drivers/md/md-cluster.c

   1 /*
   2  * Copyright (C) 2015, SUSE
   3  *
   4  * This program is free software; you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation; either version 2, or (at your option)
   7  * any later version.
   8  *
   9  */
  10
  11
  12 #include <linux/module.h>
  13 #include <linux/kthread.h>
  14 #include <linux/dlm.h>
  15 #include <linux/sched.h>
  16 #include <linux/raid/md_p.h>
  17 #include "md.h"
  18 #include "bitmap.h"
  19 #include "md-cluster.h"
  20
  21 #define LVB_SIZE        64
  22 #define NEW_DEV_TIMEOUT 5000
  23
  24 struct dlm_lock_resource {
  25         dlm_lockspace_t *ls;
  26         struct dlm_lksb lksb;
  27         char *name; /* lock name. */
  28         uint32_t flags; /* flags to pass to dlm_lock() */
  29         wait_queue_head_t sync_locking; /* wait queue for synchronized locking */
  30         bool sync_locking_done;
  31         void (*bast)(void *arg, int mode); /* blocking AST function pointer*/
  32         struct mddev *mddev; /* pointing back to mddev. */
  33         int mode;
  34 };
  35
  36 struct suspend_info {
  37         int slot;
  38         sector_t lo;
  39         sector_t hi;
  40         struct list_head list;
  41 };
  42
  43 struct resync_info {
  44         __le64 lo;
  45         __le64 hi;
  46 };
  47
  48 /* md_cluster_info flags */
  49 #define         MD_CLUSTER_WAITING_FOR_NEWDISK          1
  50 #define         MD_CLUSTER_SUSPEND_READ_BALANCING       2
  51 #define         MD_CLUSTER_BEGIN_JOIN_CLUSTER           3
  52
  53 /* Lock the send communication. This is done through
  54  * bit manipulation as opposed to a mutex in order to
  55  * accomodate lock and hold. See next comment.
  56  */
  57 #define         MD_CLUSTER_SEND_LOCK                    4
  58 /* If cluster operations (such as adding a disk) must lock the
  59  * communication channel, so as to perform extra operations
  60  * (update metadata) and no other operation is allowed on the
  61  * MD. Token needs to be locked and held until the operation
  62  * completes witha md_update_sb(), which would eventually release
  63  * the lock.
  64  */
  65 #define         MD_CLUSTER_SEND_LOCKED_ALREADY          5
  66 /* We should receive message after node joined cluster and
  67  * set up all the related infos such as bitmap and personality */
  68 #define         MD_CLUSTER_ALREADY_IN_CLUSTER           6
  69 #define         MD_CLUSTER_PENDING_RECV_EVENT           7
  70 #define         MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD              8
  71
  72 struct md_cluster_info {
  73         struct mddev *mddev; /* the md device which md_cluster_info belongs to */
  74         /* dlm lock space and resources for clustered raid. */
  75         dlm_lockspace_t *lockspace;
  76         int slot_number;
  77         struct completion completion;
  78         struct mutex recv_mutex;
  79         struct dlm_lock_resource *bitmap_lockres;
  80         struct dlm_lock_resource **other_bitmap_lockres;
  81         struct dlm_lock_resource *resync_lockres;
  82         struct list_head suspend_list;
  83         spinlock_t suspend_lock;
  84         struct md_thread *recovery_thread;
  85         unsigned long recovery_map;
  86         /* communication loc resources */
  87         struct dlm_lock_resource *ack_lockres;
  88         struct dlm_lock_resource *message_lockres;
  89         struct dlm_lock_resource *token_lockres;
  90         struct dlm_lock_resource *no_new_dev_lockres;
  91         struct md_thread *recv_thread;
  92         struct completion newdisk_completion;
  93         wait_queue_head_t wait;
  94         unsigned long state;
  95         /* record the region in RESYNCING message */
  96         sector_t sync_low;
  97         sector_t sync_hi;
  98 };
  99
 100 enum msg_type {
 101         METADATA_UPDATED = 0,
 102         RESYNCING,
 103         NEWDISK,
 104         REMOVE,
 105         RE_ADD,
 106         BITMAP_NEEDS_SYNC,
 107         CHANGE_CAPACITY,
 108 };
 109
 110 struct cluster_msg {
 111         __le32 type;
 112         __le32 slot;
 113         /* TODO: Unionize this for smaller footprint */
 114         __le64 low;
 115         __le64 high;
 116         char uuid[16];
 117         __le32 raid_slot;
 118 };
 119
 120 static void sync_ast(void *arg)
 121 {
 122         struct dlm_lock_resource *res;
 123
 124         res = arg;
 125         res->sync_locking_done = true;
 126         wake_up(&res->sync_locking);
 127 }
 128
 129 static int dlm_lock_sync(struct dlm_lock_resource *res, int mode)
 130 {
 131         int ret = 0;
 132
 133         ret = dlm_lock(res->ls, mode, &res->lksb,
 134                         res->flags, res->name, strlen(res->name),
 135                         0, sync_ast, res, res->bast);
 136         if (ret)
 137                 return ret;
 138         wait_event(res->sync_locking, res->sync_locking_done);
 139         res->sync_locking_done = false;
 140         if (res->lksb.sb_status == 0)
 141                 res->mode = mode;
 142         return res->lksb.sb_status;
 143 }
 144
 145 static int dlm_unlock_sync(struct dlm_lock_resource *res)
 146 {
 147         return dlm_lock_sync(res, DLM_LOCK_NL);
 148 }
 149
 150 /*
 151  * An variation of dlm_lock_sync, which make lock request could
 152  * be interrupted
 153  */
 154 static int dlm_lock_sync_interruptible(struct dlm_lock_resource *res, int mode,
 155                                        struct mddev *mddev)
 156 {
 157         int ret = 0;
 158
 159         ret = dlm_lock(res->ls, mode, &res->lksb,
 160                         res->flags, res->name, strlen(res->name),
 161                         0, sync_ast, res, res->bast);
 162         if (ret)
 163                 return ret;
 164
 165         wait_event(res->sync_locking, res->sync_locking_done
 166                                       || kthread_should_stop()
 167                                       || test_bit(MD_CLOSING, &mddev->flags));
 168         if (!res->sync_locking_done) {
 169                 /*
 170                  * the convert queue contains the lock request when request is
 171                  * interrupted, and sync_ast could still be run, so need to
 172                  * cancel the request and reset completion
 173                  */
 174                 ret = dlm_unlock(res->ls, res->lksb.sb_lkid, DLM_LKF_CANCEL,
 175                         &res->lksb, res);
 176                 res->sync_locking_done = false;
 177                 if (unlikely(ret != 0))
 178                         pr_info("failed to cancel previous lock request "
 179                                  "%s return %d\n", res->name, ret);
 180                 return -EPERM;
 181         } else
 182                 res->sync_locking_done = false;
 183         if (res->lksb.sb_status == 0)
 184                 res->mode = mode;
 185         return res->lksb.sb_status;
 186 }
 187
 188 static struct dlm_lock_resource *lockres_init(struct mddev *mddev,
 189                 char *name, void (*bastfn)(void *arg, int mode), int with_lvb)
 190 {
 191         struct dlm_lock_resource *res = NULL;
 192         int ret, namelen;
 193         struct md_cluster_info *cinfo = mddev->cluster_info;
 194
 195         res = kzalloc(sizeof(struct dlm_lock_resource), GFP_KERNEL);
 196         if (!res)
 197                 return NULL;
 198         init_waitqueue_head(&res->sync_locking);
 199         res->sync_locking_done = false;
 200         res->ls = cinfo->lockspace;
 201         res->mddev = mddev;
 202         res->mode = DLM_LOCK_IV;
 203         namelen = strlen(name);
 204         res->name = kzalloc(namelen + 1, GFP_KERNEL);
 205         if (!res->name) {
 206                 pr_err("md-cluster: Unable to allocate resource name for resource %s\n", name);
 207                 goto out_err;
 208         }
 209         strlcpy(res->name, name, namelen + 1);
 210         if (with_lvb) {
 211                 res->lksb.sb_lvbptr = kzalloc(LVB_SIZE, GFP_KERNEL);
