIVE_THRESHOLD | |
1387 | * times. We're not steady until we've given them a chance | |
1388 | * _after_ our first read. | |
1389 | */ | |
1390 | atomic_set(®->hr_steady_iterations, O2HB_LIVE_THRESHOLD + 1); | |
1391 | ||
1392 | reg->hr_task = kthread_run(o2hb_thread, reg, "o2hb-%s", | |
1393 | reg->hr_item.ci_name); | |
1394 | if (IS_ERR(reg->hr_task)) { | |
1395 | ret = PTR_ERR(reg->hr_task); | |
1396 | mlog_errno(ret); | |
1397 | reg->hr_task = NULL; | |
1398 | goto out; | |
1399 | } | |
1400 | ||
1401 | ret = wait_event_interruptible(o2hb_steady_queue, | |
1402 | atomic_read(®->hr_steady_iterations) == 0); | |
1403 | if (ret) { | |
1404 | kthread_stop(reg->hr_task); | |
1405 | reg->hr_task = NULL; | |
1406 | goto out; | |
1407 | } | |
1408 | ||
1409 | ret = count; | |
1410 | out: | |
1411 | if (filp) | |
1412 | fput(filp); | |
1413 | if (inode) | |
1414 | iput(inode); | |
1415 | if (ret < 0) { | |
1416 | if (reg->hr_bdev) { | |
1417 | blkdev_put(reg->hr_bdev); | |
1418 | reg->hr_bdev = NULL; | |
1419 | } | |
1420 | } | |
1421 | return ret; | |
1422 | } | |
1423 | ||
1424 | struct o2hb_region_attribute { | |
1425 | struct configfs_attribute attr; | |
1426 | ssize_t (*show)(struct o2hb_region *, char *); | |
1427 | ssize_t (*store)(struct o2hb_region *, const char *, size_t); | |
1428 | }; | |
1429 | ||
1430 | static struct o2hb_region_attribute o2hb_region_attr_block_bytes = { | |
1431 | .attr = { .ca_owner = THIS_MODULE, | |
1432 | .ca_name = "block_bytes", | |
1433 | .ca_mode = S_IRUGO | S_IWUSR }, | |
1434 | .show = o2hb_region_block_bytes_read, | |
1435 | .store = o2hb_region_block_bytes_write, | |
1436 | }; | |
1437 | ||
1438 | static struct o2hb_region_attribute o2hb_region_attr_start_block = { | |
1439 | .attr = { .ca_owner = THIS_MODULE, | |
1440 | .ca_name = "start_block", | |
1441 | .ca_mode = S_IRUGO | S_IWUSR }, | |
1442 | .show = o2hb_region_start_block_read, | |
1443 | .store = o2hb_region_start_block_write, | |
1444 | }; | |
1445 | ||
1446 | static struct o2hb_region_attribute o2hb_region_attr_blocks = { | |
1447 | .attr = { .ca_owner = THIS_MODULE, | |
1448 | .ca_name = "blocks", | |
1449 | .ca_mode = S_IRUGO | S_IWUSR }, | |
1450 | .show = o2hb_region_blocks_read, | |
1451 | .store = o2hb_region_blocks_write, | |
1452 | }; | |
1453 | ||
1454 | static struct o2hb_region_attribute o2hb_region_attr_dev = { | |
1455 | .attr = { .ca_owner = THIS_MODULE, | |
1456 | .ca_name = "dev", | |
1457 | .ca_mode = S_IRUGO | S_IWUSR }, | |
1458 | .show = o2hb_region_dev_read, | |
1459 | .store = o2hb_region_dev_write, | |
1460 | }; | |
1461 | ||
1462 | static struct configfs_attribute *o2hb_region_attrs[] = { | |
1463 | &o2hb_region_attr_block_bytes.attr, | |
1464 | &o2hb_region_attr_start_block.attr, | |
1465 | &o2hb_region_attr_blocks.attr, | |
1466 | &o2hb_region_attr_dev.attr, | |
1467 | NULL, | |
1468 | }; | |
1469 | ||
1470 | static ssize_t o2hb_region_show(struct config_item *item, | |
1471 | struct configfs_attribute *attr, | |
1472 | char *page) | |
1473 | { | |
1474 | struct o2hb_region *reg = to_o2hb_region(item); | |
