1 // SPDX-License-Identifier: GPL-2.0-or-later 2 /* 3 * Copyright (C) 2004, 2005 Oracle. All rights reserved. 4 */ 5 6 #include <linux/kernel.h> 7 #include <linux/sched.h> 8 #include <linux/jiffies.h> 9 #include <linux/module.h> 10 #include <linux/fs.h> 11 #include <linux/bio.h> 12 #include <linux/blkdev.h> 13 #include <linux/delay.h> 14 #include <linux/file.h> 15 #include <linux/kthread.h> 16 #include <linux/configfs.h> 17 #include <linux/random.h> 18 #include <linux/crc32.h> 19 #include <linux/time.h> 20 #include <linux/debugfs.h> 21 #include <linux/slab.h> 22 #include <linux/bitmap.h> 23 #include <linux/ktime.h> 24 #include "heartbeat.h" 25 #include "tcp.h" 26 #include "nodemanager.h" 27 #include "quorum.h" 28 29 #include "masklog.h" 30 31 32 /* 33 * The first heartbeat pass had one global thread that would serialize all hb 34 * callback calls. This global serializing sem should only be removed once 35 * we've made sure that all callees can deal with being called concurrently 36 * from multiple hb region threads. 37 */ 38 static DECLARE_RWSEM(o2hb_callback_sem); 39 40 /* 41 * multiple hb threads are watching multiple regions. A node is live 42 * whenever any of the threads sees activity from the node in its region. 43 */ 44 static DEFINE_SPINLOCK(o2hb_live_lock); 45 static struct list_head o2hb_live_slots[O2NM_MAX_NODES]; 46 static unsigned long o2hb_live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 47 static LIST_HEAD(o2hb_node_events); 48 static DECLARE_WAIT_QUEUE_HEAD(o2hb_steady_queue); 49 50 /* 51 * In global heartbeat, we maintain a series of region bitmaps. 52 * - o2hb_region_bitmap allows us to limit the region number to max region. 53 * - o2hb_live_region_bitmap tracks live regions (seen steady iterations). 54 * - o2hb_quorum_region_bitmap tracks live regions that have seen all nodes 55 * heartbeat on it. 56 * - o2hb_failed_region_bitmap tracks the regions that have seen io timeouts. 57 */ 58 static unsigned long o2hb_region_bitmap[BITS_TO_LONGS(O2NM_MAX_REGIONS)]; 59 static unsigned long o2hb_live_region_bitmap[BITS_TO_LONGS(O2NM_MAX_REGIONS)]; 60 static unsigned long o2hb_quorum_region_bitmap[BITS_TO_LONGS(O2NM_MAX_REGIONS)]; 61 static unsigned long o2hb_failed_region_bitmap[BITS_TO_LONGS(O2NM_MAX_REGIONS)]; 62 63 #define O2HB_DB_TYPE_LIVENODES 0 64 #define O2HB_DB_TYPE_LIVEREGIONS 1 65 #define O2HB_DB_TYPE_QUORUMREGIONS 2 66 #define O2HB_DB_TYPE_FAILEDREGIONS 3 67 #define O2HB_DB_TYPE_REGION_LIVENODES 4 68 #define O2HB_DB_TYPE_REGION_NUMBER 5 69 #define O2HB_DB_TYPE_REGION_ELAPSED_TIME 6 70 #define O2HB_DB_TYPE_REGION_PINNED 7 71 struct o2hb_debug_buf { 72 int db_type; 73 int db_size; 74 int db_len; 75 void *db_data; 76 }; 77 78 static struct o2hb_debug_buf *o2hb_db_livenodes; 79 static struct o2hb_debug_buf *o2hb_db_liveregions; 80 static struct o2hb_debug_buf *o2hb_db_quorumregions; 81 static struct o2hb_debug_buf *o2hb_db_failedregions; 82 83 #define O2HB_DEBUG_DIR "o2hb" 84 #define O2HB_DEBUG_LIVENODES "livenodes" 85 #define O2HB_DEBUG_LIVEREGIONS "live_regions" 86 #define O2HB_DEBUG_QUORUMREGIONS "quorum_regions" 87 #define O2HB_DEBUG_FAILEDREGIONS "failed_regions" 88 #define O2HB_DEBUG_REGION_NUMBER "num" 89 #define O2HB_DEBUG_REGION_ELAPSED_TIME "elapsed_time_in_ms" 90 #define O2HB_DEBUG_REGION_PINNED "pinned" 91 92 static struct dentry *o2hb_debug_dir; 93 94 static LIST_HEAD(o2hb_all_regions); 95 96 static struct o2hb_callback { 97 struct list_head list; 98 } o2hb_callbacks[O2HB_NUM_CB]; 99 100 static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type); 101 102 enum o2hb_heartbeat_modes { 103 O2HB_HEARTBEAT_LOCAL = 0, 104 O2HB_HEARTBEAT_GLOBAL, 105 O2HB_HEARTBEAT_NUM_MODES, 106 }; 107 108 static const char *o2hb_heartbeat_mode_desc[O2HB_HEARTBEAT_NUM_MODES] = { 109 "local", /* O2HB_HEARTBEAT_LOCAL */ 110 "global", /* O2HB_HEARTBEAT_GLOBAL */ 111 }; 112 113 unsigned int o2hb_dead_threshold = O2HB_DEFAULT_DEAD_THRESHOLD; 114 static unsigned int o2hb_heartbeat_mode = O2HB_HEARTBEAT_LOCAL; 115 116 /* 117 * o2hb_dependent_users tracks the number of registered callbacks that depend 118 * on heartbeat. o2net and o2dlm are two entities that register this callback. 119 * However only o2dlm depends on the heartbeat. It does not want the heartbeat 120 * to stop while a dlm domain is still active. 121 */ 122 static unsigned int o2hb_dependent_users; 123 124 /* 125 * In global heartbeat mode, all regions are pinned if there are one or more 126 * dependent users and the quorum region count is <= O2HB_PIN_CUT_OFF. All 127 * regions are unpinned if the region count exceeds the cut off or the number 128 * of dependent users falls to zero. 129 */ 130 #define O2HB_PIN_CUT_OFF 3 131 132 /* 133 * In local heartbeat mode, we assume the dlm domain name to be the same as 134 * region uuid. This is true for domains created for the file system but not 135 * necessarily true for userdlm domains. This is a known limitation. 136 * 137 * In global heartbeat mode, we pin/unpin all o2hb regions. This solution 138 * works for both file system and userdlm domains. 139 */ 140 static int o2hb_region_pin(const char *region_uuid); 141 static void o2hb_region_unpin(const char *region_uuid); 142 143 /* Only sets a new threshold if there are no active regions. 144 * 145 * No locking or otherwise interesting code is required for reading 146 * o2hb_dead_threshold as it can't change once regions are active and 147 * it's not interesting to anyone until then anyway. */ 148 static void o2hb_dead_threshold_set(unsigned int threshold) 149 { 150 if (threshold > O2HB_MIN_DEAD_THRESHOLD) { 151 spin_lock(&o2hb_live_lock); 152 if (list_empty(&o2hb_all_regions)) 153 o2hb_dead_threshold = threshold; 154 spin_unlock(&o2hb_live_lock); 155 } 156 } 157 158 static int o2hb_global_heartbeat_mode_set(unsigned int hb_mode) 159 { 160 int ret = -1; 161 162 if (hb_mode < O2HB_HEARTBEAT_NUM_MODES) { 163 spin_lock(&o2hb_live_lock); 164 if (list_empty(&o2hb_all_regions)) { 165 o2hb_heartbeat_mode = hb_mode; 166 ret = 0; 167 } 168 spin_unlock(&o2hb_live_lock); 169 } 170 171 return ret; 172 } 173 174 struct o2hb_node_event { 175 struct list_head hn_item; 176 enum o2hb_callback_type hn_event_type; 177 struct o2nm_node *hn_node; 178 int hn_node_num; 179 }; 180 181 struct o2hb_disk_slot { 182 struct o2hb_disk_heartbeat_block *ds_raw_block; 183 u8 ds_node_num; 184 u64 ds_last_time; 185 u64 ds_last_generation; 186 u16 ds_equal_samples; 187 u16 ds_changed_samples; 188 struct list_head ds_live_item; 189 }; 190 191 /* each thread owns a region.. when we're asked to tear down the region 192 * we ask the thread to stop, who cleans up the region */ 193 struct o2hb_region { 194 struct config_item hr_item; 195 196 struct list_head hr_all_item; 197 unsigned hr_unclean_stop:1, 198 hr_aborted_start:1, 199 hr_item_pinned:1, 200 hr_item_dropped:1, 201 hr_node_deleted:1; 202 203 /* protected by the hr_callback_sem */ 204 struct task_struct *hr_task; 205 206 unsigned int hr_blocks; 207 unsigned long long hr_start_block; 208 209 unsigned int hr_block_bits; 210 unsigned int hr_block_bytes; 211 212 unsigned int hr_slots_per_page; 213 unsigned int hr_num_pages; 214 215 struct page **hr_slot_data; 216 struct block_device *hr_bdev; 217 struct o2hb_disk_slot *hr_slots; 218 219 /* live node map of this region */ 220 unsigned long hr_live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 221 unsigned int hr_region_num; 222 223 struct dentry *hr_debug_dir; 224 struct o2hb_debug_buf *hr_db_livenodes; 225 struct o2hb_debug_buf *hr_db_regnum; 226 struct o2hb_debug_buf *hr_db_elapsed_time; 227 struct o2hb_debug_buf *hr_db_pinned; 228 229 /* let the person setting up hb wait for it to return until it 230 * has reached a 'steady' state. This will be fixed when we have 231 * a more complete api that doesn't lead to this sort of fragility. */ 232 atomic_t hr_steady_iterations; 233 234 /* terminate o2hb thread if it does not reach steady state 235 * (hr_steady_iterations == 0) within hr_unsteady_iterations */ 236 atomic_t hr_unsteady_iterations; 237 238 char hr_dev_name[BDEVNAME_SIZE]; 239 240 unsigned int hr_timeout_ms; 241 242 /* randomized as the region goes up and down so that a node 243 * recognizes a node going up and down in one iteration */ 244 u64 hr_generation; 245 246 struct delayed_work hr_write_timeout_work; 247 unsigned long hr_last_timeout_start; 248 249 /* negotiate timer, used to negotiate extending hb timeout. */ 250 struct delayed_work hr_nego_timeout_work; 251 unsigned long hr_nego_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 252 253 /* Used during o2hb_check_slot to hold a copy of the block 254 * being checked because we temporarily have to zero out the 255 * crc field. */ 256 struct o2hb_disk_heartbeat_block *hr_tmp_block; 257 258 /* Message key for negotiate timeout message. */ 259 unsigned int hr_key; 260 struct list_head hr_handler_list; 261 262 /* last hb status, 0 for success, other value for error. */ 263 int hr_last_hb_status; 264 }; 265 266 struct o2hb_bio_wait_ctxt { 267 atomic_t wc_num_reqs; 268 struct completion wc_io_complete; 269 int wc_error; 270 }; 271 272 #define O2HB_NEGO_TIMEOUT_MS (O2HB_MAX_WRITE_TIMEOUT_MS/2) 273 274 enum { 275 O2HB_NEGO_TIMEOUT_MSG = 1, 276 O2HB_NEGO_APPROVE_MSG = 2, 277 }; 278 279 struct o2hb_nego_msg { 280 u8 node_num; 281 }; 282 283 static void o2hb_write_timeout(struct work_struct *work) 284 { 285 int failed, quorum; 286 struct o2hb_region *reg = 287 container_of(work, struct o2hb_region, 288 hr_write_timeout_work.work); 