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 "md-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                 snprintf(str, 64, "bitmap%04d", slot);
 308                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 309                 if (!bm_lockres) {
 310                         pr_err("md-cluster: Cannot initialize bitmaps\n");
 311                         goto clear_bit;
 312                 }
 313
 314                 ret = dlm_lock_sync_interruptible(bm_lockres, DLM_LOCK_PW, mddev);
 315                 if (ret) {
 316                         pr_err("md-cluster: Could not DLM lock %s: %d\n",
 317                                         str, ret);
 318                         goto clear_bit;
 319                 }
 320                 ret = bitmap_copy_from_slot(mddev, slot, &lo, &hi, true);
 321                 if (ret) {
 322                         pr_err("md-cluster: Could not copy data from bitmap %d\n", slot);
 323                         goto clear_bit;
 324                 }
 325
 326                 /* Clear suspend_area associated with the bitmap */
 327                 spin_lock_irq(&cinfo->suspend_lock);
 328                 list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
 329                         if (slot == s->slot) {
 330                                 list_del(&s->list);
 331                                 kfree(s);
 332                         }
 333                 spin_unlock_irq(&cinfo->suspend_lock);
 334
 335                 if (hi > 0) {
 336                         if (lo < mddev->recovery_cp)
 337                                 mddev->recovery_cp = lo;
 338                         /* wake up thread to continue resync in case resync
 339                          * is not finished */
 340                         if (mddev->recovery_cp != MaxSector) {
 341                                 /*
 342                                  * clear the REMOTE flag since we will launch
 343                                  * resync thread in current node.
 344                                  */
 345                                 clear_bit(MD_RESYNCING_REMOTE,
 346                                           &mddev->recovery);
 347                                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 348                                 md_wakeup_thread(mddev->thread);
 349                         }
 350                 }
 351 clear_bit:
 352                 lockres_free(bm_lockres);
 353                 clear_bit(slot, &cinfo->recovery_map);
 354         }
 355 }
 356
 357 static void recover_prep(void *arg)
 358 {
 359         struct mddev *mddev = arg;
 360         struct md_cluster_info *cinfo = mddev->cluster_info;
 361         set_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 362 }
 363
 364 static void __recover_slot(struct mddev *mddev, int slot)
 365 {
 366         struct md_cluster_info *cinfo = mddev->cluster_info;
 367
 368         set_bit(slot, &cinfo->recovery_map);
 369         if (!cinfo->recovery_thread) {
 370                 cinfo->recovery_thread = md_register_thread(recover_bitmaps,
 371                                 mddev, "recover");
 372                 if (!cinfo->recovery_thread) {
 373                         pr_warn("md-cluster: Could not create recovery thread\n");
 374                         return;
 375                 }
 376         }
 377         md_wakeup_thread(cinfo->recovery_thread);
 378 }
 379
 380 static void recover_slot(void *arg, struct dlm_slot *slot)
 381 {
 382         struct mddev *mddev = arg;
 383         struct md_cluster_info *cinfo = mddev->cluster_info;
 384
 385         pr_info("md-cluster: %s Node %d/%d down. My slot: %d. Initiating recovery.\n",
 386                         mddev->bitmap_info.cluster_name,
 387                         slot->nodeid, slot->slot,
 388                         cinfo->slot_number);
 389         /* deduct one since dlm slot starts from one while the num of
 390          * cluster-md begins with 0 */
 391         __recover_slot(mddev, slot->slot - 1);
 392 }
 393
 394 static void recover_done(void *arg, struct dlm_slot *slots,
 395                 int num_slots, int our_slot,
 396                 uint32_t generation)
 397 {
 398         struct mddev *mddev = arg;
 399         struct md_cluster_info *cinfo = mddev->cluster_info;
 400
 401         cinfo->slot_number = our_slot;
 402         /* completion is only need to be complete when node join cluster,
 403          * it doesn't need to run during another node's failure */
 404         if (test_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state)) {
 405                 complete(&cinfo->completion);
 406                 clear_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 407         }
 408         clear_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 409 }
 410
 411 /* the ops is called when node join the cluster, and do lock recovery
 412  * if node failure occurs */
 413 static const struct dlm_lockspace_ops md_ls_ops = {
 414         .recover_prep = recover_prep,
 415         .recover_slot = recover_slot,
 416         .recover_done = recover_done,
 417 };
 418
 419 /*
 420  * The BAST function for the ack lock resource
 421  * This function wakes up the receive thread in
 422  * order to receive and process the message.