1475 | struct o2hb_region_attribute *o2hb_region_attr = | |
1476 | container_of(attr, struct o2hb_region_attribute, attr); | |
1477 | ssize_t ret = 0; | |
1478 | ||
1479 | if (o2hb_region_attr->show) | |
1480 | ret = o2hb_region_attr->show(reg, page); | |
1481 | return ret; | |
1482 | } | |
1483 | ||
1484 | static ssize_t o2hb_region_store(struct config_item *item, | |
1485 | struct configfs_attribute *attr, | |
1486 | const char *page, size_t count) | |
1487 | { | |
1488 | struct o2hb_region *reg = to_o2hb_region(item); | |
1489 | struct o2hb_region_attribute *o2hb_region_attr = | |
1490 | container_of(attr, struct o2hb_region_attribute, attr); | |
1491 | ssize_t ret = -EINVAL; | |
1492 | ||
1493 | if (o2hb_region_attr->store) | |
1494 | ret = o2hb_region_attr->store(reg, page, count); | |
1495 | return ret; | |
1496 | } | |
1497 | ||
1498 | static struct configfs_item_operations o2hb_region_item_ops = { | |
1499 | .release = o2hb_region_release, | |
1500 | .show_attribute = o2hb_region_show, | |
1501 | .store_attribute = o2hb_region_store, | |
1502 | }; | |
1503 | ||
1504 | static struct config_item_type o2hb_region_type = { | |
1505 | .ct_item_ops = &o2hb_region_item_ops, | |
1506 | .ct_attrs = o2hb_region_attrs, | |
1507 | .ct_owner = THIS_MODULE, | |
1508 | }; | |
1509 | ||
1510 | /* heartbeat set */ | |
1511 | ||
1512 | struct o2hb_heartbeat_group { | |
1513 | struct config_group hs_group; | |
1514 | /* some stuff? */ | |
1515 | }; | |
1516 | ||
1517 | static struct o2hb_heartbeat_group *to_o2hb_heartbeat_group(struct config_group *group) | |
1518 | { | |
1519 | return group ? | |
1520 | container_of(group, struct o2hb_heartbeat_group, hs_group) | |
1521 | : NULL; | |
1522 | } | |
1523 | ||
1524 | static struct config_item *o2hb_heartbeat_group_make_item(struct config_group *group, | |
1525 | const char *name) | |
1526 | { | |
1527 | struct o2hb_region *reg = NULL; | |
1528 | struct config_item *ret = NULL; | |
1529 | ||
1530 | reg = kcalloc(1, sizeof(struct o2hb_region), GFP_KERNEL); | |
1531 | if (reg == NULL) | |
1532 | goto out; /* ENOMEM */ | |
1533 | ||
1534 | config_item_init_type_name(®->hr_item, name, &o2hb_region_type); | |
1535 | ||
1536 | ret = ®->hr_item; | |
1537 | ||
1538 | spin_lock(&o2hb_live_lock); | |
1539 | list_add_tail(®->hr_all_item, &o2hb_all_regions); | |
1540 | spin_unlock(&o2hb_live_lock); | |
1541 | out: | |
1542 | if (ret == NULL) | |
1543 | kfree(reg); | |
1544 | ||
1545 | return ret; | |
1546 | } | |
1547 | ||
1548 | static void o2hb_heartbeat_group_drop_item(struct config_group *group, | |
1549 | struct config_item *item) | |
1550 | { | |
1551 | struct o2hb_region *reg = to_o2hb_region(item); | |
1552 | ||
1553 | /* stop the thread when the user removes the region dir */ | |
1554 | if (reg->hr_task) { | |
1555 | kthread_stop(reg->hr_task); | |
1556 | reg->hr_task = NULL; | |
1557 | } | |
1558 | ||
1559 | config_item_put(item); | |
1560 | } | |
1561 | ||
1562 | struct o2hb_heartbeat_group_attribute { | |
1563 | struct configfs_attribute attr; | |
1564 | ssize_t (*show)(struct o2hb_heartbeat_group *, char *); | |