289 290 mlog(ML_ERROR, "Heartbeat write timeout to device %s after %u " 291 "milliseconds\n", reg->hr_dev_name, 292 jiffies_to_msecs(jiffies - reg->hr_last_timeout_start)); 293 294 if (o2hb_global_heartbeat_active()) { 295 spin_lock(&o2hb_live_lock); 296 if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) 297 set_bit(reg->hr_region_num, o2hb_failed_region_bitmap); 298 failed = bitmap_weight(o2hb_failed_region_bitmap, 299 O2NM_MAX_REGIONS); 300 quorum = bitmap_weight(o2hb_quorum_region_bitmap, 301 O2NM_MAX_REGIONS); 302 spin_unlock(&o2hb_live_lock); 303 304 mlog(ML_HEARTBEAT, "Number of regions %d, failed regions %d\n", 305 quorum, failed); 306 307 /* 308 * Fence if the number of failed regions >= half the number 309 * of quorum regions 310 */ 311 if ((failed << 1) < quorum) 312 return; 313 } 314 315 o2quo_disk_timeout(); 316 } 317 318 static void o2hb_arm_timeout(struct o2hb_region *reg) 319 { 320 /* Arm writeout only after thread reaches steady state */ 321 if (atomic_read(®->hr_steady_iterations) != 0) 322 return; 323 324 mlog(ML_HEARTBEAT, "Queue write timeout for %u ms\n", 325 O2HB_MAX_WRITE_TIMEOUT_MS); 326 327 if (o2hb_global_heartbeat_active()) { 328 spin_lock(&o2hb_live_lock); 329 clear_bit(reg->hr_region_num, o2hb_failed_region_bitmap); 330 spin_unlock(&o2hb_live_lock); 331 } 332 cancel_delayed_work(®->hr_write_timeout_work); 333 schedule_delayed_work(®->hr_write_timeout_work, 334 msecs_to_jiffies(O2HB_MAX_WRITE_TIMEOUT_MS)); 335 336 cancel_delayed_work(®->hr_nego_timeout_work); 337 /* negotiate timeout must be less than write timeout. */ 338 schedule_delayed_work(®->hr_nego_timeout_work, 339 msecs_to_jiffies(O2HB_NEGO_TIMEOUT_MS)); 340 memset(reg->hr_nego_node_bitmap, 0, sizeof(reg->hr_nego_node_bitmap)); 341 } 342 343 static void o2hb_disarm_timeout(struct o2hb_region *reg) 344 { 345 cancel_delayed_work_sync(®->hr_write_timeout_work); 346 cancel_delayed_work_sync(®->hr_nego_timeout_work); 347 } 348 349 static int o2hb_send_nego_msg(int key, int type, u8 target) 350 { 351 struct o2hb_nego_msg msg; 352 int status, ret; 353 354 msg.node_num = o2nm_this_node(); 355 again: 356 ret = o2net_send_message(type, key, &msg, sizeof(msg), 357 target, &status); 358 359 if (ret == -EAGAIN || ret == -ENOMEM) { 360 msleep(100); 361 goto again; 362 } 363 364 return ret; 365 } 366 367 static void o2hb_nego_timeout(struct work_struct *work) 368 { 369 unsigned long live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 370 int master_node, i, ret; 371 struct o2hb_region *reg; 372 373 reg = container_of(work, struct o2hb_region, hr_nego_timeout_work.work); 374 /* don't negotiate timeout if last hb failed since it is very 375 * possible io failed. Should let write timeout fence self. 376 */ 377 if (reg->hr_last_hb_status) 378 return; 379 380 o2hb_fill_node_map(live_node_bitmap, sizeof(live_node_bitmap)); 381 /* lowest node as master node to make negotiate decision. */ 382 master_node = find_first_bit(live_node_bitmap, O2NM_MAX_NODES); 383 384 if (master_node == o2nm_this_node()) { 385 if (!test_bit(master_node, reg->hr_nego_node_bitmap)) { 386 printk(KERN_NOTICE "o2hb: node %d hb write hung for %ds on region %s (%s).\n", 387 o2nm_this_node(), O2HB_NEGO_TIMEOUT_MS/1000, 388 config_item_name(®->hr_item), reg->hr_dev_name); 389 set_bit(master_node, reg->hr_nego_node_bitmap); 390 } 391 if (memcmp(reg->hr_nego_node_bitmap, live_node_bitmap, 392 sizeof(reg->hr_nego_node_bitmap))) { 393 /* check negotiate bitmap every second to do timeout 394 * approve decision. 395 */ 396 schedule_delayed_work(®->hr_nego_timeout_work, 397 msecs_to_jiffies(1000)); 398 399 return; 400 } 401 402 printk(KERN_NOTICE "o2hb: all nodes hb write hung, maybe region %s (%s) is down.\n", 403 config_item_name(®->hr_item), reg->hr_dev_name); 404 /* approve negotiate timeout request. */ 405 o2hb_arm_timeout(reg); 406 407 i = -1; 408 while ((i = find_next_bit(live_node_bitmap, 409 O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { 410 if (i == master_node) 411 continue; 412 413 mlog(ML_HEARTBEAT, "send NEGO_APPROVE msg to node %d\n", i); 414 ret = o2hb_send_nego_msg(reg->hr_key, 415 O2HB_NEGO_APPROVE_MSG, i); 416 if (ret) 417 mlog(ML_ERROR, "send NEGO_APPROVE msg to node %d fail %d\n", 418 i, ret); 419 } 420 } else { 421 /* negotiate timeout with master node. */ 422 printk(KERN_NOTICE "o2hb: node %d hb write hung for %ds on region %s (%s), negotiate timeout with node %d.\n", 423 o2nm_this_node(), O2HB_NEGO_TIMEOUT_MS/1000, config_item_name(®->hr_item), 424 reg->hr_dev_name, master_node); 425 ret = o2hb_send_nego_msg(reg->hr_key, O2HB_NEGO_TIMEOUT_MSG, 426 master_node); 427 if (ret) 428 mlog(ML_ERROR, "send NEGO_TIMEOUT msg to node %d fail %d\n", 429 master_node, ret); 430 } 431 } 432 433 static int o2hb_nego_timeout_handler(struct o2net_msg *msg, u32 len, void *data, 434 void **ret_data) 435 { 436 struct o2hb_region *reg = data; 437 struct o2hb_nego_msg *nego_msg; 438 439 nego_msg = (struct o2hb_nego_msg *)msg->buf; 440 printk(KERN_NOTICE "o2hb: receive negotiate timeout message from node %d on region %s (%s).\n", 441 nego_msg->node_num, config_item_name(®->hr_item), reg->hr_dev_name); 442 if (nego_msg->node_num < O2NM_MAX_NODES) 443 set_bit(nego_msg->node_num, reg->hr_nego_node_bitmap); 444 else 445 mlog(ML_ERROR, "got nego timeout message from bad node.\n"); 446 447 return 0; 448 } 449 450 static int o2hb_nego_approve_handler(struct o2net_msg *msg, u32 len, void *data, 451 void **ret_data) 452 { 453 struct o2hb_region *reg = data; 454 455 printk(KERN_NOTICE "o2hb: negotiate timeout approved by master node on region %s (%s).\n", 456 config_item_name(®->hr_item), reg->hr_dev_name); 457 o2hb_arm_timeout(reg); 458 return 0; 459 } 460 461 static inline void o2hb_bio_wait_init(struct o2hb_bio_wait_ctxt *wc) 462 { 463 atomic_set(&wc->wc_num_reqs, 1); 464 init_completion(&wc->wc_io_complete); 465 wc->wc_error = 0; 466 } 467 468 /* Used in error paths too */ 469 static inline void o2hb_bio_wait_dec(struct o2hb_bio_wait_ctxt *wc, 470 unsigned int num) 471 { 472 /* sadly atomic_sub_and_test() isn't available on all platforms. The 473 * good news is that the fast path only completes one at a time */ 474 while(num--) { 475 if (atomic_dec_and_test(&wc->wc_num_reqs)) { 476 BUG_ON(num > 0); 477 complete(&wc->wc_io_complete); 478 } 479 } 480 } 481 482 static void o2hb_wait_on_io(struct o2hb_bio_wait_ctxt *wc) 483 { 484 o2hb_bio_wait_dec(wc, 1); 485 wait_for_completion(&wc->wc_io_complete); 486 } 487 488 static void o2hb_bio_end_io(struct bio *bio) 489 { 490 struct o2hb_bio_wait_ctxt *wc = bio->bi_private; 491 492 if (bio->bi_status) { 493 mlog(ML_ERROR, "IO Error %d\n", bio->bi_status); 494 wc->wc_error = blk_status_to_errno(bio->bi_status); 495 } 496 497 o2hb_bio_wait_dec(wc, 1); 498 bio_put(bio); 499 } 500 501 /* Setup a Bio to cover I/O against num_slots slots starting at 502 * start_slot. */ 503 static struct bio *o2hb_setup_one_bio(struct o2hb_region *reg, 504 struct o2hb_bio_wait_ctxt *wc, 505 unsigned int *current_slot, 506 unsigned int max_slots, int op, 507 int op_flags) 508 { 509 int len, current_page; 510 unsigned int vec_len, vec_start; 511 unsigned int bits = reg->hr_block_bits; 512 unsigned int spp = reg->hr_slots_per_page; 513 unsigned int cs = *current_slot; 514 struct bio *bio; 515 struct page *page; 516 517 /* Testing has shown this allocation to take long enough under 518 * GFP_KERNEL that the local node can get fenced. It would be 519 * nicest if we could pre-allocate these bios and avoid this 520 * all together. */ 521 bio = bio_alloc(reg->hr_bdev, 16, op | op_flags, GFP_ATOMIC); 522 if (!bio) { 523 mlog(ML_ERROR, "Could not alloc slots BIO!\n"); 524 bio = ERR_PTR(-ENOMEM); 525 goto bail; 526 } 527 528 /* Must put everything in 512 byte sectors for the bio... */ 529 bio->bi_iter.bi_sector = (reg->hr_start_block + cs) << (bits - 9); 530 bio->bi_private = wc; 531 bio->bi_end_io = o2hb_bio_end_io; 532 533 vec_start = (cs << bits) % PAGE_SIZE; 534 while(cs < max_slots) { 535 current_page = cs / spp; 536 page = reg->hr_slot_data[current_page]; 537 538 vec_len = min(PAGE_SIZE - vec_start, 539 (max_slots-cs) * (PAGE_SIZE/spp) ); 540 541 mlog(ML_HB_BIO, "page %d, vec_len = %u, vec_start = %u\n", 542 current_page, vec_len, vec_start); 543 544 len = bio_add_page(bio, page, vec_len, vec_start); 545 if (len != vec_len) break; 546 547 cs += vec_len / (PAGE_SIZE/spp); 548 vec_start = 0; 549 } 550 551 bail: 552 *current_slot = cs; 553 return bio; 554 } 555 556 static int o2hb_read_slots(struct o2hb_region *reg, 557 unsigned int begin_slot, 558 unsigned int max_slots) 559 { 560 unsigned int current_slot = begin_slot; 561 int status; 562 struct o2hb_bio_wait_ctxt wc; 563 struct bio *bio; 564 565 o2hb_bio_wait_init(&wc); 566 567 while(current_slot < max_slots) { 568 bio = o2hb_setup_one_bio(reg, &wc, ¤t_slot, max_slots, 569 REQ_OP_READ, 0); 570 if (IS_ERR(bio)) { 571 status = PTR_ERR(bio); 572 mlog_errno(status); 573 goto bail_and_wait; 574 } 575 576 atomic_inc(&wc.wc_num_reqs); 577 submit_bio(bio); 578 } 579 580 status = 0; 581 582 bail_and_wait: 583 o2hb_wait_on_io(&wc); 584 if (wc.wc_error && !status) 585 status = wc.wc_error; 586 587 return status; 588 } 589 590 static int o2hb_issue_node_write(struct o2hb_region *reg, 591 struct o2hb_bio_wait_ctxt *write_wc) 592 { 593 int status; 594 unsigned int slot; 595 struct bio *bio; 596 597 o2hb_bio_wait_init(write_wc); 598 599 slot = o2nm_this_node(); 600 601 bio = o2hb_setup_one_bio(reg, write_wc, &slot, slot+1, REQ_OP_WRITE, 602 