 423  */
 424 static void ack_bast(void *arg, int mode)
 425 {
 426         struct dlm_lock_resource *res = arg;
 427         struct md_cluster_info *cinfo = res->mddev->cluster_info;
 428
 429         if (mode == DLM_LOCK_EX) {
 430                 if (test_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state))
 431                         md_wakeup_thread(cinfo->recv_thread);
 432                 else
 433                         set_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state);
 434         }
 435 }
 436
 437 static void __remove_suspend_info(struct md_cluster_info *cinfo, int slot)
 438 {
 439         struct suspend_info *s, *tmp;
 440
 441         list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
 442                 if (slot == s->slot) {
 443                         list_del(&s->list);
 444                         kfree(s);
 445                         break;
 446                 }
 447 }
 448
 449 static void remove_suspend_info(struct mddev *mddev, int slot)
 450 {
 451         struct md_cluster_info *cinfo = mddev->cluster_info;
 452         mddev->pers->quiesce(mddev, 1);
 453         spin_lock_irq(&cinfo->suspend_lock);
 454         __remove_suspend_info(cinfo, slot);
 455         spin_unlock_irq(&cinfo->suspend_lock);
 456         mddev->pers->quiesce(mddev, 0);
 457 }
 458
 459
 460 static void process_suspend_info(struct mddev *mddev,
 461                 int slot, sector_t lo, sector_t hi)
 462 {
 463         struct md_cluster_info *cinfo = mddev->cluster_info;
 464         struct suspend_info *s;
 465
 466         if (!hi) {
 467                 /*
 468                  * clear the REMOTE flag since resync or recovery is finished
 469                  * in remote node.
 470                  */
 471                 clear_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
 472                 remove_suspend_info(mddev, slot);
 473                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 474                 md_wakeup_thread(mddev->thread);
 475                 return;
 476         }
 477
 478         /*
 479          * The bitmaps are not same for different nodes
 480          * if RESYNCING is happening in one node, then
 481          * the node which received the RESYNCING message
 482          * probably will perform resync with the region
 483          * [lo, hi] again, so we could reduce resync time
 484          * a lot if we can ensure that the bitmaps among
 485          * different nodes are match up well.
 486          *
 487          * sync_low/hi is used to record the region which
 488          * arrived in the previous RESYNCING message,
 489          *
 490          * Call bitmap_sync_with_cluster to clear
 491          * NEEDED_MASK and set RESYNC_MASK since
 492          * resync thread is running in another node,
 493          * so we don't need to do the resync again
 494          * with the same section */
 495         bitmap_sync_with_cluster(mddev, cinfo->sync_low,
 496                                         cinfo->sync_hi,
 497                                         lo, hi);
 498         cinfo->sync_low = lo;
 499         cinfo->sync_hi = hi;
 500
 501         s = kzalloc(sizeof(struct suspend_info), GFP_KERNEL);
 502         if (!s)
 503                 return;
 504         s->slot = slot;
 505         s->lo = lo;
 506         s->hi = hi;
 507         mddev->pers->quiesce(mddev, 1);
 508         spin_lock_irq(&cinfo->suspend_lock);
 509         /* Remove existing entry (if exists) before adding */
 510         __remove_suspend_info(cinfo, slot);
 511         list_add(&s->list, &cinfo->suspend_list);
 512         spin_unlock_irq(&cinfo->suspend_lock);
 513         mddev->pers->quiesce(mddev, 0);
 514 }
 515
 516 static void process_add_new_disk(struct mddev *mddev, struct cluster_msg *cmsg)
 517 {
 518         char disk_uuid[64];
 519         struct md_cluster_info *cinfo = mddev->cluster_info;
 520         char event_name[] = "EVENT=ADD_DEVICE";
 521         char raid_slot[16];
 522         char *envp[] = {event_name, disk_uuid, raid_slot, NULL};
 523         int len;
 524
 525         len = snprintf(disk_uuid, 64, "DEVICE_UUID=");
 526         sprintf(disk_uuid + len, "%pU", cmsg->uuid);
 527         snprintf(raid_slot, 16, "RAID_DISK=%d", le32_to_cpu(cmsg->raid_slot));
 528         pr_info("%s:%d Sending kobject change with %s and %s\n", __func__, __LINE__, disk_uuid, raid_slot);