 212                 if (!res->lksb.sb_lvbptr) {
 213                         pr_err("md-cluster: Unable to allocate LVB for resource %s\n", name);
 214                         goto out_err;
 215                 }
 216                 res->flags = DLM_LKF_VALBLK;
 217         }
 218
 219         if (bastfn)
 220                 res->bast = bastfn;
 221
 222         res->flags |= DLM_LKF_EXPEDITE;
 223
 224         ret = dlm_lock_sync(res, DLM_LOCK_NL);
 225         if (ret) {
 226                 pr_err("md-cluster: Unable to lock NL on new lock resource %s\n", name);
 227                 goto out_err;
 228         }
 229         res->flags &= ~DLM_LKF_EXPEDITE;
 230         res->flags |= DLM_LKF_CONVERT;
 231
 232         return res;
 233 out_err:
 234         kfree(res->lksb.sb_lvbptr);
 235         kfree(res->name);
 236         kfree(res);
 237         return NULL;
 238 }
 239
 240 static void lockres_free(struct dlm_lock_resource *res)
 241 {
 242         int ret = 0;
 243
 244         if (!res)
 245                 return;
 246
 247         /*
 248          * use FORCEUNLOCK flag, so we can unlock even the lock is on the
 249          * waiting or convert queue
 250          */
 251         ret = dlm_unlock(res->ls, res->lksb.sb_lkid, DLM_LKF_FORCEUNLOCK,
 252                 &res->lksb, res);
 253         if (unlikely(ret != 0))
 254                 pr_err("failed to unlock %s return %d\n", res->name, ret);
 255         else
 256                 wait_event(res->sync_locking, res->sync_locking_done);
 257
 258         kfree(res->name);
 259         kfree(res->lksb.sb_lvbptr);
 260         kfree(res);
 261 }
 262
 263 static void add_resync_info(struct dlm_lock_resource *lockres,
 264                             sector_t lo, sector_t hi)
 265 {
 266         struct resync_info *ri;
 267
 268         ri = (struct resync_info *)lockres->lksb.sb_lvbptr;
 269         ri->lo = cpu_to_le64(lo);
 270         ri->hi = cpu_to_le64(hi);
 271 }
 272
 273 static struct suspend_info *read_resync_info(struct mddev *mddev, struct dlm_lock_resource *lockres)
 274 {
 275         struct resync_info ri;
 276         struct suspend_info *s = NULL;
 277         sector_t hi = 0;
 278
 279         dlm_lock_sync(lockres, DLM_LOCK_CR);
 280         memcpy(&ri, lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
 281         hi = le64_to_cpu(ri.hi);
 282         if (hi > 0) {
 283                 s = kzalloc(sizeof(struct suspend_info), GFP_KERNEL);
 284                 if (!s)
 285                         goto out;
 286                 s->hi = hi;
 287                 s->lo = le64_to_cpu(ri.lo);
 288         }
 289         dlm_unlock_sync(lockres);
 290 out:
 291         return s;
 292 }
 293
 294 static void recover_bitmaps(struct md_thread *thread)
 295 {
 296         struct mddev *mddev = thread->mddev;
 297         struct md_cluster_info *cinfo = mddev->cluster_info;
 298         struct dlm_lock_resource *bm_lockres;
 299         char str[64];
 300         int slot, ret;
 301         struct suspend_info *s, *tmp;
 302         sector_t lo, hi;
 303
 304         while (cinfo->recovery_map) {
 305                 slot = fls64((u64)cinfo->recovery_map) - 1;
 306
 307                 /* Clear suspend_area associated with the bitmap */
 308                 spin_lock_irq(&cinfo->suspend_lock);
 309                 list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
 310                         if (slot == s->slot) {
 311                                 list_del(&s->list);
 312                                 kfree(s);
 313                         }
 314                 spin_unlock_irq(&cinfo->suspend_lock);
 315
 316                 snprintf(str, 64, "bitmap%04d", slot);
 317                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 318                 if (!bm_lockres) {
 319                         pr_err("md-cluster: Cannot initialize bitmaps\n");
 320                         goto clear_bit;
 321                 }
 322
 323                 ret = dlm_lock_sync_interruptible(bm_lockres, DLM_LOCK_PW, mddev);
 324                 if (ret) {
 325                         pr_err("md-cluster: Could not DLM lock %s: %d\n",
 326                                         str, ret);
 327                         goto clear_bit;
 328                 }
 329                 ret = bitmap_copy_from_slot(mddev, slot, &lo, &hi, true);
 330                 if (ret) {
 331                         pr_err("md-cluster: Could not copy data from bitmap %d\n", slot);
 332                         goto clear_bit;
 333                 }
 334                 if (hi > 0) {
 335                         if (lo < mddev->recovery_cp)
 336                                 mddev->recovery_cp = lo;
 337                         /* wake up thread to continue resync in case resync
 338                          * is not finished */
 339                         if (mddev->recovery_cp != MaxSector) {
 340                             set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 341                             md_wakeup_thread(mddev->thread);
 342                         }
 343                 }
 344 clear_bit:
 345                 lockres_free(bm_lockres);
 346                 clear_bit(slot, &cinfo->recovery_map);
 347         }
 348 }
 349
 350 static void recover_prep(void *arg)
 351 {
 352         struct mddev *mddev = arg;
 353         struct md_cluster_info *cinfo = mddev->cluster_info;
 354         set_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 355 }
 356
 357 static void __recover_slot(struct mddev *mddev, int slot)
 358 {
 359         struct md_cluster_info *cinfo = mddev->cluster_info;
 360
 361         set_bit(slot, &cinfo->recovery_map);
 362         if (!cinfo->recovery_thread) {
 363                 cinfo->recovery_thread = md_register_thread(recover_bitmaps,
 364                                 mddev, "recover");
 365                 if (!cinfo->recovery_thread) {
 366                         pr_warn("md-cluster: Could not create recovery thread\n");
 367                         return;
 368                 }
 369         }
 370         md_wakeup_thread(cinfo->recovery_thread);
 371 }
 372
 373 static void recover_slot(void *arg, struct dlm_slot *slot)
 374 {
 375         struct mddev *mddev = arg;
 376         struct md_cluster_info *cinfo = mddev->cluster_info;
 377
 378         pr_info("md-cluster: %s Node %d/%d down. My slot: %d. Initiating recovery.\n",
 379                         mddev->bitmap_info.cluster_name,
 380                         slot->nodeid, slot->slot,
 381                         cinfo->slot_number);
 382         /* deduct one since dlm slot starts from one while the num of
 383          * cluster-md begins with 0 */
 384         __recover_slot(mddev, slot->slot - 1);
 385 }
 386
 387 static void recover_done(void *arg, struct dlm_slot *slots,
 388                 int num_slots, int our_slot,
 389                 uint32_t generation)
 390 {
 391         struct mddev *mddev = arg;
 392         struct md_cluster_info *cinfo = mddev->cluster_info;
 393
 394         cinfo->slot_number = our_slot;
 395         /* completion is only need to be complete when node join cluster,
 396          * it doesn't need to run during another node's failure */
 397         if (test_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state)) {
 398                 complete(&cinfo->completion);
 399                 clear_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 400         }
 401         clear_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 402 }
 403
 404 /* the ops is called when node join the cluster, and do lock recovery
 405  * if node failure occurs */
 406 static const struct dlm_lockspace_ops md_ls_ops = {
 407         .recover_prep = recover_prep,
 408         .recover_slot = recover_slot,
 409         .recover_done = recover_done,
 410 };
 411
 412 /*
 413  * The BAST function for the ack lock resource
 414  * This function wakes up the receive thread in
 415  * order to receive and process the message.