REQ_SYNC); 603 if (IS_ERR(bio)) { 604 status = PTR_ERR(bio); 605 mlog_errno(status); 606 goto bail; 607 } 608 609 atomic_inc(&write_wc->wc_num_reqs); 610 submit_bio(bio); 611 612 status = 0; 613 bail: 614 return status; 615 } 616 617 static u32 o2hb_compute_block_crc_le(struct o2hb_region *reg, 618 struct o2hb_disk_heartbeat_block *hb_block) 619 { 620 __le32 old_cksum; 621 u32 ret; 622 623 /* We want to compute the block crc with a 0 value in the 624 * hb_cksum field. Save it off here and replace after the 625 * crc. */ 626 old_cksum = hb_block->hb_cksum; 627 hb_block->hb_cksum = 0; 628 629 ret = crc32_le(0, (unsigned char *) hb_block, reg->hr_block_bytes); 630 631 hb_block->hb_cksum = old_cksum; 632 633 return ret; 634 } 635 636 static void o2hb_dump_slot(struct o2hb_disk_heartbeat_block *hb_block) 637 { 638 mlog(ML_ERROR, "Dump slot information: seq = 0x%llx, node = %u, " 639 "cksum = 0x%x, generation 0x%llx\n", 640 (long long)le64_to_cpu(hb_block->hb_seq), 641 hb_block->hb_node, le32_to_cpu(hb_block->hb_cksum), 642 (long long)le64_to_cpu(hb_block->hb_generation)); 643 } 644 645 static int o2hb_verify_crc(struct o2hb_region *reg, 646 struct o2hb_disk_heartbeat_block *hb_block) 647 { 648 u32 read, computed; 649 650 read = le32_to_cpu(hb_block->hb_cksum); 651 computed = o2hb_compute_block_crc_le(reg, hb_block); 652 653 return read == computed; 654 } 655 656 /* 657 * Compare the slot data with what we wrote in the last iteration. 658 * If the match fails, print an appropriate error message. This is to 659 * detect errors like... another node hearting on the same slot, 660 * flaky device that is losing writes, etc. 661 * Returns 1 if check succeeds, 0 otherwise. 662 */ 663 static int o2hb_check_own_slot(struct o2hb_region *reg) 664 { 665 struct o2hb_disk_slot *slot; 666 struct o2hb_disk_heartbeat_block *hb_block; 667 char *errstr; 668 669 slot = ®->hr_slots[o2nm_this_node()]; 670 /* Don't check on our 1st timestamp */ 671 if (!slot->ds_last_time) 672 return 0; 673 674 hb_block = slot->ds_raw_block; 675 if (le64_to_cpu(hb_block->hb_seq) == slot->ds_last_time && 676 le64_to_cpu(hb_block->hb_generation) == slot->ds_last_generation && 677 hb_block->hb_node == slot->ds_node_num) 678 return 1; 679 680 #define ERRSTR1 "Another node is heartbeating on device" 681 #define ERRSTR2 "Heartbeat generation mismatch on device" 682 #define ERRSTR3 "Heartbeat sequence mismatch on device" 683 684 if (hb_block->hb_node != slot->ds_node_num) 685 errstr = ERRSTR1; 686 else if (le64_to_cpu(hb_block->hb_generation) != 687 slot->ds_last_generation) 688 errstr = ERRSTR2; 689 else 690 errstr = ERRSTR3; 691 692 mlog(ML_ERROR, "%s (%s): expected(%u:0x%llx, 0x%llx), " 693 "ondisk(%u:0x%llx, 0x%llx)\n", errstr, reg->hr_dev_name, 694 slot->ds_node_num, (unsigned long long)slot->ds_last_generation, 695 (unsigned long long)slot->ds_last_time, hb_block->hb_node, 696 (unsigned long long)le64_to_cpu(hb_block->hb_generation), 697 (unsigned long long)le64_to_cpu(hb_block->hb_seq)); 698 699 return 0; 700 } 701 702 static inline void o2hb_prepare_block(struct o2hb_region *reg, 703 u64 generation) 704 { 705 int node_num; 706 u64 cputime; 707 struct o2hb_disk_slot *slot; 708 struct o2hb_disk_heartbeat_block *hb_block; 709 710 node_num = o2nm_this_node(); 711 slot = ®->hr_slots[node_num]; 712 713 hb_block = (struct o2hb_disk_heartbeat_block *)slot->ds_raw_block; 714 memset(hb_block, 0, reg->hr_block_bytes); 715 /* TODO: time stuff */ 716 cputime = ktime_get_real_seconds(); 717 if (!cputime) 718 cputime = 1; 719 720 hb_block->hb_seq = cpu_to_le64(cputime); 721 hb_block->hb_node = node_num; 722 hb_block->hb_generation = cpu_to_le64(generation); 723 hb_block->hb_dead_ms = cpu_to_le32(o2hb_dead_threshold * O2HB_REGION_TIMEOUT_MS); 724 725 /* This step must always happen last! */ 726 hb_block->hb_cksum = cpu_to_le32(o2hb_compute_block_crc_le(reg, 727 hb_block)); 728 729 mlog(ML_HB_BIO, "our node generation = 0x%llx, cksum = 0x%x\n", 730 (long long)generation, 731 le32_to_cpu(hb_block->hb_cksum)); 732 } 733 734 static void o2hb_fire_callbacks(struct o2hb_callback *hbcall, 735 struct o2nm_node *node, 736 int idx) 737 { 738 struct o2hb_callback_func *f; 739 740 list_for_each_entry(f, &hbcall->list, hc_item) { 741 mlog(ML_HEARTBEAT, "calling funcs %p\n", f); 742 (f->hc_func)(node, idx, f->hc_data); 743 } 744 } 745 746 /* Will run the list in order until we process the passed event */ 747 static void o2hb_run_event_list(struct o2hb_node_event *queued_event) 748 { 749 struct o2hb_callback *hbcall; 750 struct o2hb_node_event *event; 751 752 /* Holding callback sem assures we don't alter the callback 753 * lists when doing this, and serializes ourselves with other 754 * processes wanting callbacks. */ 755 down_write(&o2hb_callback_sem); 756 757 spin_lock(&o2hb_live_lock); 758 while (!list_empty(&o2hb_node_events) 759 && !list_empty(&queued_event->hn_item)) { 760 event = list_entry(o2hb_node_events.next, 761 struct o2hb_node_event, 762 hn_item); 763 list_del_init(&event->hn_item); 764 spin_unlock(&o2hb_live_lock); 765 766 mlog(ML_HEARTBEAT, "Node %s event for %d\n", 767 event->hn_event_type == O2HB_NODE_UP_CB ? "UP" : "DOWN", 768 event->hn_node_num); 769 770 hbcall = hbcall_from_type(event->hn_event_type); 771 772 /* We should *never* have gotten on to the list with a 773 * bad type... This isn't something that we should try 774 * to recover from. */ 775 BUG_ON(IS_ERR(hbcall)); 776 777 o2hb_fire_callbacks(hbcall, event->hn_node, event->hn_node_num); 778 779 spin_lock(&o2hb_live_lock); 780 } 781 spin_unlock(&o2hb_live_lock); 782 783 up_write(&o2hb_callback_sem); 784 } 785 786 static void o2hb_queue_node_event(struct o2hb_node_event *event, 787 enum o2hb_callback_type type, 788 struct o2nm_node *node, 789 int node_num) 790 { 791 assert_spin_locked(&o2hb_live_lock); 792 793 BUG_ON((!node) && (type != O2HB_NODE_DOWN_CB)); 794 795 event->hn_event_type = type; 796 event->hn_node = node; 797 event->hn_node_num = node_num; 798 799 mlog(ML_HEARTBEAT, "Queue node %s event for node %d\n", 800 type == O2HB_NODE_UP_CB ? "UP" : "DOWN", node_num); 801 802 list_add_tail(&event->hn_item, &o2hb_node_events); 803 } 804 805 static void o2hb_shutdown_slot(struct o2hb_disk_slot *slot) 806 { 807 struct o2hb_node_event event = 808 { .hn_item = LIST_HEAD_INIT(event.hn_item), }; 809 struct o2nm_node *node; 810 int queued = 0; 811 812 node = o2nm_get_node_by_num(slot->ds_node_num); 813 if (!node) 814 return; 815 816 spin_lock(&o2hb_live_lock); 817 if (!list_empty(&slot->ds_live_item)) { 818 mlog(ML_HEARTBEAT, "Shutdown, node %d leaves region\n", 819 slot->ds_node_num); 820 821 list_del_init(&slot->ds_live_item); 822 823 if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { 824 clear_bit(slot->ds_node_num, o2hb_live_node_bitmap); 825 826 o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node, 827 slot->ds_node_num); 828 queued = 1; 829 } 830 } 831 spin_unlock(&o2hb_live_lock); 832 833 if (queued) 834 o2hb_run_event_list(&event); 835 836 o2nm_node_put(node); 837 } 838 839 static void o2hb_set_quorum_device(struct o2hb_region *reg) 840 { 841 if (!o2hb_global_heartbeat_active()) 842 return; 843 844 /* Prevent race with o2hb_heartbeat_group_drop_item() */ 845 if (kthread_should_stop()) 846 return; 847 848 /* Tag region as quorum only after thread reaches steady state */ 849 if (atomic_read(®->hr_steady_iterations) != 0) 850 return; 851 852 spin_lock(&o2hb_live_lock); 853 854 if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) 855 goto unlock; 856 857 /* 858 * A region can be added to the quorum only when it sees all 859 * live nodes heartbeat on it. In other words, the region has been 860 * added to all nodes. 861 */ 862 if (memcmp(reg->hr_live_node_bitmap, o2hb_live_node_bitmap, 863 sizeof(o2hb_live_node_bitmap))) 864 goto unlock; 865 866 printk(KERN_NOTICE "o2hb: Region %s (%s) is now a quorum device\n", 867 config_item_name(®->hr_item), reg->hr_dev_name); 868 869 set_bit(reg->hr_region_num, o2hb_quorum_region_bitmap); 870 871 /* 872 * If global heartbeat active, unpin all regions if the 873 * region count > CUT_OFF 874 */ 875 if (bitmap_weight(o2hb_quorum_region_bitmap, 876 O2NM_MAX_REGIONS) > O2HB_PIN_CUT_OFF) 877 o2hb_region_unpin(NULL); 878 unlock: 879 spin_unlock(&o2hb_live_lock); 880 } 881 882 static int o2hb_check_slot(struct o2hb_region *reg, 883 struct o2hb_disk_slot *slot) 884 { 885 int changed = 0, gen_changed = 0; 886 struct o2hb_node_event event = 887 { .hn_item = LIST_HEAD_INIT(event.hn_item), }; 888 struct o2nm_node *node; 889 struct o2hb_disk_heartbeat_block *hb_block = reg->hr_tmp_block; 890 u64 cputime; 891 unsigned int dead_ms = o2hb_dead_threshold * O2HB_REGION_TIMEOUT_MS; 892 unsigned int slot_dead_ms; 893 int tmp; 894 int queued = 0; 895 896 memcpy(hb_block, slot->ds_raw_block, reg->hr_block_bytes); 897 898 /* 899 * If a node is no longer configured but is still in the livemap, we 900 * may need to clear that bit from the livemap. 