 529         init_completion(&cinfo->newdisk_completion);
 530         set_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
 531         kobject_uevent_env(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE, envp);
 532         wait_for_completion_timeout(&cinfo->newdisk_completion,
 533                         NEW_DEV_TIMEOUT);
 534         clear_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
 535 }
 536
 537
 538 static void process_metadata_update(struct mddev *mddev, struct cluster_msg *msg)
 539 {
 540         int got_lock = 0;
 541         struct md_cluster_info *cinfo = mddev->cluster_info;
 542         mddev->good_device_nr = le32_to_cpu(msg->raid_slot);
 543
 544         dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
 545         wait_event(mddev->thread->wqueue,
 546                    (got_lock = mddev_trylock(mddev)) ||
 547                     test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state));
 548         md_reload_sb(mddev, mddev->good_device_nr);
 549         if (got_lock)
 550                 mddev_unlock(mddev);
 551 }
 552
 553 static void process_remove_disk(struct mddev *mddev, struct cluster_msg *msg)
 554 {
 555         struct md_rdev *rdev;
 556
 557         rcu_read_lock();
 558         rdev = md_find_rdev_nr_rcu(mddev, le32_to_cpu(msg->raid_slot));
 559         if (rdev) {
 560                 set_bit(ClusterRemove, &rdev->flags);
 561                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 562                 md_wakeup_thread(mddev->thread);
 563         }
 564         else
 565                 pr_warn("%s: %d Could not find disk(%d) to REMOVE\n",
 566                         __func__, __LINE__, le32_to_cpu(msg->raid_slot));
 567         rcu_read_unlock();
 568 }
 569
 570 static void process_readd_disk(struct mddev *mddev, struct cluster_msg *msg)
 571 {
 572         struct md_rdev *rdev;
 573
 574         rcu_read_lock();
 575         rdev = md_find_rdev_nr_rcu(mddev, le32_to_cpu(msg->raid_slot));
 576         if (rdev && test_bit(Faulty, &rdev->flags))
 577                 clear_bit(Faulty, &rdev->flags);
 578         else
 579                 pr_warn("%s: %d Could not find disk(%d) which is faulty",
 580                         __func__, __LINE__, le32_to_cpu(msg->raid_slot));
 581         rcu_read_unlock();
 582 }
 583
 584 static int process_recvd_msg(struct mddev *mddev, struct cluster_msg *msg)
 585 {
 586         int ret = 0;
 587
 588         if (WARN(mddev->cluster_info->slot_number - 1 == le32_to_cpu(msg->slot),
 589                 "node %d received it's own msg\n", le32_to_cpu(msg->slot)))
 590                 return -1;
 591         switch (le32_to_cpu(msg->type)) {
 592         case METADATA_UPDATED:
 593                 process_metadata_update(mddev, msg);
 594                 break;
 595         case CHANGE_CAPACITY:
 596                 set_capacity(mddev->gendisk, mddev->array_sectors);
 597                 revalidate_disk(mddev->gendisk);
 598                 break;
 599         case RESYNCING:
 600                 set_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
 601                 process_suspend_info(mddev, le32_to_cpu(msg->slot),
 602                                      le64_to_cpu(msg->low),
 603                                      le64_to_cpu(msg->high));
 604                 break;
 605         case NEWDISK:
 606                 process_add_new_disk(mddev, msg);
 607                 break;
 608         case REMOVE:
 609                 process_remove_disk(mddev, msg);
 610                 break;
 611         case RE_ADD:
 612                 process_readd_disk(mddev, msg);
 613                 break;
 614         case BITMAP_NEEDS_SYNC:
 615                 __recover_slot(mddev, le32_to_cpu(msg->slot));
 616                 break;
 617         default:
 618                 ret = -1;
 619                 pr_warn("%s:%d Received unknown message from %d\n",
 620                         __func__, __LINE__, msg->slot);
 621         }
 622         return ret;
 623 }
 624
 625 /*
 626  * thread for receiving message
 627  */
 628 static void recv_daemon(struct md_thread *thread)
 629 {
 630         struct md_cluster_info *cinfo = thread->mddev->cluster_info;
 631         struct dlm_lock_resource *ack_lockres = cinfo->ack_lockres;
 632         struct dlm_lock_resource *message_lockres = cinfo->message_lockres;
 633         struct cluster_msg msg;