1565 | ssize_t (*store)(struct o2hb_heartbeat_group *, const char *, size_t); | |
1566 | }; | |
1567 | ||
1568 | static ssize_t o2hb_heartbeat_group_show(struct config_item *item, | |
1569 | struct configfs_attribute *attr, | |
1570 | char *page) | |
1571 | { | |
1572 | struct o2hb_heartbeat_group *reg = to_o2hb_heartbeat_group(to_config_group(item)); | |
1573 | struct o2hb_heartbeat_group_attribute *o2hb_heartbeat_group_attr = | |
1574 | container_of(attr, struct o2hb_heartbeat_group_attribute, attr); | |
1575 | ssize_t ret = 0; | |
1576 | ||
1577 | if (o2hb_heartbeat_group_attr->show) | |
1578 | ret = o2hb_heartbeat_group_attr->show(reg, page); | |
1579 | return ret; | |
1580 | } | |
1581 | ||
1582 | static ssize_t o2hb_heartbeat_group_store(struct config_item *item, | |
1583 | struct configfs_attribute *attr, | |
1584 | const char *page, size_t count) | |
1585 | { | |
1586 | struct o2hb_heartbeat_group *reg = to_o2hb_heartbeat_group(to_config_group(item)); | |
1587 | struct o2hb_heartbeat_group_attribute *o2hb_heartbeat_group_attr = | |
1588 | container_of(attr, struct o2hb_heartbeat_group_attribute, attr); | |
1589 | ssize_t ret = -EINVAL; | |
1590 | ||
1591 | if (o2hb_heartbeat_group_attr->store) | |
1592 | ret = o2hb_heartbeat_group_attr->store(reg, page, count); | |
1593 | return ret; | |
1594 | } | |
1595 | ||
1596 | static ssize_t o2hb_heartbeat_group_threshold_show(struct o2hb_heartbeat_group *group, | |
1597 | char *page) | |
1598 | { | |
1599 | return sprintf(page, "%u\n", o2hb_dead_threshold); | |
1600 | } | |
1601 | ||
1602 | static ssize_t o2hb_heartbeat_group_threshold_store(struct o2hb_heartbeat_group *group, | |
1603 | const char *page, | |
1604 | size_t count) | |
1605 | { | |
1606 | unsigned long tmp; | |
1607 | char *p = (char *)page; | |
1608 | ||
1609 | tmp = simple_strtoul(p, &p, 10); | |
1610 | if (!p || (*p && (*p != '\n'))) | |
1611 | return -EINVAL; | |
1612 | ||
1613 | /* this will validate ranges for us. */ | |
1614 | o2hb_dead_threshold_set((unsigned int) tmp); | |
1615 | ||
1616 | return count; | |
1617 | } | |
1618 | ||
1619 | static struct o2hb_heartbeat_group_attribute o2hb_heartbeat_group_attr_threshold = { | |
1620 | .attr = { .ca_owner = THIS_MODULE, | |
1621 | .ca_name = "dead_threshold", | |
1622 | .ca_mode = S_IRUGO | S_IWUSR }, | |
1623 | .show = o2hb_heartbeat_group_threshold_show, | |
1624 | .store = o2hb_heartbeat_group_threshold_store, | |
1625 | }; | |
1626 | ||
1627 | static struct configfs_attribute *o2hb_heartbeat_group_attrs[] = { | |
1628 | &o2hb_heartbeat_group_attr_threshold.attr, | |
1629 | NULL, | |
1630 | }; | |
1631 | ||
1632 | static struct configfs_item_operations o2hb_hearbeat_group_item_ops = { | |
1633 | .show_attribute = o2hb_heartbeat_group_show, | |
1634 | .store_attribute = o2hb_heartbeat_group_store, | |
1635 | }; | |
1636 | ||
1637 | static struct configfs_group_operations o2hb_heartbeat_group_group_ops = { | |
1638 | .make_item = o2hb_heartbeat_group_make_item, | |
1639 | .drop_item = o2hb_heartbeat_group_drop_item, | |
1640 | }; | |
1641 | ||