901 */ 902 node = o2nm_get_node_by_num(slot->ds_node_num); 903 if (!node) { 904 spin_lock(&o2hb_live_lock); 905 tmp = test_bit(slot->ds_node_num, o2hb_live_node_bitmap); 906 spin_unlock(&o2hb_live_lock); 907 if (!tmp) 908 return 0; 909 } 910 911 if (!o2hb_verify_crc(reg, hb_block)) { 912 /* all paths from here will drop o2hb_live_lock for 913 * us. */ 914 spin_lock(&o2hb_live_lock); 915 916 /* Don't print an error on the console in this case - 917 * a freshly formatted heartbeat area will not have a 918 * crc set on it. */ 919 if (list_empty(&slot->ds_live_item)) 920 goto out; 921 922 /* The node is live but pushed out a bad crc. We 923 * consider it a transient miss but don't populate any 924 * other values as they may be junk. */ 925 mlog(ML_ERROR, "Node %d has written a bad crc to %s\n", 926 slot->ds_node_num, reg->hr_dev_name); 927 o2hb_dump_slot(hb_block); 928 929 slot->ds_equal_samples++; 930 goto fire_callbacks; 931 } 932 933 /* we don't care if these wrap.. the state transitions below 934 * clear at the right places */ 935 cputime = le64_to_cpu(hb_block->hb_seq); 936 if (slot->ds_last_time != cputime) 937 slot->ds_changed_samples++; 938 else 939 slot->ds_equal_samples++; 940 slot->ds_last_time = cputime; 941 942 /* The node changed heartbeat generations. We assume this to 943 * mean it dropped off but came back before we timed out. We 944 * want to consider it down for the time being but don't want 945 * to lose any changed_samples state we might build up to 946 * considering it live again. */ 947 if (slot->ds_last_generation != le64_to_cpu(hb_block->hb_generation)) { 948 gen_changed = 1; 949 slot->ds_equal_samples = 0; 950 mlog(ML_HEARTBEAT, "Node %d changed generation (0x%llx " 951 "to 0x%llx)\n", slot->ds_node_num, 952 (long long)slot->ds_last_generation, 953 (long long)le64_to_cpu(hb_block->hb_generation)); 954 } 955 956 slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation); 957 958 mlog(ML_HEARTBEAT, "Slot %d gen 0x%llx cksum 0x%x " 959 "seq %llu last %llu changed %u equal %u\n", 960 slot->ds_node_num, (long long)slot->ds_last_generation, 961 le32_to_cpu(hb_block->hb_cksum), 962 (unsigned long long)le64_to_cpu(hb_block->hb_seq), 963 (unsigned long long)slot->ds_last_time, slot->ds_changed_samples, 964 slot->ds_equal_samples); 965 966 spin_lock(&o2hb_live_lock); 967 968 fire_callbacks: 969 /* dead nodes only come to life after some number of 970 * changes at any time during their dead time */ 971 if (list_empty(&slot->ds_live_item) && 972 slot->ds_changed_samples >= O2HB_LIVE_THRESHOLD) { 973 mlog(ML_HEARTBEAT, "Node %d (id 0x%llx) joined my region\n", 974 slot->ds_node_num, (long long)slot->ds_last_generation); 975 976 set_bit(slot->ds_node_num, reg->hr_live_node_bitmap); 977 978 /* first on the list generates a callback */ 979 if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { 980 mlog(ML_HEARTBEAT, "o2hb: Add node %d to live nodes " 981 "bitmap\n", slot->ds_node_num); 982 set_bit(slot->ds_node_num, o2hb_live_node_bitmap); 983 984 o2hb_queue_node_event(&event, O2HB_NODE_UP_CB, node, 985 slot->ds_node_num); 986 987 changed = 1; 988 queued = 1; 989 } 990 991 list_add_tail(&slot->ds_live_item, 992 &o2hb_live_slots[slot->ds_node_num]); 993 994 slot->ds_equal_samples = 0; 995 996 /* We want to be sure that all nodes agree on the 997 * number of milliseconds before a node will be 998 * considered dead. The self-fencing timeout is 999 * computed from this value, and a discrepancy might 1000 * result in heartbeat calling a node dead when it 1001 * hasn't self-fenced yet. */ 1002 slot_dead_ms = le32_to_cpu(hb_block->hb_dead_ms); 1003 if (slot_dead_ms && slot_dead_ms != dead_ms) { 1004 /* TODO: Perhaps we can fail the region here. */ 1005 mlog(ML_ERROR, "Node %d on device %s has a dead count " 1006 "of %u ms, but our count is %u ms.\n" 1007 "Please double check your configuration values " 1008 "for 'O2CB_HEARTBEAT_THRESHOLD'\n", 1009 slot->ds_node_num, reg->hr_dev_name, slot_dead_ms, 1010 dead_ms); 1011 } 1012 goto out; 1013 } 1014 1015 /* if the list is dead, we're done.. */ 1016 if (list_empty(&slot->ds_live_item)) 1017 goto out; 1018 1019 /* live nodes only go dead after enough consequtive missed 1020 * samples.. reset the missed counter whenever we see 1021 * activity */ 1022 if (slot->ds_equal_samples >= o2hb_dead_threshold || gen_changed) { 1023 mlog(ML_HEARTBEAT, "Node %d left my region\n", 1024 slot->ds_node_num); 1025 1026 clear_bit(slot->ds_node_num, reg->hr_live_node_bitmap); 1027 1028 /* last off the live_slot generates a callback */ 1029 list_del_init(&slot->ds_live_item); 1030 if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { 1031 mlog(ML_HEARTBEAT, "o2hb: Remove node %d from live " 1032 "nodes bitmap\n", slot->ds_node_num); 1033 clear_bit(slot->ds_node_num, o2hb_live_node_bitmap); 1034 1035 /* node can be null */ 1036 o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, 1037 node, slot->ds_node_num); 1038 1039 changed = 1; 1040 queued = 1; 1041 } 1042 1043 /* We don't clear this because the node is still 1044 * actually writing new blocks. */ 1045 if (!gen_changed) 1046 slot->ds_changed_samples = 0; 1047 goto out; 1048 } 1049 if (slot->ds_changed_samples) { 1050 slot->ds_changed_samples = 0; 1051 slot->ds_equal_samples = 0; 1052 } 1053 out: 1054 spin_unlock(&o2hb_live_lock); 1055 1056 if (queued) 1057 o2hb_run_event_list(&event); 1058 1059 if (node) 1060 o2nm_node_put(node); 1061 return changed; 1062 } 1063 1064 static int o2hb_highest_node(unsigned long *nodes, int numbits) 1065 { 1066 return find_last_bit(nodes, numbits); 1067 } 1068 1069 static int o2hb_lowest_node(unsigned long *nodes, int numbits) 1070 { 1071 return find_first_bit(nodes, numbits); 1072 } 1073 1074 static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) 1075 { 1076 int i, ret, highest_node, lowest_node; 1077 int membership_change = 0, own_slot_ok = 0; 1078 unsigned long configured_nodes[BITS_TO_LONGS(O2NM_MAX_NODES)]; 1079 unsigned long live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 1080 struct o2hb_bio_wait_ctxt write_wc; 1081 1082 ret = o2nm_configured_node_map(configured_nodes, 1083 sizeof(configured_nodes)); 1084 if (ret) { 1085 mlog_errno(ret); 1086 goto bail; 1087 } 1088 1089 /* 1090 * If a node is not configured but is in the livemap, we still need 1091 * to read the slot so as to be able to remove it from the livemap. 1092 */ 1093 o2hb_fill_node_map(live_node_bitmap, sizeof(live_node_bitmap)); 1094 i = -1; 1095 while ((i = find_next_bit(live_node_bitmap, 1096 O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { 1097 set_bit(i, configured_nodes); 1098 } 1099 1100 highest_node = o2hb_highest_node(configured_nodes, O2NM_MAX_NODES); 1101 lowest_node = o2hb_lowest_node(configured_nodes, O2NM_MAX_NODES); 1102 if (highest_node >= O2NM_MAX_NODES || lowest_node >= O2NM_MAX_NODES) { 1103 mlog(ML_NOTICE, "o2hb: No configured nodes found!\n"); 1104 ret = -EINVAL; 1105 goto bail; 1106 } 1107 1108 /* No sense in reading the slots of nodes that don't exist 1109 * yet. Of course, if the node definitions have holes in them 1110 * then we're reading an empty slot anyway... Consider this 1111 * best-effort. */ 1112 ret = o2hb_read_slots(reg, lowest_node, highest_node + 1); 1113 if (ret < 0) { 1114 mlog_errno(ret); 1115 goto bail; 1116 } 1117 1118 /* With an up to date view of the slots, we can check that no 1119 * other node has been improperly configured to heartbeat in 1120 * our slot. */ 1121 own_slot_ok = o2hb_check_own_slot(reg); 1122 1123 /* fill in the proper info for our next heartbeat */ 1124 o2hb_prepare_block(reg, reg->hr_generation); 1125 1126 ret = o2hb_issue_node_write(reg, &write_wc); 1127 if (ret < 0) { 1128 mlog_errno(ret); 1129 goto bail; 1130 } 1131 1132 i = -1; 1133 while((i = find_next_bit(configured_nodes, 1134 O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { 1135 membership_change |= o2hb_check_slot(reg, ®->hr_slots[i]); 1136 } 1137 1138 /* 1139 * We have to be sure we've advertised ourselves on disk 1140 * before we can go to steady state. This ensures that 1141 * people we find in our steady state have seen us. 1142 */ 1143 o2hb_wait_on_io(&write_wc); 1144 if (write_wc.wc_error) { 1145 /* Do not re-arm the write timeout on I/O error - we 1146 * can't be sure that the new block ever made it to 1147 * disk */ 1148 mlog(ML_ERROR, "Write error %d on device \"%s\"\n", 1149 write_wc.wc_error, reg->hr_dev_name); 1150 ret = write_wc.wc_error; 1151 goto bail; 1152 } 1153 1154 /* Skip disarming the timeout if own slot has stale/bad data */ 1155 if (own_slot_ok) { 1156 o2hb_set_quorum_device(reg); 1157 o2hb_arm_timeout(reg); 1158 reg->hr_last_timeout_start = jiffies; 1159 } 1160 1161 bail: 1162 /* let the person who launched us know when things are steady */ 1163 if (atomic_read(®->hr_steady_iterations) != 0) { 1164 if (!ret && own_slot_ok && !membership_change) { 1165 if (atomic_dec_and_test(®->hr_steady_iterations)) 1166 wake_up(&o2hb_steady_queue); 1167 } 1168 } 1169 1170 if (atomic_read(®->hr_steady_iterations) != 0) { 1171 if (atomic_dec_and_test(®->hr_unsteady_iterations)) { 1172 printk(KERN_NOTICE "o2hb: Unable to stabilize " 1173 "heartbeat on region %s (%s)\n", 1174 config_item_name(®->hr_item), 1175 reg->hr_dev_name); 1176 atomic_set(®->hr_steady_iterations, 0); 1177 reg->hr_aborted_start = 1; 1178 wake_up(&o2hb_steady_queue); 1179 ret = -EIO; 1180 } 1181 } 1182 1183 return ret; 1184 } 1185 1186 /* 1187 * we ride the region ref that the region dir holds. before the region 1188 * dir is removed and drops it ref it will wait to tear down this 1189 * thread. 