 634         int ret;
 635
 636         mutex_lock(&cinfo->recv_mutex);
 637         /*get CR on Message*/
 638         if (dlm_lock_sync(message_lockres, DLM_LOCK_CR)) {
 639                 pr_err("md/raid1:failed to get CR on MESSAGE\n");
 640                 mutex_unlock(&cinfo->recv_mutex);
 641                 return;
 642         }
 643
 644         /* read lvb and wake up thread to process this message_lockres */
 645         memcpy(&msg, message_lockres->lksb.sb_lvbptr, sizeof(struct cluster_msg));
 646         ret = process_recvd_msg(thread->mddev, &msg);
 647         if (ret)
 648                 goto out;
 649
 650         /*release CR on ack_lockres*/
 651         ret = dlm_unlock_sync(ack_lockres);
 652         if (unlikely(ret != 0))
 653                 pr_info("unlock ack failed return %d\n", ret);
 654         /*up-convert to PR on message_lockres*/
 655         ret = dlm_lock_sync(message_lockres, DLM_LOCK_PR);
 656         if (unlikely(ret != 0))
 657                 pr_info("lock PR on msg failed return %d\n", ret);
 658         /*get CR on ack_lockres again*/
 659         ret = dlm_lock_sync(ack_lockres, DLM_LOCK_CR);
 660         if (unlikely(ret != 0))
 661                 pr_info("lock CR on ack failed return %d\n", ret);
 662 out:
 663         /*release CR on message_lockres*/
 664         ret = dlm_unlock_sync(message_lockres);
 665         if (unlikely(ret != 0))
 666                 pr_info("unlock msg failed return %d\n", ret);
 667         mutex_unlock(&cinfo->recv_mutex);
 668 }
 669
 670 /* lock_token()
 671  * Takes the lock on the TOKEN lock resource so no other
 672  * node can communicate while the operation is underway.
 673  */
 674 static int lock_token(struct md_cluster_info *cinfo, bool mddev_locked)
 675 {
 676         int error, set_bit = 0;
 677         struct mddev *mddev = cinfo->mddev;
 678
 679         /*
 680          * If resync thread run after raid1d thread, then process_metadata_update
 681          * could not continue if raid1d held reconfig_mutex (and raid1d is blocked
 682          * since another node already got EX on Token and waitting the EX of Ack),
 683          * so let resync wake up thread in case flag is set.
 684          */
 685         if (mddev_locked && !test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 686                                       &cinfo->state)) {
 687                 error = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 688                                               &cinfo->state);
 689                 WARN_ON_ONCE(error);
 690                 md_wakeup_thread(mddev->thread);
 691                 set_bit = 1;
 692         }
 693         error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 694         if (set_bit)
 695                 clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 696
 697         if (error)
 698                 pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
 699                                 __func__, __LINE__, error);
 700
 701         /* Lock the receive sequence */
 702         mutex_lock(&cinfo->recv_mutex);
 703         return error;
 704 }
 705
 706 /* lock_comm()
 707  * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
 708  */
 709 static int lock_comm(struct md_cluster_info *cinfo, bool mddev_locked)
 710 {
 711         wait_event(cinfo->wait,
 712                    !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));
 713
 714         return lock_token(cinfo, mddev_locked);
 715 }
 716
 717 static void unlock_comm(struct md_cluster_info *cinfo)
 718 {
 719         WARN_ON(cinfo->token_lockres->mode != DLM_LOCK_EX);
 720         mutex_unlock(&cinfo->recv_mutex);
 721         dlm_unlock_sync(cinfo->token_lockres);
 722         clear_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state);
 723         wake_up(&cinfo->wait);
 724 }
 725
 726 /* __sendmsg()
 727  * This function performs the actual sending of the message. This function is
 728  * usually called after performing the encompassing operation
 729  * The function:
 730  * 1. Grabs the message lockresource in EX mode
 731  * 2. Copies the message to the message LVB
 732  * 3. Downconverts message lockresource to CW
 733  * 4. Upconverts ack lock resource from CR to EX. This forces the BAST on other nodes
 734  *    and the other nodes read the message. The thread will wait here until all other
 735  *    nodes have released ack lock resource.