 416  */
 417 static void ack_bast(void *arg, int mode)
 418 {
 419         struct dlm_lock_resource *res = arg;
 420         struct md_cluster_info *cinfo = res->mddev->cluster_info;
 421
 422         if (mode == DLM_LOCK_EX) {
 423                 if (test_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state))
 424                         md_wakeup_thread(cinfo->recv_thread);
 425                 else
 426                         set_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state);
 427         }
 428 }
 429
 430 static void __remove_suspend_info(struct md_cluster_info *cinfo, int slot)
 431 {
 432         struct suspend_info *s, *tmp;
 433
 434         list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
 435                 if (slot == s->slot) {
 436                         list_del(&s->list);
 437                         kfree(s);
 438                         break;
 439                 }
 440 }
 441
 442 static void remove_suspend_info(struct mddev *mddev, int slot)
 443 {
 444         struct md_cluster_info *cinfo = mddev->cluster_info;
 445         spin_lock_irq(&cinfo->suspend_lock);
 446         __remove_suspend_info(cinfo, slot);
 447         spin_unlock_irq(&cinfo->suspend_lock);
 448         mddev->pers->quiesce(mddev, 2);
 449 }
 450
 451
 452 static void process_suspend_info(struct mddev *mddev,
 453                 int slot, sector_t lo, sector_t hi)
 454 {
 455         struct md_cluster_info *cinfo = mddev->cluster_info;
 456         struct suspend_info *s;
 457
 458         if (!hi) {
 459                 remove_suspend_info(mddev, slot);
 460                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 461                 md_wakeup_thread(mddev->thread);
 462                 return;
 463         }
 464
 465         /*
 466          * The bitmaps are not same for different nodes
 467          * if RESYNCING is happening in one node, then
 468          * the node which received the RESYNCING message
 469          * probably will perform resync with the region
 470          * [lo, hi] again, so we could reduce resync time
 471          * a lot if we can ensure that the bitmaps among
 472          * different nodes are match up well.
 473          *
 474          * sync_low/hi is used to record the region which
 475          * arrived in the previous RESYNCING message,
 476          *
 477          * Call bitmap_sync_with_cluster to clear
 478          * NEEDED_MASK and set RESYNC_MASK since
 479          * resync thread is running in another node,
 480          * so we don't need to do the resync again
 481          * with the same section */
 482         bitmap_sync_with_cluster(mddev, cinfo->sync_low,
 483                                         cinfo->sync_hi,
 484                                         lo, hi);
 485         cinfo->sync_low = lo;
 486         cinfo->sync_hi = hi;
 487
 488         s = kzalloc(sizeof(struct suspend_info), GFP_KERNEL);
 489         if (!s)
 490                 return;
 491         s->slot = slot;
 492         s->lo = lo;
 493         s->hi = hi;
 494         mddev->pers->quiesce(mddev, 1);
 495         mddev->pers->quiesce(mddev, 0);
 496         spin_lock_irq(&cinfo->suspend_lock);
 497         /* Remove existing entry (if exists) before adding */
 498         __remove_suspend_info(cinfo, slot);
 499         list_add(&s->list, &cinfo->suspend_list);
 500         spin_unlock_irq(&cinfo->suspend_lock);
 501         mddev->pers->quiesce(mddev, 2);
 502 }
 503
 504 static void process_add_new_disk(struct mddev *mddev, struct cluster_msg *cmsg)
 505 {
 506         char disk_uuid[64];
 507         struct md_cluster_info *cinfo = mddev->cluster_info;
 508         char event_name[] = "EVENT=ADD_DEVICE";
 509         char raid_slot[16];
 510         char *envp[] = {event_name, disk_uuid, raid_slot, NULL};
 511         int len;
 512
 513         len = snprintf(disk_uuid, 64, "DEVICE_UUID=");
 514         sprintf(disk_uuid + len, "%pU", cmsg->uuid);
 515         snprintf(raid_slot, 16, "RAID_DISK=%d", le32_to_cpu(cmsg->raid_slot));
 516         pr_info("%s:%d Sending kobject change with %s and %s\n", __func__, __LINE__, disk_uuid, raid_slot);
 517         init_completion(&cinfo->newdisk_completion);
 518         set_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
 519         kobject_uevent_env(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE, envp);
 520         wait_for_completion_timeout(&cinfo->newdisk_completion,
 521                         NEW_DEV_TIMEOUT);
 522         clear_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
 523 }
 524
 525
 526 static void process_metadata_update(struct mddev *mddev, struct cluster_msg *msg)
 527 {
 528         int got_lock = 0;
 529         struct md_cluster_info *cinfo = mddev->cluster_info;
 530         mddev->good_device_nr = le32_to_cpu(msg->raid_slot);
 531
 532         dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
 533         wait_event(mddev->thread->wqueue,
 534                    (got_lock = mddev_trylock(mddev)) ||
 535                     test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state));
 536         md_reload_sb(mddev, mddev->good_device_nr);
 537         if (got_lock)
 538                 mddev_unlock(mddev);
 539 }
 540
 541 static void process_remove_disk(struct mddev *mddev, struct cluster_msg *msg)
 542 {
 543         struct md_rdev *rdev;
 544
 545         rcu_read_lock();
 546         rdev = md_find_rdev_nr_rcu(mddev, le32_to_cpu(msg->raid_slot));
 547         if (rdev) {
 548                 set_bit(ClusterRemove, &rdev->flags);
 549                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 550                 md_wakeup_thread(mddev->thread);
 551         }
 552         else
 553                 pr_warn("%s: %d Could not find disk(%d) to REMOVE\n",
 554                         __func__, __LINE__, le32_to_cpu(msg->raid_slot));
 555         rcu_read_unlock();
 556 }
 557
 558 static void process_readd_disk(struct mddev *mddev, struct cluster_msg *msg)
 559 {
 560         struct md_rdev *rdev;
 561
 562         rcu_read_lock();
 563         rdev = md_find_rdev_nr_rcu(mddev, le32_to_cpu(msg->raid_slot));
 564         if (rdev && test_bit(Faulty, &rdev->flags))
 565                 clear_bit(Faulty, &rdev->flags);
 566         else
 567                 pr_warn("%s: %d Could not find disk(%d) which is faulty",
 568                         __func__, __LINE__, le32_to_cpu(msg->raid_slot));