1190 */ 1191 static int o2hb_thread(void *data) 1192 { 1193 int i, ret; 1194 struct o2hb_region *reg = data; 1195 struct o2hb_bio_wait_ctxt write_wc; 1196 ktime_t before_hb, after_hb; 1197 unsigned int elapsed_msec; 1198 1199 mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread running\n"); 1200 1201 set_user_nice(current, MIN_NICE); 1202 1203 /* Pin node */ 1204 ret = o2nm_depend_this_node(); 1205 if (ret) { 1206 mlog(ML_ERROR, "Node has been deleted, ret = %d\n", ret); 1207 reg->hr_node_deleted = 1; 1208 wake_up(&o2hb_steady_queue); 1209 return 0; 1210 } 1211 1212 while (!kthread_should_stop() && 1213 !reg->hr_unclean_stop && !reg->hr_aborted_start) { 1214 /* We track the time spent inside 1215 * o2hb_do_disk_heartbeat so that we avoid more than 1216 * hr_timeout_ms between disk writes. On busy systems 1217 * this should result in a heartbeat which is less 1218 * likely to time itself out. */ 1219 before_hb = ktime_get_real(); 1220 1221 ret = o2hb_do_disk_heartbeat(reg); 1222 reg->hr_last_hb_status = ret; 1223 1224 after_hb = ktime_get_real(); 1225 1226 elapsed_msec = (unsigned int) 1227 ktime_ms_delta(after_hb, before_hb); 1228 1229 mlog(ML_HEARTBEAT, 1230 "start = %lld, end = %lld, msec = %u, ret = %d\n", 1231 before_hb, after_hb, elapsed_msec, ret); 1232 1233 if (!kthread_should_stop() && 1234 elapsed_msec < reg->hr_timeout_ms) { 1235 /* the kthread api has blocked signals for us so no 1236 * need to record the return value. */ 1237 msleep_interruptible(reg->hr_timeout_ms - elapsed_msec); 1238 } 1239 } 1240 1241 o2hb_disarm_timeout(reg); 1242 1243 /* unclean stop is only used in very bad situation */ 1244 for(i = 0; !reg->hr_unclean_stop && i < reg->hr_blocks; i++) 1245 o2hb_shutdown_slot(®->hr_slots[i]); 1246 1247 /* Explicit down notification - avoid forcing the other nodes 1248 * to timeout on this region when we could just as easily 1249 * write a clear generation - thus indicating to them that 1250 * this node has left this region. 1251 */ 1252 if (!reg->hr_unclean_stop && !reg->hr_aborted_start) { 1253 o2hb_prepare_block(reg, 0); 1254 ret = o2hb_issue_node_write(reg, &write_wc); 1255 if (ret == 0) 1256 o2hb_wait_on_io(&write_wc); 1257 else 1258 mlog_errno(ret); 1259 } 1260 1261 /* Unpin node */ 1262 o2nm_undepend_this_node(); 1263 1264 mlog(ML_HEARTBEAT|ML_KTHREAD, "o2hb thread exiting\n"); 1265 1266 return 0; 1267 } 1268 1269 #ifdef CONFIG_DEBUG_FS 1270 static int o2hb_debug_open(struct inode *inode, struct file *file) 1271 { 1272 struct o2hb_debug_buf *db = inode->i_private; 1273 struct o2hb_region *reg; 1274 unsigned long map[BITS_TO_LONGS(O2NM_MAX_NODES)]; 1275 unsigned long lts; 1276 char *buf = NULL; 1277 int i = -1; 1278 int out = 0; 1279 1280 /* max_nodes should be the largest bitmap we pass here */ 1281 BUG_ON(sizeof(map) < db->db_size); 1282 1283 buf = kmalloc(PAGE_SIZE, GFP_KERNEL); 1284 if (!buf) 1285 goto bail; 1286 1287 switch (db->db_type) { 1288 case O2HB_DB_TYPE_LIVENODES: 1289 case O2HB_DB_TYPE_LIVEREGIONS: 1290 case O2HB_DB_TYPE_QUORUMREGIONS: 1291 case O2HB_DB_TYPE_FAILEDREGIONS: 1292 spin_lock(&o2hb_live_lock); 1293 memcpy(map, db->db_data, db->db_size); 1294 spin_unlock(&o2hb_live_lock); 1295 break; 1296 1297 case O2HB_DB_TYPE_REGION_LIVENODES: 1298 spin_lock(&o2hb_live_lock); 1299 reg = (struct o2hb_region *)db->db_data; 1300 memcpy(map, reg->hr_live_node_bitmap, db->db_size); 1301 spin_unlock(&o2hb_live_lock); 1302 break; 1303 1304 case O2HB_DB_TYPE_REGION_NUMBER: 1305 reg = (struct o2hb_region *)db->db_data; 1306 out += scnprintf(buf + out, PAGE_SIZE - out, "%d\n", 1307 reg->hr_region_num); 1308 goto done; 1309 1310 case O2HB_DB_TYPE_REGION_ELAPSED_TIME: 1311 reg = (struct o2hb_region *)db->db_data; 1312 lts = reg->hr_last_timeout_start; 1313 /* If 0, it has never been set before */ 1314 if (lts) 1315 lts = jiffies_to_msecs(jiffies - lts); 1316 out += scnprintf(buf + out, PAGE_SIZE - out, "%lu\n", lts); 1317 goto done; 1318 1319 case O2HB_DB_TYPE_REGION_PINNED: 1320 reg = (struct o2hb_region *)db->db_data; 1321 out += scnprintf(buf + out, PAGE_SIZE - out, "%u\n", 1322 !!reg->hr_item_pinned); 1323 goto done; 1324 1325 default: 1326 goto done; 1327 } 1328 1329 while ((i = find_next_bit(map, db->db_len, i + 1)) < db->db_len) 1330 out += scnprintf(buf + out, PAGE_SIZE - out, "%d ", i); 1331 out += scnprintf(buf + out, PAGE_SIZE - out, "\n"); 1332 1333 done: 1334 i_size_write(inode, out); 1335 1336 file->private_data = buf; 1337 1338 return 0; 1339 bail: 1340 return -ENOMEM; 1341 } 1342 1343 static int o2hb_debug_release(struct inode *inode, struct file *file) 1344 { 1345 kfree(file->private_data); 1346 return 0; 1347 } 1348 1349 static ssize_t o2hb_debug_read(struct file *file, char __user *buf, 1350 size_t nbytes, loff_t *ppos) 1351 { 1352 return simple_read_from_buffer(buf, nbytes, ppos, file->private_data, 1353 i_size_read(file->f_mapping->host)); 1354 } 1355 #else 1356 static int o2hb_debug_open(struct inode *inode, struct file *file) 1357 { 1358 return 0; 1359 } 1360 static int o2hb_debug_release(struct inode *inode, struct file *file) 1361 { 1362 return 0; 1363 } 1364 static ssize_t o2hb_debug_read(struct file *file, char __user *buf, 1365 size_t nbytes, loff_t *ppos) 1366 { 1367 return 0; 1368 } 1369 #endif /* CONFIG_DEBUG_FS */ 1370 1371 static const struct file_operations o2hb_debug_fops = { 1372 .open = o2hb_debug_open, 1373 .release = o2hb_debug_release, 1374 .read = o2hb_debug_read, 1375 .llseek = generic_file_llseek, 1376 }; 1377 1378 void o2hb_exit(void) 1379 { 1380 debugfs_remove_recursive(o2hb_debug_dir); 1381 kfree(o2hb_db_livenodes); 1382 kfree(o2hb_db_liveregions); 1383 kfree(o2hb_db_quorumregions); 1384 kfree(o2hb_db_failedregions); 1385 } 1386 1387 static void o2hb_debug_create(const char *name, struct dentry *dir, 1388 struct o2hb_debug_buf **db, int db_len, int type, 1389 int size, int len, void *data) 1390 { 1391 *db = kmalloc(db_len, GFP_KERNEL); 1392 if (!*db) 1393 return; 1394 1395 (*db)->db_type = type; 1396 (*db)->db_size = size; 1397 (*db)->db_len = len; 1398 (*db)->db_data = data; 1399 1400 debugfs_create_file(name, S_IFREG|S_IRUSR, dir, *db, &o2hb_debug_fops); 1401 } 1402 1403 static void o2hb_debug_init(void) 1404 { 1405 o2hb_debug_dir = debugfs_create_dir(O2HB_DEBUG_DIR, NULL); 1406 1407 o2hb_debug_create(O2HB_DEBUG_LIVENODES, o2hb_debug_dir, 1408 &o2hb_db_livenodes, sizeof(*o2hb_db_livenodes), 1409 O2HB_DB_TYPE_LIVENODES, sizeof(o2hb_live_node_bitmap), 1410 O2NM_MAX_NODES, o2hb_live_node_bitmap); 1411 1412 o2hb_debug_create(O2HB_DEBUG_LIVEREGIONS, o2hb_debug_dir, 1413 &o2hb_db_liveregions, sizeof(*o2hb_db_liveregions), 1414 O2HB_DB_TYPE_LIVEREGIONS, 1415 sizeof(o2hb_live_region_bitmap), O2NM_MAX_REGIONS, 1416 o2hb_live_region_bitmap); 1417 1418 o2hb_debug_create(O2HB_DEBUG_QUORUMREGIONS, o2hb_debug_dir, 1419 &o2hb_db_quorumregions, 1420 sizeof(*o2hb_db_quorumregions), 1421 O2HB_DB_TYPE_QUORUMREGIONS, 1422 sizeof(o2hb_quorum_region_bitmap), O2NM_MAX_REGIONS, 1423 o2hb_quorum_region_bitmap); 1424 1425 o2hb_debug_create(O2HB_DEBUG_FAILEDREGIONS, o2hb_debug_dir, 1426 &o2hb_db_failedregions, 1427 sizeof(*o2hb_db_failedregions), 1428 O2HB_DB_TYPE_FAILEDREGIONS, 1429 sizeof(o2hb_failed_region_bitmap), O2NM_MAX_REGIONS, 1430 o2hb_failed_region_bitmap); 1431 } 1432 1433 void o2hb_init(void) 1434 { 1435 int i; 1436 1437 for (i = 0; i < ARRAY_SIZE(o2hb_callbacks); i++) 1438 INIT_LIST_HEAD(&o2hb_callbacks[i].list); 1439 1440 for (i = 0; i < ARRAY_SIZE(o2hb_live_slots); i++) 1441 INIT_LIST_HEAD(&o2hb_live_slots[i]); 1442 1443 memset(o2hb_live_node_bitmap, 0, sizeof(o2hb_live_node_bitmap)); 1444 memset(o2hb_region_bitmap, 0, sizeof(o2hb_region_bitmap)); 1445 memset(o2hb_live_region_bitmap, 0, sizeof(o2hb_live_region_bitmap)); 1446 memset(o2hb_quorum_region_bitmap, 0, sizeof(o2hb_quorum_region_bitmap)); 1447 memset(o2hb_failed_region_bitmap, 0, sizeof(o2hb_failed_region_bitmap)); 1448 1449 o2hb_dependent_users = 0; 1450 1451 o2hb_debug_init(); 1452 } 1453 1454 /* if we're already in a callback then we're already serialized by the sem */ 1455 static void o2hb_fill_node_map_from_callback(unsigned long *map, 1456 unsigned bytes) 1457 { 1458 BUG_ON(bytes < (BITS_TO_LONGS(O2NM_MAX_NODES) * sizeof(unsigned long))); 1459 1460 memcpy(map, &o2hb_live_node_bitmap, bytes); 1461 } 1462 1463 /* 1464 * get a map of all nodes that are heartbeating in any regions 1465 */ 1466 void o2hb_fill_node_map(unsigned long *map, unsigned bytes) 1467 { 1468 /* callers want to serialize this map and callbacks so that they 1469 * can trust that they don't miss nodes coming to the party */ 1470 down_read(&o2hb_callback_sem); 1471 spin_lock(&o2hb_live_lock); 1472 o2hb_fill_node_map_from_callback(map, bytes); 1473 spin_unlock(&o2hb_live_lock); 1474 up_read(&o2hb_callback_sem); 1475 } 1476 EXPORT_SYMBOL_GPL(o2hb_fill_node_map); 1477 1478 /* 1479 * heartbeat configfs bits. The heartbeat set is a default set under 1480 * the cluster set in nodemanager.c. 1481 */ 1482 1483 static struct o2hb_region *to_o2hb_region(struct config_item *item) 1484 { 1485 return item ? container_of(item, struct o2hb_region, hr_item) : NULL; 1486 } 1487 1488 /* drop_item only drops its ref after killing the thread, nothing should 1489 * be using the region anymore. this has to clean up any state that 1490 * attributes might have built up. */ 1491 static void o2hb_region_release(struct config_item *item) 1492 { 1493 int i; 1494 struct page *page; 1495 struct o2hb_region *reg = to_o2hb_region(item); 1496 1497 mlog(ML_HEARTBEAT, "hb region release (%s)\n", reg->hr_dev_name); 1498 1499 kfree(reg->hr_tmp_block); 1500 1501 if (reg->hr_slot_data) { 1502 for (i = 0; i < reg->hr_num_pages; i++) { 1503 page = reg->hr_slot_data[i]; 1504 if (page) 1505 __free_page(page); 1506 } 1507 kfree(reg->hr_slot_data); 1508 } 1509 1510 if (reg->hr_bdev) 1511 blkdev_put(reg->hr_bdev, FMODE_READ|FMODE_WRITE); 1512 1513 kfree(reg->hr_slots); 1514 1515 debugfs_remove_recursive(reg->hr_debug_dir); 1516 kfree(reg->hr_db_livenodes); 