 736  * 5. Downconvert ack lockresource to CR
 737  */
 738 static int __sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
 739 {
 740         int error;
 741         int slot = cinfo->slot_number - 1;
 742
 743         cmsg->slot = cpu_to_le32(slot);
 744         /*get EX on Message*/
 745         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_EX);
 746         if (error) {
 747                 pr_err("md-cluster: failed to get EX on MESSAGE (%d)\n", error);
 748                 goto failed_message;
 749         }
 750
 751         memcpy(cinfo->message_lockres->lksb.sb_lvbptr, (void *)cmsg,
 752                         sizeof(struct cluster_msg));
 753         /*down-convert EX to CW on Message*/
 754         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_CW);
 755         if (error) {
 756                 pr_err("md-cluster: failed to convert EX to CW on MESSAGE(%d)\n",
 757                                 error);
 758                 goto failed_ack;
 759         }
 760
 761         /*up-convert CR to EX on Ack*/
 762         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_EX);
 763         if (error) {
 764                 pr_err("md-cluster: failed to convert CR to EX on ACK(%d)\n",
 765                                 error);
 766                 goto failed_ack;
 767         }
 768
 769         /*down-convert EX to CR on Ack*/
 770         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR);
 771         if (error) {
 772                 pr_err("md-cluster: failed to convert EX to CR on ACK(%d)\n",
 773                                 error);
 774                 goto failed_ack;
 775         }
 776
 777 failed_ack:
 778         error = dlm_unlock_sync(cinfo->message_lockres);
 779         if (unlikely(error != 0)) {
 780                 pr_err("md-cluster: failed convert to NL on MESSAGE(%d)\n",
 781                         error);
 782                 /* in case the message can't be released due to some reason */
 783                 goto failed_ack;
 784         }
 785 failed_message:
 786         return error;
 787 }
 788
 789 static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg,
 790                    bool mddev_locked)
 791 {
 792         int ret;
 793
 794         lock_comm(cinfo, mddev_locked);
 795         ret = __sendmsg(cinfo, cmsg);
 796         unlock_comm(cinfo);
 797         return ret;
 798 }
 799
 800 static int gather_all_resync_info(struct mddev *mddev, int total_slots)
 801 {
 802         struct md_cluster_info *cinfo = mddev->cluster_info;
 803         int i, ret = 0;
 804         struct dlm_lock_resource *bm_lockres;
 805         struct suspend_info *s;
 806         char str[64];
 807         sector_t lo, hi;
 808
 809
 810         for (i = 0; i < total_slots; i++) {
 811                 memset(str, '\0', 64);
 812                 snprintf(str, 64, "bitmap%04d", i);
 813                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 814                 if (!bm_lockres)
 815                         return -ENOMEM;
 816                 if (i == (cinfo->slot_number - 1)) {
 817                         lockres_free(bm_lockres);
 818                         continue;
 819                 }
 820
 821                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
 822                 ret = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
 823                 if (ret == -EAGAIN) {
 824                         s = read_resync_info(mddev, bm_lockres);
 825                         if (s) {
 826                                 pr_info("%s:%d Resync[%llu..%llu] in progress on %d\n",
 827                                                 __func__, __LINE__,
 828                                                 (unsigned long long) s->lo,
 829                                                 (unsigned long long) s->hi, i);
 830                                 spin_lock_irq(&cinfo->suspend_lock);
 831                                 s->slot = i;
 832                                 list_add(&s->list, &cinfo->suspend_list);
 833                                 spin_unlock_irq(&cinfo->suspend_lock);