1642 | static struct config_item_type o2hb_heartbeat_group_type = { | |
1643 | .ct_group_ops = &o2hb_heartbeat_group_group_ops, | |
1644 | .ct_item_ops = &o2hb_hearbeat_group_item_ops, | |
1645 | .ct_attrs = o2hb_heartbeat_group_attrs, | |
1646 | .ct_owner = THIS_MODULE, | |
1647 | }; | |
1648 | ||
1649 | /* this is just here to avoid touching group in heartbeat.h which the | |
1650 | * entire damn world #includes */ | |
1651 | struct config_group *o2hb_alloc_hb_set(void) | |
1652 | { | |
1653 | struct o2hb_heartbeat_group *hs = NULL; | |
1654 | struct config_group *ret = NULL; | |
1655 | ||
1656 | hs = kcalloc(1, sizeof(struct o2hb_heartbeat_group), GFP_KERNEL); | |
1657 | if (hs == NULL) | |
1658 | goto out; | |
1659 | ||
1660 | config_group_init_type_name(&hs->hs_group, "heartbeat", | |
1661 | &o2hb_heartbeat_group_type); | |
1662 | ||
1663 | ret = &hs->hs_group; | |
1664 | out: | |
1665 | if (ret == NULL) | |
1666 | kfree(hs); | |
1667 | return ret; | |
1668 | } | |
1669 | ||
1670 | void o2hb_free_hb_set(struct config_group *group) | |
1671 | { | |
1672 | struct o2hb_heartbeat_group *hs = to_o2hb_heartbeat_group(group); | |
1673 | kfree(hs); | |
1674 | } | |
1675 | ||
1676 | /* hb callback registration and issueing */ | |
1677 | ||
1678 | static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type) | |
1679 | { | |
1680 | if (type == O2HB_NUM_CB) | |
1681 | return ERR_PTR(-EINVAL); | |
1682 | ||
1683 | return &o2hb_callbacks[type]; | |
1684 | } | |
1685 | ||
1686 | void o2hb_setup_callback(struct o2hb_callback_func *hc, | |
1687 | enum o2hb_callback_type type, | |
1688 | o2hb_cb_func *func, | |
1689 | void *data, | |
1690 | int priority) | |
1691 | { | |
1692 | INIT_LIST_HEAD(&hc->hc_item); | |
1693 | hc->hc_func = func; | |
1694 | hc->hc_data = data; | |
1695 | hc->hc_priority = priority; | |
1696 | hc->hc_type = type; | |
1697 | hc->hc_magic = O2HB_CB_MAGIC; | |
1698 | } | |
1699 | EXPORT_SYMBOL_GPL(o2hb_setup_callback); | |
1700 | ||
1701 | int o2hb_register_callback(struct o2hb_callback_func *hc) | |
1702 | { | |
1703 | struct o2hb_callback_func *tmp; | |
1704 | struct list_head *iter; | |
1705 | struct o2hb_callback *hbcall; | |
1706 | int ret; | |
1707 | ||
1708 | BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); | |
1709 | BUG_ON(!list_empty(&hc->hc_item)); | |
1710 | ||
1711 | hbcall = hbcall_from_type(hc->hc_type); | |
1712 | if (IS_ERR(hbcall)) { | |
1713 | ret = PTR_ERR(hbcall); | |
1714 | goto out; | |
1715 | } | |
1716 | ||
1717 | down_write(&o2hb_callback_sem); | |
1718 | ||
1719 | list_for_each(iter, &hbcall->list) { | |
1720 | tmp = list_entry(iter, struct o2hb_callback_func, hc_item); | |
1721 | if (hc->hc_priority < tmp->hc_priority) { | |
1722 | list_add_tail(&hc->hc_item, iter); | |
1723 | break; | |
1724 | } | |
1725 | } | |
1726 | if (list_empty(&hc->hc_item)) | |
1727 | list_add_tail(&hc->hc_item, &hbcall->list); | |
1728 | ||
1729 | up_write(&o2hb_callback_sem); | |
1730 | ret = 0; | |
1731 | out: | |
1732 | mlog(ML_HEARTBEAT, "returning %d on behalf of %p for funcs %p\n", | |
1733 | ret, __builtin_return_address(0), hc); | |