1517 kfree(reg->hr_db_regnum); 1518 kfree(reg->hr_db_elapsed_time); 1519 kfree(reg->hr_db_pinned); 1520 1521 spin_lock(&o2hb_live_lock); 1522 list_del(®->hr_all_item); 1523 spin_unlock(&o2hb_live_lock); 1524 1525 o2net_unregister_handler_list(®->hr_handler_list); 1526 kfree(reg); 1527 } 1528 1529 static int o2hb_read_block_input(struct o2hb_region *reg, 1530 const char *page, 1531 unsigned long *ret_bytes, 1532 unsigned int *ret_bits) 1533 { 1534 unsigned long bytes; 1535 char *p = (char *)page; 1536 1537 bytes = simple_strtoul(p, &p, 0); 1538 if (!p || (*p && (*p != '\n'))) 1539 return -EINVAL; 1540 1541 /* Heartbeat and fs min / max block sizes are the same. */ 1542 if (bytes > 4096 || bytes < 512) 1543 return -ERANGE; 1544 if (hweight16(bytes) != 1) 1545 return -EINVAL; 1546 1547 if (ret_bytes) 1548 *ret_bytes = bytes; 1549 if (ret_bits) 1550 *ret_bits = ffs(bytes) - 1; 1551 1552 return 0; 1553 } 1554 1555 static ssize_t o2hb_region_block_bytes_show(struct config_item *item, 1556 char *page) 1557 { 1558 return sprintf(page, "%u\n", to_o2hb_region(item)->hr_block_bytes); 1559 } 1560 1561 static ssize_t o2hb_region_block_bytes_store(struct config_item *item, 1562 const char *page, 1563 size_t count) 1564 { 1565 struct o2hb_region *reg = to_o2hb_region(item); 1566 int status; 1567 unsigned long block_bytes; 1568 unsigned int block_bits; 1569 1570 if (reg->hr_bdev) 1571 return -EINVAL; 1572 1573 status = o2hb_read_block_input(reg, page, &block_bytes, 1574 &block_bits); 1575 if (status) 1576 return status; 1577 1578 reg->hr_block_bytes = (unsigned int)block_bytes; 1579 reg->hr_block_bits = block_bits; 1580 1581 return count; 1582 } 1583 1584 static ssize_t o2hb_region_start_block_show(struct config_item *item, 1585 char *page) 1586 { 1587 return sprintf(page, "%llu\n", to_o2hb_region(item)->hr_start_block); 1588 } 1589 1590 static ssize_t o2hb_region_start_block_store(struct config_item *item, 1591 const char *page, 1592 size_t count) 1593 { 1594 struct o2hb_region *reg = to_o2hb_region(item); 1595 unsigned long long tmp; 1596 char *p = (char *)page; 1597 ssize_t ret; 1598 1599 if (reg->hr_bdev) 1600 return -EINVAL; 1601 1602 ret = kstrtoull(p, 0, &tmp); 1603 if (ret) 1604 return -EINVAL; 1605 1606 reg->hr_start_block = tmp; 1607 1608 return count; 1609 } 1610 1611 static ssize_t o2hb_region_blocks_show(struct config_item *item, char *page) 1612 { 1613 return sprintf(page, "%d\n", to_o2hb_region(item)->hr_blocks); 1614 } 1615 1616 static ssize_t o2hb_region_blocks_store(struct config_item *item, 1617 const char *page, 1618 size_t count) 1619 { 1620 struct o2hb_region *reg = to_o2hb_region(item); 1621 unsigned long tmp; 1622 char *p = (char *)page; 1623 1624 if (reg->hr_bdev) 1625 return -EINVAL; 1626 1627 tmp = simple_strtoul(p, &p, 0); 1628 if (!p || (*p && (*p != '\n'))) 1629 return -EINVAL; 1630 1631 if (tmp > O2NM_MAX_NODES || tmp == 0) 1632 return -ERANGE; 1633 1634 reg->hr_blocks = (unsigned int)tmp; 1635 1636 return count; 1637 } 1638 1639 static ssize_t o2hb_region_dev_show(struct config_item *item, char *page) 1640 { 1641 unsigned int ret = 0; 1642 1643 if (to_o2hb_region(item)->hr_bdev) 1644 ret = sprintf(page, "%s\n", to_o2hb_region(item)->hr_dev_name); 1645 1646 return ret; 1647 } 1648 1649 static void o2hb_init_region_params(struct o2hb_region *reg) 1650 { 1651 reg->hr_slots_per_page = PAGE_SIZE >> reg->hr_block_bits; 1652 reg->hr_timeout_ms = O2HB_REGION_TIMEOUT_MS; 1653 1654 mlog(ML_HEARTBEAT, "hr_start_block = %llu, hr_blocks = %u\n", 1655 reg->hr_start_block, reg->hr_blocks); 1656 mlog(ML_HEARTBEAT, "hr_block_bytes = %u, hr_block_bits = %u\n", 1657 reg->hr_block_bytes, reg->hr_block_bits); 1658 mlog(ML_HEARTBEAT, "hr_timeout_ms = %u\n", reg->hr_timeout_ms); 1659 mlog(ML_HEARTBEAT, "dead threshold = %u\n", o2hb_dead_threshold); 1660 } 1661 1662 static int o2hb_map_slot_data(struct o2hb_region *reg) 1663 { 1664 int i, j; 1665 unsigned int last_slot; 1666 unsigned int spp = reg->hr_slots_per_page; 1667 struct page *page; 1668 char *raw; 1669 struct o2hb_disk_slot *slot; 1670 1671 reg->hr_tmp_block = kmalloc(reg->hr_block_bytes, GFP_KERNEL); 1672 if (reg->hr_tmp_block == NULL) 1673 return -ENOMEM; 1674 1675 reg->hr_slots = kcalloc(reg->hr_blocks, 1676 sizeof(struct o2hb_disk_slot), GFP_KERNEL); 1677 if (reg->hr_slots == NULL) 1678 return -ENOMEM; 1679 1680 for(i = 0; i < reg->hr_blocks; i++) { 1681 slot = ®->hr_slots[i]; 1682 slot->ds_node_num = i; 1683 INIT_LIST_HEAD(&slot->ds_live_item); 1684 slot->ds_raw_block = NULL; 1685 } 1686 1687 reg->hr_num_pages = (reg->hr_blocks + spp - 1) / spp; 1688 mlog(ML_HEARTBEAT, "Going to require %u pages to cover %u blocks " 1689 "at %u blocks per page\n", 1690 reg->hr_num_pages, reg->hr_blocks, spp); 1691 1692 reg->hr_slot_data = kcalloc(reg->hr_num_pages, sizeof(struct page *), 1693 GFP_KERNEL); 1694 if (!reg->hr_slot_data) 1695 return -ENOMEM; 1696 1697 for(i = 0; i < reg->hr_num_pages; i++) { 1698 page = alloc_page(GFP_KERNEL); 1699 if (!page) 1700 return -ENOMEM; 1701 1702 reg->hr_slot_data[i] = page; 1703 1704 last_slot = i * spp; 1705 raw = page_address(page); 1706 for (j = 0; 1707 (j < spp) && ((j + last_slot) < reg->hr_blocks); 1708 j++) { 1709 BUG_ON((j + last_slot) >= reg->hr_blocks); 1710 1711 slot = ®->hr_slots[j + last_slot]; 1712 slot->ds_raw_block = 1713 (struct o2hb_disk_heartbeat_block *) raw; 1714 1715 raw += reg->hr_block_bytes; 1716 } 1717 } 1718 1719 return 0; 1720 } 1721 1722 /* Read in all the slots available and populate the tracking 1723 * structures so that we can start with a baseline idea of what's 1724 * there. */ 1725 static int o2hb_populate_slot_data(struct o2hb_region *reg) 1726 { 1727 int ret, i; 1728 struct o2hb_disk_slot *slot; 1729 struct o2hb_disk_heartbeat_block *hb_block; 1730 1731 ret = o2hb_read_slots(reg, 0, reg->hr_blocks); 1732 if (ret) 1733 goto out; 1734 1735 /* We only want to get an idea of the values initially in each 1736 * slot, so we do no verification - o2hb_check_slot will 1737 * actually determine if each configured slot is valid and 1738 * whether any values have changed. */ 1739 for(i = 0; i < reg->hr_blocks; i++) { 1740 slot = ®->hr_slots[i]; 1741 hb_block = (struct o2hb_disk_heartbeat_block *) slot->ds_raw_block; 1742 1743 /* Only fill the values that o2hb_check_slot uses to 1744 * determine changing slots */ 1745 slot->ds_last_time = le64_to_cpu(hb_block->hb_seq); 1746 slot->ds_last_generation = le64_to_cpu(hb_block->hb_generation); 1747 } 1748 1749 out: 1750 return ret; 1751 } 1752 1753 /* this is acting as commit; we set up all of hr_bdev and hr_task or nothing */ 1754 static ssize_t o2hb_region_dev_store(struct config_item *item, 1755 const char *page, 1756 size_t count) 1757 { 1758 struct o2hb_region *reg = to_o2hb_region(item); 1759 struct task_struct *hb_task; 1760 long fd; 1761 int sectsize; 1762 char *p = (char *)page; 1763 struct fd f; 1764 ssize_t ret = -EINVAL; 1765 int live_threshold; 1766 1767 if (reg->hr_bdev) 1768 goto out; 1769 1770 /* We can't heartbeat without having had our node number 1771 * configured yet. */ 1772 if (o2nm_this_node() == O2NM_MAX_NODES) 1773 goto out; 1774 1775 fd = simple_strtol(p, &p, 0); 1776 if (!p || (*p && (*p != '\n'))) 1777 goto out; 1778 1779 if (fd < 0 || fd >= INT_MAX) 1780 goto out; 1781 1782 f = fdget(fd); 1783 if (f.file == NULL) 1784 goto out; 1785 1786 if (reg->hr_blocks == 0 || reg->hr_start_block == 0 || 1787 reg->hr_block_bytes == 0) 1788 goto out2; 1789 1790 if (!S_ISBLK(f.file->f_mapping->host->i_mode)) 1791 goto out2; 1792 1793 reg->hr_bdev = blkdev_get_by_dev(f.file->f_mapping->host->i_rdev, 1794 FMODE_WRITE | FMODE_READ, NULL); 1795 if (IS_ERR(reg->hr_bdev)) { 1796 ret = PTR_ERR(reg->hr_bdev); 1797 reg->hr_bdev = NULL; 1798 goto out2; 1799 } 1800 1801 bdevname(reg->hr_bdev, reg->hr_dev_name); 1802 1803 sectsize = bdev_logical_block_size(reg->hr_bdev); 1804 if (sectsize != reg->hr_block_bytes) { 1805 mlog(ML_ERROR, 1806 "blocksize %u incorrect for device, expected %d", 1807 reg->hr_block_bytes, sectsize); 1808 ret = -EINVAL; 1809 goto out3; 1810 } 1811 1812 o2hb_init_region_params(reg); 1813 1814 /* Generation of zero is invalid */ 1815 do { 1816 get_random_bytes(®->hr_generation, 1817 sizeof(reg->hr_generation)); 1818 } while (reg->hr_generation == 0); 1819 1820 ret = o2hb_map_slot_data(reg); 1821 if (ret) { 1822 mlog_errno(ret); 1823 goto out3; 1824 } 1825 1826 ret = o2hb_populate_slot_data(reg); 1827 if (ret) { 1828 mlog_errno(ret); 1829 goto out3; 1830 } 1831 1832 INIT_DELAYED_WORK(®->hr_write_timeout_work, o2hb_write_timeout); 1833 INIT_DELAYED_WORK(®->hr_nego_timeout_work, o2hb_nego_timeout); 1834 1835 /* 1836 * A node is considered live after it has beat LIVE_THRESHOLD 1837 * times. We're not steady until we've given them a chance 1838 * _after_ our first read. 1839 * The default threshold is bare minimum so as to limit the delay 1840 * during mounts. For global heartbeat, the threshold doubled for the 1841 * first region. 