 569         rcu_read_unlock();
 570 }
 571
 572 static int process_recvd_msg(struct mddev *mddev, struct cluster_msg *msg)
 573 {
 574         int ret = 0;
 575
 576         if (WARN(mddev->cluster_info->slot_number - 1 == le32_to_cpu(msg->slot),
 577                 "node %d received it's own msg\n", le32_to_cpu(msg->slot)))
 578                 return -1;
 579         switch (le32_to_cpu(msg->type)) {
 580         case METADATA_UPDATED:
 581                 process_metadata_update(mddev, msg);
 582                 break;
 583         case CHANGE_CAPACITY:
 584                 set_capacity(mddev->gendisk, mddev->array_sectors);
 585                 revalidate_disk(mddev->gendisk);
 586                 break;
 587         case RESYNCING:
 588                 process_suspend_info(mddev, le32_to_cpu(msg->slot),
 589                                      le64_to_cpu(msg->low),
 590                                      le64_to_cpu(msg->high));
 591                 break;
 592         case NEWDISK:
 593                 process_add_new_disk(mddev, msg);
 594                 break;
 595         case REMOVE:
 596                 process_remove_disk(mddev, msg);
 597                 break;
 598         case RE_ADD:
 599                 process_readd_disk(mddev, msg);
 600                 break;
 601         case BITMAP_NEEDS_SYNC:
 602                 __recover_slot(mddev, le32_to_cpu(msg->slot));
 603                 break;
 604         default:
 605                 ret = -1;
 606                 pr_warn("%s:%d Received unknown message from %d\n",
 607                         __func__, __LINE__, msg->slot);
 608         }
 609         return ret;
 610 }
 611
 612 /*
 613  * thread for receiving message
 614  */
 615 static void recv_daemon(struct md_thread *thread)
 616 {
 617         struct md_cluster_info *cinfo = thread->mddev->cluster_info;
 618         struct dlm_lock_resource *ack_lockres = cinfo->ack_lockres;
 619         struct dlm_lock_resource *message_lockres = cinfo->message_lockres;
 620         struct cluster_msg msg;
 621         int ret;
 622
 623         mutex_lock(&cinfo->recv_mutex);
 624         /*get CR on Message*/
 625         if (dlm_lock_sync(message_lockres, DLM_LOCK_CR)) {
 626                 pr_err("md/raid1:failed to get CR on MESSAGE\n");
 627                 mutex_unlock(&cinfo->recv_mutex);
 628                 return;
 629         }
 630
 631         /* read lvb and wake up thread to process this message_lockres */
 632         memcpy(&msg, message_lockres->lksb.sb_lvbptr, sizeof(struct cluster_msg));
 633         ret = process_recvd_msg(thread->mddev, &msg);
 634         if (ret)
 635                 goto out;
 636
 637         /*release CR on ack_lockres*/
 638         ret = dlm_unlock_sync(ack_lockres);
 639         if (unlikely(ret != 0))
 640                 pr_info("unlock ack failed return %d\n", ret);
 641         /*up-convert to PR on message_lockres*/
 642         ret = dlm_lock_sync(message_lockres, DLM_LOCK_PR);
 643         if (unlikely(ret != 0))
 644                 pr_info("lock PR on msg failed return %d\n", ret);
 645         /*get CR on ack_lockres again*/
 646         ret = dlm_lock_sync(ack_lockres, DLM_LOCK_CR);
 647         if (unlikely(ret != 0))
 648                 pr_info("lock CR on ack failed return %d\n", ret);
 649 out:
 650         /*release CR on message_lockres*/
 651         ret = dlm_unlock_sync(message_lockres);
 652         if (unlikely(ret != 0))
 653                 pr_info("unlock msg failed return %d\n", ret);
 654         mutex_unlock(&cinfo->recv_mutex);
 655 }
 656
 657 /* lock_token()
 658  * Takes the lock on the TOKEN lock resource so no other
 659  * node can communicate while the operation is underway.
 660  */
 661 static int lock_token(struct md_cluster_info *cinfo, bool mddev_locked)
 662 {
 663         int error, set_bit = 0;
 664         struct mddev *mddev = cinfo->mddev;
 665
 666         /*
 667          * If resync thread run after raid1d thread, then process_metadata_update
 668          * could not continue if raid1d held reconfig_mutex (and raid1d is blocked
 669          * since another node already got EX on Token and waitting the EX of Ack),
 670          * so let resync wake up thread in case flag is set.
 671          */
 672         if (mddev_locked && !test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 673                                       &cinfo->state)) {
 674                 error = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 675                                               &cinfo->state);
 676                 WARN_ON_ONCE(error);
 677                 md_wakeup_thread(mddev->thread);
 678                 set_bit = 1;
 679         }
 680         error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 681         if (set_bit)
 682                 clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 683
 684         if (error)
 685                 pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
 686                                 __func__, __LINE__, error);
 687
 688         /* Lock the receive sequence */
 689         mutex_lock(&cinfo->recv_mutex);
 690         return error;
 691 }
 692
 693 /* lock_comm()
 694  * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
 695  */
 696 static int lock_comm(struct md_cluster_info *cinfo, bool mddev_locked)
 697 {
 698         wait_event(cinfo->wait,
 699                    !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));
 700
 701         return lock_token(cinfo, mddev_locked);
 702 }
 703
 704 static void unlock_comm(struct md_cluster_info *cinfo)
 705 {
 706         WARN_ON(cinfo->token_lockres->mode != DLM_LOCK_EX);
 707         mutex_unlock(&cinfo->recv_mutex);
 708         dlm_unlock_sync(cinfo->token_lockres);
 709         clear_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state);
 710         wake_up(&cinfo->wait);
 711 }
 712
 713 /* __sendmsg()
 714  * This function performs the actual sending of the message. This function is
 715  * usually called after performing the encompassing operation
 716  * The function:
 717  * 1. Grabs the message lockresource in EX mode
 718  * 2. Copies the message to the message LVB
 719  * 3. Downconverts message lockresource to CW
 720  * 4. Upconverts ack lock resource from CR to EX. This forces the BAST on other nodes
 721  *    and the other nodes read the message. The thread will wait here until all other
 722  *    nodes have released ack lock resource.