1734 | return ret; | |
1735 | } | |
1736 | EXPORT_SYMBOL_GPL(o2hb_register_callback); | |
1737 | ||
1738 | int o2hb_unregister_callback(struct o2hb_callback_func *hc) | |
1739 | { | |
1740 | BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); | |
1741 | ||
1742 | mlog(ML_HEARTBEAT, "on behalf of %p for funcs %p\n", | |
1743 | __builtin_return_address(0), hc); | |
1744 | ||
1745 | if (list_empty(&hc->hc_item)) | |
1746 | return 0; | |
1747 | ||
1748 | down_write(&o2hb_callback_sem); | |
1749 | ||
1750 | list_del_init(&hc->hc_item); | |
1751 | ||
1752 | up_write(&o2hb_callback_sem); | |
1753 | ||
1754 | return 0; | |
1755 | } | |
1756 | EXPORT_SYMBOL_GPL(o2hb_unregister_callback); | |
1757 | ||
1758 | int o2hb_check_node_heartbeating(u8 node_num) | |
1759 | { | |
1760 | unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)]; | |
1761 | ||
1762 | o2hb_fill_node_map(testing_map, sizeof(testing_map)); | |
1763 | if (!test_bit(node_num, testing_map)) { | |
1764 | mlog(ML_HEARTBEAT, | |
1765 | "node (%u) does not have heartbeating enabled.\n", | |
1766 | node_num); | |
1767 | return 0; | |
1768 | } | |
1769 | ||
1770 | return 1; | |
1771 | } | |
1772 | EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating); | |
1773 | ||
1774 | int o2hb_check_node_heartbeating_from_callback(u8 node_num) | |
1775 | { | |
1776 | unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)]; | |
1777 | ||
1778 | o2hb_fill_node_map_from_callback(testing_map, sizeof(testing_map)); | |
1779 | if (!test_bit(node_num, testing_map)) { | |
1780 | mlog(ML_HEARTBEAT, | |
1781 | "node (%u) does not have heartbeating enabled.\n", | |
1782 | node_num); | |
1783 | return 0; | |
1784 | } | |
1785 | ||
1786 | return 1; | |
1787 | } | |
1788 | EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating_from_callback); | |
1789 | ||
1790 | /* Makes sure our local node is configured with a node number, and is | |
1791 | * heartbeating. */ | |
1792 | int o2hb_check_local_node_heartbeating(void) | |
1793 | { | |
1794 | u8 node_num; | |
1795 | ||
1796 | /* if this node was set then we have networking */ | |
1797 | node_num = o2nm_this_node(); | |
1798 | if (node_num == O2NM_MAX_NODES) { | |
1799 | mlog(ML_HEARTBEAT, "this node has not been configured.\n"); | |
1800 | return 0; | |
1801 | } | |
1802 | ||
1803 | return o2hb_check_node_heartbeating(node_num); | |
1804 | } | |
1805 | EXPORT_SYMBOL_GPL(o2hb_check_local_node_heartbeating); | |
1806 | ||
1807 | /* | |
1808 | * this is just a hack until we get the plumbing which flips file systems | |
1809 | * read only and drops the hb ref instead of killing the node dead. | |
1810 | */ | |
1811 | void o2hb_stop_all_regions(void) | |
1812 | { | |
1813 | struct o2hb_region *reg; | |
1814 | ||
1815 | mlog(ML_ERROR, "stopping heartbeat on all active regions.\n"); | |
1816 | ||
1817 | spin_lock(&o2hb_live_lock); | |
1818 | ||
1819 | list_for_each_entry(reg, &o2hb_all_regions, hr_all_item) | |
1820 | reg->hr_unclean_stop = 1; | |
1821 | ||
1822 | spin_unlock(&o2hb_live_lock); | |
1823 | } | |
1824 | EXPORT_SYMBOL_GPL(o2hb_stop_all_regions); |