1842 */ 1843 live_threshold = O2HB_LIVE_THRESHOLD; 1844 if (o2hb_global_heartbeat_active()) { 1845 spin_lock(&o2hb_live_lock); 1846 if (bitmap_weight(o2hb_region_bitmap, O2NM_MAX_REGIONS) == 1) 1847 live_threshold <<= 1; 1848 spin_unlock(&o2hb_live_lock); 1849 } 1850 ++live_threshold; 1851 atomic_set(®->hr_steady_iterations, live_threshold); 1852 /* unsteady_iterations is triple the steady_iterations */ 1853 atomic_set(®->hr_unsteady_iterations, (live_threshold * 3)); 1854 1855 hb_task = kthread_run(o2hb_thread, reg, "o2hb-%s", 1856 reg->hr_item.ci_name); 1857 if (IS_ERR(hb_task)) { 1858 ret = PTR_ERR(hb_task); 1859 mlog_errno(ret); 1860 goto out3; 1861 } 1862 1863 spin_lock(&o2hb_live_lock); 1864 reg->hr_task = hb_task; 1865 spin_unlock(&o2hb_live_lock); 1866 1867 ret = wait_event_interruptible(o2hb_steady_queue, 1868 atomic_read(®->hr_steady_iterations) == 0 || 1869 reg->hr_node_deleted); 1870 if (ret) { 1871 atomic_set(®->hr_steady_iterations, 0); 1872 reg->hr_aborted_start = 1; 1873 } 1874 1875 if (reg->hr_aborted_start) { 1876 ret = -EIO; 1877 goto out3; 1878 } 1879 1880 if (reg->hr_node_deleted) { 1881 ret = -EINVAL; 1882 goto out3; 1883 } 1884 1885 /* Ok, we were woken. Make sure it wasn't by drop_item() */ 1886 spin_lock(&o2hb_live_lock); 1887 hb_task = reg->hr_task; 1888 if (o2hb_global_heartbeat_active()) 1889 set_bit(reg->hr_region_num, o2hb_live_region_bitmap); 1890 spin_unlock(&o2hb_live_lock); 1891 1892 if (hb_task) 1893 ret = count; 1894 else 1895 ret = -EIO; 1896 1897 if (hb_task && o2hb_global_heartbeat_active()) 1898 printk(KERN_NOTICE "o2hb: Heartbeat started on region %s (%s)\n", 1899 config_item_name(®->hr_item), reg->hr_dev_name); 1900 1901 out3: 1902 if (ret < 0) { 1903 blkdev_put(reg->hr_bdev, FMODE_READ | FMODE_WRITE); 1904 reg->hr_bdev = NULL; 1905 } 1906 out2: 1907 fdput(f); 1908 out: 1909 return ret; 1910 } 1911 1912 static ssize_t o2hb_region_pid_show(struct config_item *item, char *page) 1913 { 1914 struct o2hb_region *reg = to_o2hb_region(item); 1915 pid_t pid = 0; 1916 1917 spin_lock(&o2hb_live_lock); 1918 if (reg->hr_task) 1919 pid = task_pid_nr(reg->hr_task); 1920 spin_unlock(&o2hb_live_lock); 1921 1922 if (!pid) 1923 return 0; 1924 1925 return sprintf(page, "%u\n", pid); 1926 } 1927 1928 CONFIGFS_ATTR(o2hb_region_, block_bytes); 1929 CONFIGFS_ATTR(o2hb_region_, start_block); 1930 CONFIGFS_ATTR(o2hb_region_, blocks); 1931 CONFIGFS_ATTR(o2hb_region_, dev); 1932 CONFIGFS_ATTR_RO(o2hb_region_, pid); 1933 1934 static struct configfs_attribute *o2hb_region_attrs[] = { 1935 &o2hb_region_attr_block_bytes, 1936 &o2hb_region_attr_start_block, 1937 &o2hb_region_attr_blocks, 1938 &o2hb_region_attr_dev, 1939 &o2hb_region_attr_pid, 1940 NULL, 1941 }; 1942 1943 static struct configfs_item_operations o2hb_region_item_ops = { 1944 .release = o2hb_region_release, 1945 }; 1946 1947 static const struct config_item_type o2hb_region_type = { 1948 .ct_item_ops = &o2hb_region_item_ops, 1949 .ct_attrs = o2hb_region_attrs, 1950 .ct_owner = THIS_MODULE, 1951 }; 1952 1953 /* heartbeat set */ 1954 1955 struct o2hb_heartbeat_group { 1956 struct config_group hs_group; 1957 /* some stuff? */ 1958 }; 1959 1960 static struct o2hb_heartbeat_group *to_o2hb_heartbeat_group(struct config_group *group) 1961 { 1962 return group ? 1963 container_of(group, struct o2hb_heartbeat_group, hs_group) 1964 : NULL; 1965 } 1966 1967 static void o2hb_debug_region_init(struct o2hb_region *reg, 1968 struct dentry *parent) 1969 { 1970 struct dentry *dir; 1971 1972 dir = debugfs_create_dir(config_item_name(®->hr_item), parent); 1973 reg->hr_debug_dir = dir; 1974 1975 o2hb_debug_create(O2HB_DEBUG_LIVENODES, dir, &(reg->hr_db_livenodes), 1976 sizeof(*(reg->hr_db_livenodes)), 1977 O2HB_DB_TYPE_REGION_LIVENODES, 1978 sizeof(reg->hr_live_node_bitmap), O2NM_MAX_NODES, 1979 reg); 1980 1981 o2hb_debug_create(O2HB_DEBUG_REGION_NUMBER, dir, &(reg->hr_db_regnum), 1982 sizeof(*(reg->hr_db_regnum)), 1983 O2HB_DB_TYPE_REGION_NUMBER, 0, O2NM_MAX_NODES, reg); 1984 1985 o2hb_debug_create(O2HB_DEBUG_REGION_ELAPSED_TIME, dir, 1986 &(reg->hr_db_elapsed_time), 1987 sizeof(*(reg->hr_db_elapsed_time)), 1988 O2HB_DB_TYPE_REGION_ELAPSED_TIME, 0, 0, reg); 1989 1990 o2hb_debug_create(O2HB_DEBUG_REGION_PINNED, dir, &(reg->hr_db_pinned), 1991 sizeof(*(reg->hr_db_pinned)), 1992 O2HB_DB_TYPE_REGION_PINNED, 0, 0, reg); 1993 1994 } 1995 1996 static struct config_item *o2hb_heartbeat_group_make_item(struct config_group *group, 1997 const char *name) 1998 { 1999 struct o2hb_region *reg = NULL; 2000 int ret; 2001 2002 reg = kzalloc(sizeof(struct o2hb_region), GFP_KERNEL); 2003 if (reg == NULL) 2004 return ERR_PTR(-ENOMEM); 2005 2006 if (strlen(name) > O2HB_MAX_REGION_NAME_LEN) { 2007 ret = -ENAMETOOLONG; 2008 goto free; 2009 } 2010 2011 spin_lock(&o2hb_live_lock); 2012 reg->hr_region_num = 0; 2013 if (o2hb_global_heartbeat_active()) { 2014 reg->hr_region_num = find_first_zero_bit(o2hb_region_bitmap, 2015 O2NM_MAX_REGIONS); 2016 if (reg->hr_region_num >= O2NM_MAX_REGIONS) { 2017 spin_unlock(&o2hb_live_lock); 2018 ret = -EFBIG; 2019 goto free; 2020 } 2021 set_bit(reg->hr_region_num, o2hb_region_bitmap); 2022 } 2023 list_add_tail(®->hr_all_item, &o2hb_all_regions); 2024 spin_unlock(&o2hb_live_lock); 2025 2026 config_item_init_type_name(®->hr_item, name, &o2hb_region_type); 2027 2028 /* this is the same way to generate msg key as dlm, for local heartbeat, 2029 * name is also the same, so make initial crc value different to avoid 2030 * message key conflict. 2031 */ 2032 reg->hr_key = crc32_le(reg->hr_region_num + O2NM_MAX_REGIONS, 2033 name, strlen(name)); 2034 INIT_LIST_HEAD(®->hr_handler_list); 2035 ret = o2net_register_handler(O2HB_NEGO_TIMEOUT_MSG, reg->hr_key, 2036 sizeof(struct o2hb_nego_msg), 2037 o2hb_nego_timeout_handler, 2038 reg, NULL, ®->hr_handler_list); 2039 if (ret) 2040 goto remove_item; 2041 2042 ret = o2net_register_handler(O2HB_NEGO_APPROVE_MSG, reg->hr_key, 2043 sizeof(struct o2hb_nego_msg), 2044 o2hb_nego_approve_handler, 2045 reg, NULL, ®->hr_handler_list); 2046 if (ret) 2047 goto unregister_handler; 2048 2049 o2hb_debug_region_init(reg, o2hb_debug_dir); 2050 2051 return ®->hr_item; 2052 2053 unregister_handler: 2054 o2net_unregister_handler_list(®->hr_handler_list); 2055 remove_item: 2056 spin_lock(&o2hb_live_lock); 2057 list_del(®->hr_all_item); 2058 if (o2hb_global_heartbeat_active()) 2059 clear_bit(reg->hr_region_num, o2hb_region_bitmap); 2060 spin_unlock(&o2hb_live_lock); 2061 free: 2062 kfree(reg); 2063 return ERR_PTR(ret); 2064 } 2065 2066 static void o2hb_heartbeat_group_drop_item(struct config_group *group, 2067 struct config_item *item) 2068 { 2069 struct task_struct *hb_task; 2070 struct o2hb_region *reg = to_o2hb_region(item); 2071 int quorum_region = 0; 2072 2073 /* stop the thread when the user removes the region dir */ 2074 spin_lock(&o2hb_live_lock); 2075 hb_task = reg->hr_task; 2076 reg->hr_task = NULL; 2077 reg->hr_item_dropped = 1; 2078 spin_unlock(&o2hb_live_lock); 2079 2080 if (hb_task) 2081 kthread_stop(hb_task); 2082 2083 if (o2hb_global_heartbeat_active()) { 2084 spin_lock(&o2hb_live_lock); 2085 clear_bit(reg->hr_region_num, o2hb_region_bitmap); 2086 clear_bit(reg->hr_region_num, o2hb_live_region_bitmap); 2087 if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) 2088 quorum_region = 1; 2089 clear_bit(reg->hr_region_num, o2hb_quorum_region_bitmap); 2090 spin_unlock(&o2hb_live_lock); 2091 printk(KERN_NOTICE "o2hb: Heartbeat %s on region %s (%s)\n", 2092 ((atomic_read(®->hr_steady_iterations) == 0) ? 2093 "stopped" : "start aborted"), config_item_name(item), 2094 reg->hr_dev_name); 2095 } 2096 2097 /* 2098 * If we're racing a dev_write(), we need to wake them. They will 2099 * check reg->hr_task 2100 */ 2101 if (atomic_read(®->hr_steady_iterations) != 0) { 2102 reg->hr_aborted_start = 1; 2103 atomic_set(®->hr_steady_iterations, 0); 2104 wake_up(&o2hb_steady_queue); 2105 } 2106 2107 config_item_put(item); 2108 2109 if (!o2hb_global_heartbeat_active() || !quorum_region) 2110 return; 2111 2112 /* 2113 * If global heartbeat active and there are dependent users, 2114 * pin all regions if quorum region count <= CUT_OFF 2115 */ 2116 spin_lock(&o2hb_live_lock); 2117 2118 if (!o2hb_dependent_users) 2119 goto unlock; 2120 2121 if (bitmap_weight(o2hb_quorum_region_bitmap, 2122 O2NM_MAX_REGIONS) <= O2HB_PIN_CUT_OFF) 2123 o2hb_region_pin(NULL); 2124 2125 unlock: 2126 spin_unlock(&o2hb_live_lock); 2127 } 2128 2129 static ssize_t o2hb_heartbeat_group_dead_threshold_show(struct config_item *item, 2130 char *page) 2131 { 2132 return sprintf(page, "%u\n", o2hb_dead_threshold); 2133 } 2134 2135 static ssize_t o2hb_heartbeat_group_dead_threshold_store(struct config_item *item, 2136 const char *page, size_t count) 2137 { 2138 unsigned long tmp; 2139 char *p = (char *)page; 2140 2141 tmp = simple_strtoul(p, &p, 10); 2142 if (!p || (*p && (*p != '\n'))) 2143 return -EINVAL; 2144 2145 /* this will validate ranges for us. */ 2146 o2hb_dead_threshold_set((unsigned int) tmp); 2147 2148 return count; 2149 } 2150 2151 static ssize_t o2hb_heartbeat_group_mode_show(struct config_item *item, 2152 char *page) 2153 { 2154 return sprintf(page, "%s\n", 2155 o2hb_heartbeat_mode_desc[o2hb_heartbeat_mode]); 2156 } 2157 2158 static ssize_t o2hb_heartbeat_group_mode_store(struct config_item *item, 2159 const char *page, size_t count) 2160 { 2161 unsigned int i; 2162 int ret; 2163 size_t len; 2164 2165 len = (page[count - 1] == '\n') ? count - 1 : count; 2166 if (!len) 2167 return -EINVAL; 2168 2169 for (i = 0; i < O2HB_HEARTBEAT_NUM_MODES; ++i) { 2170 if (strncasecmp(page, o2hb_heartbeat_mode_desc[i], len)) 2171 continue; 2172 2173 ret = o2hb_global_heartbeat_mode_set(i); 2174 if (!ret) 2175 printk(KERN_NOTICE "o2hb: Heartbeat mode set to %s\n", 2176 o2hb_heartbeat_mode_desc[i]); 2177 return count; 2178 } 2179 2180 return -EINVAL; 2181 2182 } 2183 2184 CONFIGFS_ATTR(o2hb_heartbeat_group_, dead_threshold); 2185 CONFIGFS_ATTR(o2hb_heartbeat_group_, mode); 2186 2187 static struct configfs_attribute *o2hb_heartbeat_group_attrs[] = { 2188 &o2hb_heartbeat_group_attr_dead_threshold, 2189 &o2hb_heartbeat_group_attr_mode, 2190 NULL, 2191 }; 2192 2193 static struct configfs_group_operations o2hb_heartbeat_group_group_ops = { 2194 .make_item = o2hb_heartbeat_group_make_item, 2195 .drop_item = o2hb_heartbeat_group_drop_item, 2196 }; 2197 2198 static const struct config_item_type o2hb_heartbeat_group_type = { 2199 .ct_group_ops = &o2hb_heartbeat_group_group_ops, 2200 .ct_attrs = o2hb_heartbeat_group_attrs, 2201 .ct_owner = THIS_MODULE, 2202 }; 2203 2204 /* this is just here to avoid touching group in heartbeat.h which the 2205 * entire damn world #includes */ 2206 struct config_group *o2hb_alloc_hb_set(void) 2207 { 2208 struct o2hb_heartbeat_group *hs = NULL; 2209 struct config_group *ret = NULL; 2210 2211 hs = kzalloc(sizeof(struct o2hb_heartbeat_group), GFP_KERNEL); 2212 if (hs == NULL) 2213 goto out; 2214 2215 config_group_init_type_name(&hs->hs_group, "heartbeat", 2216 &o2hb_heartbeat_group_type); 2217 2218 ret = &hs->hs_group; 2219 out: 2220 if (ret == NULL) 2221 kfree(hs); 2222 return ret; 2223 } 2224 2225 void o2hb_free_hb_set(struct config_group *group) 2226 { 2227 struct o2hb_heartbeat_group *hs = to_o2hb_heartbeat_group(group); 2228 kfree(hs); 2229 } 2230 2231 /* hb callback registration and issuing */ 2232 2233 static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type) 2234 { 2235 if (type == O2HB_NUM_CB) 2236 return ERR_PTR(-EINVAL); 2237 2238 return &o2hb_callbacks[type]; 2239 } 2240 2241 void o2hb_setup_callback(struct o2hb_callback_func *hc, 2242 enum o2hb_callback_type type, 2243 o2hb_cb_func *func, 2244 void *data, 2245 int priority) 2246 { 2247 INIT_LIST_HEAD(&hc->hc_item); 2248 hc->hc_func = func; 2249 hc->hc_data = data; 2250 hc->hc_priority = priority; 2251 hc->hc_type = type; 2252 hc->hc_magic = O2HB_CB_MAGIC; 2253 } 2254 EXPORT_SYMBOL_GPL(o2hb_setup_callback); 2255 2256 /* 2257 * In local heartbeat mode, region_uuid passed matches the dlm domain name. 2258 * In global heartbeat mode, region_uuid passed is NULL. 2259 * 2260 * In local, we only pin the matching region. In global we pin all the active 2261 * regions. 2262 */ 2263 static int o2hb_region_pin(const char *region_uuid) 2264 { 2265 int ret = 0, found = 0; 2266 struct o2hb_region *reg; 2267 char *uuid; 2268 2269 assert_spin_locked(&o2hb_live_lock); 2270 2271 list_for_each_entry(reg, &o2hb_all_regions, hr_all_item) { 2272 if (reg->hr_item_dropped) 2273 continue; 2274 2275 uuid = config_item_name(®->hr_item); 2276 2277 /* local heartbeat */ 2278 if (region_uuid) { 2279 if (strcmp(region_uuid, uuid)) 2280 continue; 2281 found = 1; 2282 } 2283 2284 if (reg->hr_item_pinned || reg->hr_item_dropped) 2285 goto skip_pin; 2286 2287 /* Ignore ENOENT only for local hb (userdlm domain) */ 2288 ret = o2nm_depend_item(®->hr_item); 2289 if (!ret) { 2290 mlog(ML_CLUSTER, "Pin region %s\n", uuid); 2291 reg->hr_item_pinned = 1; 2292 } else { 2293 if (ret == -ENOENT && found) 2294 ret = 0; 2295 else { 2296 mlog(ML_ERROR, "Pin region %s fails with %d\n", 2297 uuid, ret); 2298 break; 2299 } 2300 } 2301 skip_pin: 2302 if (found) 2303 break; 2304 } 2305 2306 return ret; 2307 } 2308 2309 /* 2310 * In local heartbeat mode, region_uuid passed matches the dlm domain name. 2311 * In global heartbeat mode, region_uuid passed is NULL. 2312 * 2313 * In local, we only unpin the matching region. In global we unpin all the 2314 * active regions. 2315 */ 2316 static void o2hb_region_unpin(const char *region_uuid) 2317 { 2318 struct o2hb_region *reg; 2319 char *uuid; 2320 int found = 0; 2321 2322 assert_spin_locked(&o2hb_live_lock); 2323 2324 list_for_each_entry(reg, &o2hb_all_regions, hr_all_item) { 2325 if (reg->hr_item_dropped) 2326 continue; 2327 2328 uuid = config_item_name(®->hr_item); 2329 if (region_uuid) { 2330 if (strcmp(region_uuid, uuid)) 2331 continue; 2332 found = 1; 2333 } 2334 2335 if (reg->hr_item_pinned) { 2336 mlog(ML_CLUSTER, "Unpin region %s\n", uuid); 2337 o2nm_undepend_item(®->hr_item); 2338 reg->hr_item_pinned = 0; 2339 } 2340 if (found) 2341 break; 2342 } 2343 } 2344 2345 static int o2hb_region_inc_user(const char *region_uuid) 2346 { 2347 int ret = 0; 2348 2349 spin_lock(&o2hb_live_lock); 2350 2351 /* local heartbeat */ 2352 if (!o2hb_global_heartbeat_active()) { 2353 ret = o2hb_region_pin(region_uuid); 2354 goto unlock; 2355 } 2356 2357 /* 2358 * if global heartbeat active and this is the first dependent user, 2359 * pin all regions if quorum region count <= CUT_OFF 2360 */ 2361 o2hb_dependent_users++; 2362 if (o2hb_dependent_users > 1) 2363 goto unlock; 2364 2365 if (bitmap_weight(o2hb_quorum_region_bitmap, 2366 O2NM_MAX_REGIONS) <= O2HB_PIN_CUT_OFF) 2367 ret = o2hb_region_pin(NULL); 2368 2369 unlock: 2370 spin_unlock(&o2hb_live_lock); 2371 return ret; 2372 } 2373 2374 static void o2hb_region_dec_user(const char *region_uuid) 2375 { 2376 spin_lock(&o2hb_live_lock); 2377 2378 /* local heartbeat */ 2379 if (!o2hb_global_heartbeat_active()) { 2380 o2hb_region_unpin(region_uuid); 2381 goto unlock; 2382 } 2383 2384 /* 2385 * if global heartbeat active and there are no dependent users, 2386 * unpin all quorum regions 2387 */ 2388 o2hb_dependent_users--; 2389 if (!o2hb_dependent_users) 2390 o2hb_region_unpin(NULL); 2391 2392 unlock: 2393 spin_unlock(&o2hb_live_lock); 2394 } 2395 2396 int o2hb_register_callback(const char *region_uuid, 2397 struct o2hb_callback_func *hc) 2398 { 2399 struct o2hb_callback_func *f; 2400 struct o2hb_callback *hbcall; 2401 int ret; 2402 2403 BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); 2404 BUG_ON(!list_empty(&hc->hc_item)); 2405 2406 hbcall = hbcall_from_type(hc->hc_type); 2407 if (IS_ERR(hbcall)) { 2408 ret = PTR_ERR(hbcall); 2409 goto out; 2410 } 2411 2412 if (region_uuid) { 2413 ret = o2hb_region_inc_user(region_uuid); 2414 if (ret) { 2415 mlog_errno(ret); 2416 goto out; 2417 } 2418 } 2419 2420 down_write(&o2hb_callback_sem); 2421 2422 list_for_each_entry(f, &hbcall->list, hc_item) { 2423 if (hc->hc_priority < f->hc_priority) { 2424 list_add_tail(&hc->hc_item, &f->hc_item); 2425 break; 2426 } 2427 } 2428 if (list_empty(&hc->hc_item)) 2429 list_add_tail(&hc->hc_item, &hbcall->list); 2430 2431 up_write(&o2hb_callback_sem); 2432 ret = 0; 2433 out: 2434 mlog(ML_CLUSTER, "returning %d on behalf of %p for funcs %p\n", 2435 ret, __builtin_return_address(0), hc); 2436 return ret; 2437 } 2438 EXPORT_SYMBOL_GPL(o2hb_register_callback); 2439 2440 void o2hb_unregister_callback(const char *region_uuid, 2441 struct o2hb_callback_func *hc) 2442 { 2443 BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); 2444 2445 mlog(ML_CLUSTER, "on behalf of %p for funcs %p\n", 2446 __builtin_return_address(0), hc); 2447 2448 /* XXX Can this happen _with_ a region reference? */ 2449 if (list_empty(&hc->hc_item)) 2450 return; 2451 2452 if (region_uuid) 2453 o2hb_region_dec_user(region_uuid); 2454 2455 down_write(&o2hb_callback_sem); 2456 2457 list_del_init(&hc->hc_item); 2458 2459 up_write(&o2hb_callback_sem); 2460 } 2461 EXPORT_SYMBOL_GPL(o2hb_unregister_callback); 2462 2463 int o2hb_check_node_heartbeating_no_sem(u8 node_num) 2464 { 2465 unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)]; 2466 2467 spin_lock(&o2hb_live_lock); 2468 o2hb_fill_node_map_from_callback(testing_map, sizeof(testing_map)); 2469 spin_unlock(&o2hb_live_lock); 2470 if (!test_bit(node_num, testing_map)) { 2471 mlog(ML_HEARTBEAT, 2472 "node (%u) does not have heartbeating enabled.\n", 2473 node_num); 2474 return 0; 2475 } 2476 2477 return 1; 2478 } 2479 EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating_no_sem); 2480 2481 int o2hb_check_node_heartbeating_from_callback(u8 node_num) 2482 { 2483 unsigned long testing_map[BITS_TO_LONGS(O2NM_MAX_NODES)]; 2484 2485 o2hb_fill_node_map_from_callback(testing_map, sizeof(testing_map)); 2486 if (!test_bit(node_num, testing_map)) { 2487 mlog(ML_HEARTBEAT, 2488 "node (%u) does not have heartbeating enabled.\n", 2489 node_num); 2490 return 0; 2491 } 2492 2493 return 1; 2494 } 2495 EXPORT_SYMBOL_GPL(o2hb_check_node_heartbeating_from_callback); 2496 2497 /* 2498 * this is just a hack until we get the plumbing which flips file systems 2499 * read only and drops the hb ref instead of killing the node dead. 2500 */ 2501 void o2hb_stop_all_regions(void) 2502 { 2503 struct o2hb_region *reg; 2504 2505 mlog(ML_ERROR, "stopping heartbeat on all active regions.\n"); 2506 2507 spin_lock(&o2hb_live_lock); 2508 2509 list_for_each_entry(reg, &o2hb_all_regions, hr_all_item) 2510 reg->hr_unclean_stop = 1; 2511 2512 spin_unlock(&o2hb_live_lock); 2513 } 2514 EXPORT_SYMBOL_GPL(o2hb_stop_all_regions); 2515 2516 int o2hb_get_all_regions(char *region_uuids, u8 max_regions) 2517 { 2518 struct o2hb_region *reg; 2519 int numregs = 0; 2520 char *p; 2521 2522 spin_lock(&o2hb_live_lock); 2523 2524 p = region_uuids; 2525 list_for_each_entry(reg, &o2hb_all_regions, hr_all_item) { 2526 if (reg->hr_item_dropped) 2527 continue; 2528 2529 mlog(0, "Region: %s\n", config_item_name(®->hr_item)); 2530 if (numregs < max_regions) { 2531 memcpy(p, config_item_name(®->hr_item), 2532 O2HB_MAX_REGION_NAME_LEN); 2533 p += O2HB_MAX_REGION_NAME_LEN; 2534 } 2535 numregs++; 2536 } 2537 2538 spin_unlock(&o2hb_live_lock); 2539 2540 return numregs; 2541 } 2542 EXPORT_SYMBOL_GPL(o2hb_get_all_regions); 2543 2544 int o2hb_global_heartbeat_active(void) 2545 { 2546 return (o2hb_heartbeat_mode == O2HB_HEARTBEAT_GLOBAL); 2547 } 2548 EXPORT_SYMBOL(o2hb_global_heartbeat_active); 2549