 723  * 5. Downconvert ack lockresource to CR
 724  */
 725 static int __sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
 726 {
 727         int error;
 728         int slot = cinfo->slot_number - 1;
 729
 730         cmsg->slot = cpu_to_le32(slot);
 731         /*get EX on Message*/
 732         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_EX);
 733         if (error) {
 734                 pr_err("md-cluster: failed to get EX on MESSAGE (%d)\n", error);
 735                 goto failed_message;
 736         }
 737
 738         memcpy(cinfo->message_lockres->lksb.sb_lvbptr, (void *)cmsg,
 739                         sizeof(struct cluster_msg));
 740         /*down-convert EX to CW on Message*/
 741         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_CW);
 742         if (error) {
 743                 pr_err("md-cluster: failed to convert EX to CW on MESSAGE(%d)\n",
 744                                 error);
 745                 goto failed_ack;
 746         }
 747
 748         /*up-convert CR to EX on Ack*/
 749         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_EX);
 750         if (error) {
 751                 pr_err("md-cluster: failed to convert CR to EX on ACK(%d)\n",
 752                                 error);
 753                 goto failed_ack;
 754         }
 755
 756         /*down-convert EX to CR on Ack*/
 757         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR);
 758         if (error) {
 759                 pr_err("md-cluster: failed to convert EX to CR on ACK(%d)\n",
 760                                 error);
 761                 goto failed_ack;
 762         }
 763
 764 failed_ack:
 765         error = dlm_unlock_sync(cinfo->message_lockres);
 766         if (unlikely(error != 0)) {
 767                 pr_err("md-cluster: failed convert to NL on MESSAGE(%d)\n",
 768                         error);
 769                 /* in case the message can't be released due to some reason */
 770                 goto failed_ack;
 771         }
 772 failed_message:
 773         return error;
 774 }
 775
 776 static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg,
 777                    bool mddev_locked)
 778 {
 779         int ret;
 780
 781         lock_comm(cinfo, mddev_locked);
 782         ret = __sendmsg(cinfo, cmsg);
 783         unlock_comm(cinfo);
 784         return ret;
 785 }
 786
 787 static int gather_all_resync_info(struct mddev *mddev, int total_slots)
 788 {
 789         struct md_cluster_info *cinfo = mddev->cluster_info;
 790         int i, ret = 0;
 791         struct dlm_lock_resource *bm_lockres;
 792         struct suspend_info *s;
 793         char str[64];
 794         sector_t lo, hi;
 795
 796
 797         for (i = 0; i < total_slots; i++) {
 798                 memset(str, '\0', 64);
 799                 snprintf(str, 64, "bitmap%04d", i);
 800                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 801                 if (!bm_lockres)
 802                         return -ENOMEM;
 803                 if (i == (cinfo->slot_number - 1)) {
 804                         lockres_free(bm_lockres);
 805                         continue;
 806                 }
 807
 808                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
 809                 ret = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
 810                 if (ret == -EAGAIN) {
 811                         s = read_resync_info(mddev, bm_lockres);
 812                         if (s) {
 813                                 pr_info("%s:%d Resync[%llu..%llu] in progress on %d\n",
 814                                                 __func__, __LINE__,
 815                                                 (unsigned long long) s->lo,
 816                                                 (unsigned long long) s->hi, i);
 817                                 spin_lock_irq(&cinfo->suspend_lock);
 818                                 s->slot = i;
 819                                 list_add(&s->list, &cinfo->suspend_list);
 820                                 spin_unlock_irq(&cinfo->suspend_lock);
 821                         }
 822                         ret = 0;
 823                         lockres_free(bm_lockres);
 824                         continue;
 825                 }
 826                 if (ret) {
 827                         lockres_free(bm_lockres);
 828                         goto out;
 829                 }
 830
 831                 /* Read the disk bitmap sb and check if it needs recovery */
 832                 ret = bitmap_copy_from_slot(mddev, i, &lo, &hi, false);
 833                 if (ret) {
 834                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", i);
 835                         lockres_free(bm_lockres);
 836                         continue;
 837                 }
 838                 if ((hi > 0) && (lo < mddev->recovery_cp)) {
 839                         set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 840                         mddev->recovery_cp = lo;
 841                         md_check_recovery(mddev);
 842                 }
 843
 844                 lockres_free(bm_lockres);
 845         }
 846 out:
 847         return ret;
 848 }
 849
 850 static int join(struct mddev *mddev, int nodes)
 851 {
 852         struct md_cluster_info *cinfo;
 853         int ret, ops_rv;
 854         char str[64];
 855
 856         cinfo = kzalloc(sizeof(struct md_cluster_info), GFP_KERNEL);
 857         if (!cinfo)
 858                 return -ENOMEM;
 859
 860         INIT_LIST_HEAD(&cinfo->suspend_list);
 861         spin_lock_init(&cinfo->suspend_lock);
 862         init_completion(&cinfo->completion);
 863         set_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 864         init_waitqueue_head(&cinfo->wait);
 865         mutex_init(&cinfo->recv_mutex);
 866
 867         mddev->cluster_info = cinfo;
 868         cinfo->mddev = mddev;
 869
 870         memset(str, 0, 64);
 871         sprintf(str, "%pU", mddev->uuid);
 872         ret = dlm_new_lockspace(str, mddev->bitmap_info.cluster_name,
 873                                 DLM_LSFL_FS, LVB_SIZE,
 874                                 &md_ls_ops, mddev, &ops_rv, &cinfo->lockspace);
 875         if (ret)
 876                 goto err;
 877         wait_for_completion(&cinfo->completion);
 878         if (nodes < cinfo->slot_number) {
 879                 pr_err("md-cluster: Slot allotted(%d) is greater than available slots(%d).",
 880                         cinfo->slot_number, nodes);
 881                 ret = -ERANGE;
 882                 goto err;
 883         }
 884         /* Initiate the communication resources */
 885         ret = -ENOMEM;
 886         cinfo->recv_thread = md_register_thread(recv_daemon, mddev, "cluster_recv");
 887         if (!cinfo->recv_thread) {
 888                 pr_err("md-cluster: cannot allocate memory for recv_thread!\n");
 889                 goto err;
 890         }
 891         cinfo->message_lockres = lockres_init(mddev, "message", NULL, 1);
 892         if (!cinfo->message_lockres)
 893                 goto err;
 894         cinfo->token_lockres = lockres_init(mddev, "token", NULL, 0);
 895         if (!cinfo->token_lockres)
 896                 goto err;
 897         cinfo->no_new_dev_lockres = lockres_init(mddev, "no-new-dev", NULL, 0);
 898         if (!cinfo->no_new_dev_lockres)
 899                 goto err;
 900
 901         ret = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 902         if (ret) {
 903                 ret = -EAGAIN;
 904                 pr_err("md-cluster: can't join cluster to avoid lock issue\n");
 905                 goto err;
 906         }
 907         cinfo->ack_lockres = lockres_init(mddev, "ack", ack_bast, 0);
 908         if (!cinfo->ack_lockres) {
 909                 ret = -ENOMEM;
 910                 goto err;
 911         }
 912         /* get sync CR lock on ACK. */
 913         if (dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR))
 914                 pr_err("md-cluster: failed to get a sync CR lock on ACK!(%d)\n",
 915                                 ret);
 916         dlm_unlock_sync(cinfo->token_lockres);
 917         /* get sync CR lock on no-new-dev. */
 918         if (dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR))
 919                 pr_err("md-cluster: failed to get a sync CR lock on no-new-dev!(%d)\n", ret);
 920
 921
 922         pr_info("md-cluster: Joined cluster %s slot %d\n", str, cinfo->slot_number);
 923         snprintf(str, 64, "bitmap%04d", cinfo->slot_number - 1);
 924         cinfo->bitmap_lockres = lockres_init(mddev, str, NULL, 1);
 925         if (!cinfo->bitmap_lockres) {
 926                 ret = -ENOMEM;
 927                 goto err;
 928         }
 929         if (dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW)) {
 930                 pr_err("Failed to get bitmap lock\n");
 931                 ret = -EINVAL;
 932                 goto err;
 933         }
 934
 935         cinfo->resync_lockres = lockres_init(mddev, "resync", NULL, 0);
 936         if (!cinfo->resync_lockres) {
 937                 ret = -ENOMEM;
 938                 goto err;
 939         }
 940
 941         return 0;
 942 err:
 943         set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 944         md_unregister_thread(&cinfo->recovery_thread);
 945         md_unregister_thread(&cinfo->recv_thread);
 946         lockres_free(cinfo->message_lockres);
 947         lockres_free(cinfo->token_lockres);
 948         lockres_free(cinfo->ack_lockres);
 949         lockres_free(cinfo->no_new_dev_lockres);
 950         lockres_free(cinfo->resync_lockres);
 951         lockres_free(cinfo->bitmap_lockres);
 952         if (cinfo->lockspace)
 953                 dlm_release_lockspace(cinfo->lockspace, 2);
 954         mddev->cluster_info = NULL;
 955         kfree(cinfo);
 956         return ret;
 957 }
 958
 959 static void load_bitmaps(struct mddev *mddev, int total_slots)
 960 {
 961         struct md_cluster_info *cinfo = mddev->cluster_info;
 962
 963         /* load all the node's bitmap info for resync */
 964         if (gather_all_resync_info(mddev, total_slots))
 965                 pr_err("md-cluster: failed to gather all resyn infos\n");
 966         set_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state);
 967         /* wake up recv thread in case something need to be handled */
 968         if (test_and_clear_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state))
 969                 md_wakeup_thread(cinfo->recv_thread);
 970 }
 971
 972 static void resync_bitmap(struct mddev *mddev)
 973 {
 974         struct md_cluster_info *cinfo = mddev->cluster_info;
 975         struct cluster_msg cmsg = {0};
 976         int err;
 977
 978         cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
 979         err = sendmsg(cinfo, &cmsg, 1);
 980         if (err)
 981                 pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
 982                         __func__, __LINE__, err);
 983 }
 984
 985 static void unlock_all_bitmaps(struct mddev *mddev);
 986 static int leave(struct mddev *mddev)
 987 {
 988         struct md_cluster_info *cinfo = mddev->cluster_info;
 989
 990         if (!cinfo)
 991                 return 0;
 992
 993         /* BITMAP_NEEDS_SYNC message should be sent when node
 994          * is leaving the cluster with dirty bitmap, also we
 995          * can only deliver it when dlm connection is available */
 996         if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector)
 997                 resync_bitmap(mddev);
 998
 999         set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1000         md_unregister_thread(&cinfo->recovery_thread);
1001         md_unregister_thread(&cinfo->recv_thread);
1002         lockres_free(cinfo->message_lockres);
1003         lockres_free(cinfo->token_lockres);
1004         lockres_free(cinfo->ack_lockres);
1005         lockres_free(cinfo->no_new_dev_lockres);
1006         lockres_free(cinfo->resync_lockres);
1007         lockres_free(cinfo->bitmap_lockres);
1008         unlock_all_bitmaps(mddev);
1009         dlm_release_lockspace(cinfo->lockspace, 2);
1010         kfree(cinfo);
1011         return 0;
1012 }
1013
1014 /* slot_number(): Returns the MD slot number to use
1015  * DLM starts the slot numbers from 1, wheras cluster-md
1016  * wants the number to be from zero, so we deduct one
1017  */
1018 static int slot_number(struct mddev *mddev)
1019 {
1020         struct md_cluster_info *cinfo = mddev->cluster_info;
1021
1022         return cinfo->slot_number - 1;
1023 }
1024
1025 /*
1026  * Check if the communication is already locked, else lock the communication
1027  * channel.
1028  * If it is already locked, token is in EX mode, and hence lock_token()
1029  * should not be called.
1030  */
1031 static int metadata_update_start(struct mddev *mddev)
1032 {
1033         struct md_cluster_info *cinfo = mddev->cluster_info;
1034         int ret;
1035
1036         /*
1037          * metadata_update_start is always called with the protection of
1038          * reconfig_mutex, so set WAITING_FOR_TOKEN here.
1039          */
1040         ret = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
1041                                     &cinfo->state);
1042         WARN_ON_ONCE(ret);
1043         md_wakeup_thread(mddev->thread);
1044
1045         wait_event(cinfo->wait,
1046                    !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state) ||
1047                    test_and_clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state));
1048
1049         /* If token is already locked, return 0 */
1050         if (cinfo->token_lockres->mode == DLM_LOCK_EX) {
1051                 clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1052                 return 0;
1053         }
1054
1055         ret = lock_token(cinfo, 1);
1056         clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1057         return ret;
1058 }
1059
1060 static int metadata_update_finish(struct mddev *mddev)
1061 {
1062         struct md_cluster_info *cinfo = mddev->cluster_info;
1063         struct cluster_msg cmsg;
1064         struct md_rdev *rdev;
1065         int ret = 0;
1066         int raid_slot = -1;
1067
1068         memset(&cmsg, 0, sizeof(cmsg));
1069         cmsg.type = cpu_to_le32(METADATA_UPDATED);
1070         /* Pick up a good active device number to send.
1071          */
1072         rdev_for_each(rdev, mddev)
1073                 if (rdev->raid_disk > -1 && !test_bit(Faulty, &rdev->flags)) {
1074                         raid_slot = rdev->desc_nr;
1075                         break;
1076                 }
1077         if (raid_slot >= 0) {
1078                 cmsg.raid_slot = cpu_to_le32(raid_slot);
1079                 ret = __sendmsg(cinfo, &cmsg);
1080         } else
1081                 pr_warn("md-cluster: No good device id found to send\n");
1082         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1083         unlock_comm(cinfo);
1084         return ret;
1085 }
1086
1087 static void metadata_update_cancel(struct mddev *mddev)
1088 {
1089         struct md_cluster_info *cinfo = mddev->cluster_info;
1090         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1091         unlock_comm(cinfo);
1092 }
1093
1094 /*
1095  * return 0 if all the bitmaps have the same sync_size
1096  */
1097 int cluster_check_sync_size(struct mddev *mddev)
1098 {
1099         int i, rv;
1100         bitmap_super_t *sb;
1101         unsigned long my_sync_size, sync_size = 0;
1102         int node_num = mddev->bitmap_info.nodes;
1103         int current_slot = md_cluster_ops->slot_number(mddev);
1104         struct bitmap *bitmap = mddev->bitmap;
1105         char str[64];
1106         struct dlm_lock_resource *bm_lockres;
1107
1108         sb = kmap_atomic(bitmap->storage.sb_page);
1109         my_sync_size = sb->sync_size;
1110         kunmap_atomic(sb);
1111
1112         for (i = 0; i < node_num; i++) {
1113                 if (i == current_slot)
1114                         continue;
1115
1116                 bitmap = get_bitmap_from_slot(mddev, i);
1117                 if (IS_ERR(bitmap)) {
1118                         pr_err("can't get bitmap from slot %d\n", i);
1119                         return -1;
1120                 }
1121
1122                 /*
1123                  * If we can hold the bitmap lock of one node then
1124                  * the slot is not occupied, update the sb.
1125                  */
1126                 snprintf(str, 64, "bitmap%04d", i);
1127                 bm_lockres = lockres_init(mddev, str, NULL, 1);
1128                 if (!bm_lockres) {
1129                         pr_err("md-cluster: Cannot initialize %s\n", str);
1130                         bitmap_free(bitmap);
1131                         return -1;
1132                 }
1133                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
1134                 rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
1135                 if (!rv)
1136                         bitmap_update_sb(bitmap);
1137                 lockres_free(bm_lockres);
1138
1139                 sb = kmap_atomic(bitmap->storage.sb_page);
1140                 if (sync_size == 0)
1141                         sync_size = sb->sync_size;
1142                 else if (sync_size != sb->sync_size) {
1143                         kunmap_atomic(sb);
1144                         bitmap_free(bitmap);
1145                         return -1;
1146                 }
1147                 kunmap_atomic(sb);
1148                 bitmap_free(bitmap);
1149         }
1150
1151         return (my_sync_size == sync_size) ? 0 : -1;
1152 }
1153
1154 /*
1155  * Update the size for cluster raid is a little more complex, we perform it
1156  * by the steps:
1157  * 1. hold token lock and update superblock in initiator node.
1158  * 2. send METADATA_UPDATED msg to other nodes.
1159  * 3. The initiator node continues to check each bitmap's sync_size, if all
1160  *    bitmaps have the same value of sync_size, then we can set capacity and
1161  *    let other nodes to perform it. If one node can't update sync_size
1162  *    accordingly, we need to revert to previous value.
1163  */
1164 static void update_size(struct mddev *mddev, sector_t old_dev_sectors)
1165 {
1166         struct md_cluster_info *cinfo = mddev->cluster_info;
1167         struct cluster_msg cmsg;
1168         struct md_rdev *rdev;
1169         int ret = 0;
1170         int raid_slot = -1;
1171
1172         md_update_sb(mddev, 1);
1173         lock_comm(cinfo, 1);
1174
1175         memset(&cmsg, 0, sizeof(cmsg));
1176         cmsg.type = cpu_to_le32(METADATA_UPDATED);
1177         rdev_for_each(rdev, mddev)
1178                 if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) {
1179                         raid_slot = rdev->desc_nr;
1180                         break;
1181                 }
1182         if (raid_slot >= 0) {
1183                 cmsg.raid_slot = cpu_to_le32(raid_slot);
1184                 /*
1185                  * We can only change capiticy after all the nodes can do it,
1186                  * so need to wait after other nodes already received the msg
1187                  * and handled the change
1188                  */
1189                 ret = __sendmsg(cinfo, &cmsg);
1190                 if (ret) {
1191                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1192                                __func__, __LINE__);
1193                         unlock_comm(cinfo);
1194                         return;
1195                 }
1196         } else {
1197                 pr_err("md-cluster: No good device id found to send\n");
1198                 unlock_comm(cinfo);
1199                 return;
1200         }
1201
1202         /*
1203          * check the sync_size from other node's bitmap, if sync_size
1204          * have already updated in other nodes as expected, send an
1205          * empty metadata msg to permit the change of capacity
1206          */
1207         if (cluster_check_sync_size(mddev) == 0) {
1208                 memset(&cmsg, 0, sizeof(cmsg));
1209                 cmsg.type = cpu_to_le32(CHANGE_CAPACITY);
1210                 ret = __sendmsg(cinfo, &cmsg);
1211                 if (ret)
1212                         pr_err("%s:%d: failed to send CHANGE_CAPACITY msg\n",
1213                                __func__, __LINE__);
1214                 set_capacity(mddev->gendisk, mddev->array_sectors);
1215                 revalidate_disk(mddev->gendisk);
1216         } else {
1217                 /* revert to previous sectors */
1218                 ret = mddev->pers->resize(mddev, old_dev_sectors);
1219                 if (!ret)
1220                         revalidate_disk(mddev->gendisk);
1221                 ret = __sendmsg(cinfo, &cmsg);
1222                 if (ret)
1223                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1224                                __func__, __LINE__);
1225         }
1226         unlock_comm(cinfo);
1227 }
1228
1229 static int resync_start(struct mddev *mddev)
1230 {
1231         struct md_cluster_info *cinfo = mddev->cluster_info;
1232         return dlm_lock_sync_interruptible(cinfo->resync_lockres, DLM_LOCK_EX, mddev);
1233 }
1234
1235 static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
1236 {
1237         struct md_cluster_info *cinfo = mddev->cluster_info;
1238         struct resync_info ri;
1239         struct cluster_msg cmsg = {0};
1240
1241         /* do not send zero again, if we have sent before */
1242         if (hi == 0) {
1243                 memcpy(&ri, cinfo->bitmap_lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
1244                 if (le64_to_cpu(ri.hi) == 0)
1245                         return 0;
1246         }
1247
1248         add_resync_info(cinfo->bitmap_lockres, lo, hi);
1249         /* Re-acquire the lock to refresh LVB */
1250         dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW);
1251         cmsg.type = cpu_to_le32(RESYNCING);
1252         cmsg.low = cpu_to_le64(lo);
1253         cmsg.high = cpu_to_le64(hi);
1254
1255         /*
1256          * mddev_lock is held if resync_info_update is called from
1257          * resync_finish (md_reap_sync_thread -> resync_finish)
1258          */
1259         if (lo == 0 && hi == 0)
1260                 return sendmsg(cinfo, &cmsg, 1);
1261         else
1262                 return sendmsg(cinfo, &cmsg, 0);
1263 }
1264
1265 static int resync_finish(struct mddev *mddev)
1266 {
1267         struct md_cluster_info *cinfo = mddev->cluster_info;
1268         dlm_unlock_sync(cinfo->resync_lockres);
1269         return resync_info_update(mddev, 0, 0);
1270 }
1271
1272 static int area_resyncing(struct mddev *mddev, int direction,
1273                 sector_t lo, sector_t hi)
1274 {
1275         struct md_cluster_info *cinfo = mddev->cluster_info;
1276         int ret = 0;
1277         struct suspend_info *s;
1278
1279         if ((direction == READ) &&
1280                 test_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state))
1281                 return 1;
1282
1283         spin_lock_irq(&cinfo->suspend_lock);
1284         if (list_empty(&cinfo->suspend_list))
1285                 goto out;
1286         list_for_each_entry(s, &cinfo->suspend_list, list)
1287                 if (hi > s->lo && lo < s->hi) {
1288                         ret = 1;
1289                         break;
1290                 }
1291 out:
1292         spin_unlock_irq(&cinfo->suspend_lock);
1293         return ret;
1294 }
1295
1296 /* add_new_disk() - initiates a disk add
1297  * However, if this fails before writing md_update_sb(),
1298  * add_new_disk_cancel() must be called to release token lock
1299  */
1300 static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
1301 {
1302         struct md_cluster_info *cinfo = mddev->cluster_info;
1303         struct cluster_msg cmsg;
1304         int ret = 0;
1305         struct mdp_superblock_1 *sb = page_address(rdev->sb_page);
1306         char *uuid = sb->device_uuid;
1307
1308         memset(&cmsg, 0, sizeof(cmsg));
1309         cmsg.type = cpu_to_le32(NEWDISK);
1310         memcpy(cmsg.uuid, uuid, 16);
1311         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1312         lock_comm(cinfo, 1);
1313         ret = __sendmsg(cinfo, &cmsg);
1314         if (ret) {
1315                 unlock_comm(cinfo);
1316                 return ret;
1317         }
1318         cinfo->no_new_dev_lockres->flags |= DLM_LKF_NOQUEUE;
1319         ret = dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_EX);
1320         cinfo->no_new_dev_lockres->flags &= ~DLM_LKF_NOQUEUE;
1321         /* Some node does not "see" the device */
1322         if (ret == -EAGAIN)
1323                 ret = -ENOENT;
1324         if (ret)
1325                 unlock_comm(cinfo);
1326         else {
1327                 dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
1328                 /* Since MD_CHANGE_DEVS will be set in add_bound_rdev which
1329                  * will run soon after add_new_disk, the below path will be
1330                  * invoked:
1331                  *   md_wakeup_thread(mddev->thread)
1332                  *      -> conf->thread (raid1d)
1333                  *      -> md_check_recovery -> md_update_sb
1334                  *      -> metadata_update_start/finish
1335                  * MD_CLUSTER_SEND_LOCKED_ALREADY will be cleared eventually.
1336                  *
1337                  * For other failure cases, metadata_update_cancel and
1338                  * add_new_disk_cancel also clear below bit as well.
1339                  * */
1340                 set_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1341                 wake_up(&cinfo->wait);
1342         }
1343         return ret;
1344 }
1345
1346 static void add_new_disk_cancel(struct mddev *mddev)
1347 {
1348         struct md_cluster_info *cinfo = mddev->cluster_info;
1349         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1350         unlock_comm(cinfo);
1351 }
1352
1353 static int new_disk_ack(struct mddev *mddev, bool ack)
1354 {
1355         struct md_cluster_info *cinfo = mddev->cluster_info;
1356
1357         if (!test_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state)) {
1358                 pr_warn("md-cluster(%s): Spurious cluster confirmation\n", mdname(mddev));
1359                 return -EINVAL;
1360         }
1361
1362         if (ack)
1363                 dlm_unlock_sync(cinfo->no_new_dev_lockres);
1364         complete(&cinfo->newdisk_completion);
1365         return 0;
1366 }
1367
1368 static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
1369 {
1370         struct cluster_msg cmsg = {0};
1371         struct md_cluster_info *cinfo = mddev->cluster_info;
1372         cmsg.type = cpu_to_le32(REMOVE);
1373         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1374         return sendmsg(cinfo, &cmsg, 1);
1375 }
1376
1377 static int lock_all_bitmaps(struct mddev *mddev)
1378 {
1379         int slot, my_slot, ret, held = 1, i = 0;
1380         char str[64];
1381         struct md_cluster_info *cinfo = mddev->cluster_info;
1382
1383         cinfo->other_bitmap_lockres = kzalloc((mddev->bitmap_info.nodes - 1) *
1384                                              sizeof(struct dlm_lock_resource *),
1385                                              GFP_KERNEL);
1386         if (!cinfo->other_bitmap_lockres) {
1387                 pr_err("md: can't alloc mem for other bitmap locks\n");
1388                 return 0;
1389         }
1390
1391         my_slot = slot_number(mddev);
1392         for (slot = 0; slot < mddev->bitmap_info.nodes; slot++) {
1393                 if (slot == my_slot)
1394                         continue;
1395
1396                 memset(str, '\0', 64);
1397                 snprintf(str, 64, "bitmap%04d", slot);
1398                 cinfo->other_bitmap_lockres[i] = lockres_init(mddev, str, NULL, 1);
1399                 if (!cinfo->other_bitmap_lockres[i])
1400                         return -ENOMEM;
1401
1402                 cinfo->other_bitmap_lockres[i]->flags |= DLM_LKF_NOQUEUE;
1403                 ret = dlm_lock_sync(cinfo->other_bitmap_lockres[i], DLM_LOCK_PW);
1404                 if (ret)
1405                         held = -1;
1406                 i++;
1407         }
1408
1409         return held;
1410 }
1411
1412 static void unlock_all_bitmaps(struct mddev *mddev)
1413 {
1414         struct md_cluster_info *cinfo = mddev->cluster_info;
1415         int i;
1416
1417         /* release other node's bitmap lock if they are existed */
1418         if (cinfo->other_bitmap_lockres) {
1419                 for (i = 0; i < mddev->bitmap_info.nodes - 1; i++) {
1420                         if (cinfo->other_bitmap_lockres[i]) {
1421                                 lockres_free(cinfo->other_bitmap_lockres[i]);
1422                         }
1423                 }
1424                 kfree(cinfo->other_bitmap_lockres);
1425         }
1426 }
1427
1428 static int gather_bitmaps(struct md_rdev *rdev)
1429 {
1430         int sn, err;
1431         sector_t lo, hi;
1432         struct cluster_msg cmsg = {0};
1433         struct mddev *mddev = rdev->mddev;
1434         struct md_cluster_info *cinfo = mddev->cluster_info;
1435
1436         cmsg.type = cpu_to_le32(RE_ADD);
1437         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1438         err = sendmsg(cinfo, &cmsg, 1);
1439         if (err)
1440                 goto out;
1441
1442         for (sn = 0; sn < mddev->bitmap_info.nodes; sn++) {
1443                 if (sn == (cinfo->slot_number - 1))
1444                         continue;
1445                 err = bitmap_copy_from_slot(mddev, sn, &lo, &hi, false);
1446                 if (err) {
1447                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", sn);
1448                         goto out;
1449                 }
1450                 if ((hi > 0) && (lo < mddev->recovery_cp))
1451                         mddev->recovery_cp = lo;
1452         }
1453 out:
1454         return err;
1455 }
1456
1457 static struct md_cluster_operations cluster_ops = {
1458         .join   = join,
1459         .leave  = leave,
1460         .slot_number = slot_number,
1461         .resync_start = resync_start,
1462         .resync_finish = resync_finish,
1463         .resync_info_update = resync_info_update,
1464         .metadata_update_start = metadata_update_start,
1465         .metadata_update_finish = metadata_update_finish,
1466         .metadata_update_cancel = metadata_update_cancel,
1467         .area_resyncing = area_resyncing,
1468         .add_new_disk = add_new_disk,
1469         .add_new_disk_cancel = add_new_disk_cancel,
1470         .new_disk_ack = new_disk_ack,
1471         .remove_disk = remove_disk,
1472         .load_bitmaps = load_bitmaps,
1473         .gather_bitmaps = gather_bitmaps,
1474         .lock_all_bitmaps = lock_all_bitmaps,
1475         .unlock_all_bitmaps = unlock_all_bitmaps,
1476         .update_size = update_size,
1477 };
1478
1479 static int __init cluster_init(void)
1480 {
1481         pr_warn("md-cluster: EXPERIMENTAL. Use with caution\n");
1482         pr_info("Registering Cluster MD functions\n");
1483         register_md_cluster_operations(&cluster_ops, THIS_MODULE);
1484         return 0;
1485 }
1486
1487 static void cluster_exit(void)
1488 {
1489         unregister_md_cluster_operations();
1490 }
1491
1492 module_init(cluster_init);
1493 module_exit(cluster_exit);
1494 MODULE_AUTHOR("SUSE");
1495 MODULE_LICENSE("GPL");
1496 MODULE_DESCRIPTION("Clustering support for MD");