 834                         }
 835                         ret = 0;
 836                         lockres_free(bm_lockres);
 837                         continue;
 838                 }
 839                 if (ret) {
 840                         lockres_free(bm_lockres);
 841                         goto out;
 842                 }
 843
 844                 /* Read the disk bitmap sb and check if it needs recovery */
 845                 ret = bitmap_copy_from_slot(mddev, i, &lo, &hi, false);
 846                 if (ret) {
 847                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", i);
 848                         lockres_free(bm_lockres);
 849                         continue;
 850                 }
 851                 if ((hi > 0) && (lo < mddev->recovery_cp)) {
 852                         set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 853                         mddev->recovery_cp = lo;
 854                         md_check_recovery(mddev);
 855                 }
 856
 857                 lockres_free(bm_lockres);
 858         }
 859 out:
 860         return ret;
 861 }
 862
 863 static int join(struct mddev *mddev, int nodes)
 864 {
 865         struct md_cluster_info *cinfo;
 866         int ret, ops_rv;
 867         char str[64];
 868
 869         cinfo = kzalloc(sizeof(struct md_cluster_info), GFP_KERNEL);
 870         if (!cinfo)
 871                 return -ENOMEM;
 872
 873         INIT_LIST_HEAD(&cinfo->suspend_list);
 874         spin_lock_init(&cinfo->suspend_lock);
 875         init_completion(&cinfo->completion);
 876         set_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 877         init_waitqueue_head(&cinfo->wait);
 878         mutex_init(&cinfo->recv_mutex);
 879
 880         mddev->cluster_info = cinfo;
 881         cinfo->mddev = mddev;
 882
 883         memset(str, 0, 64);
 884         sprintf(str, "%pU", mddev->uuid);
 885         ret = dlm_new_lockspace(str, mddev->bitmap_info.cluster_name,
 886                                 DLM_LSFL_FS, LVB_SIZE,
 887                                 &md_ls_ops, mddev, &ops_rv, &cinfo->lockspace);
 888         if (ret)
 889                 goto err;
 890         wait_for_completion(&cinfo->completion);
 891         if (nodes < cinfo->slot_number) {
 892                 pr_err("md-cluster: Slot allotted(%d) is greater than available slots(%d).",
 893                         cinfo->slot_number, nodes);
 894                 ret = -ERANGE;
 895                 goto err;
 896         }
 897         /* Initiate the communication resources */
 898         ret = -ENOMEM;
 899         cinfo->recv_thread = md_register_thread(recv_daemon, mddev, "cluster_recv");
 900         if (!cinfo->recv_thread) {
 901                 pr_err("md-cluster: cannot allocate memory for recv_thread!\n");
 902                 goto err;
 903         }
 904         cinfo->message_lockres = lockres_init(mddev, "message", NULL, 1);
 905         if (!cinfo->message_lockres)
 906                 goto err;
 907         cinfo->token_lockres = lockres_init(mddev, "token", NULL, 0);
 908         if (!cinfo->token_lockres)
 909                 goto err;
 910         cinfo->no_new_dev_lockres = lockres_init(mddev, "no-new-dev", NULL, 0);
 911         if (!cinfo->no_new_dev_lockres)
 912                 goto err;
 913
 914         ret = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 915         if (ret) {
 916                 ret = -EAGAIN;
 917                 pr_err("md-cluster: can't join cluster to avoid lock issue\n");
 918                 goto err;
 919         }
 920         cinfo->ack_lockres = lockres_init(mddev, "ack", ack_bast, 0);
 921         if (!cinfo->ack_lockres) {
 922                 ret = -ENOMEM;
 923                 goto err;
 924         }
 925         /* get sync CR lock on ACK. */
 926         if (dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR))
 927                 pr_err("md-cluster: failed to get a sync CR lock on ACK!(%d)\n",
 928                                 ret);
 929         dlm_unlock_sync(cinfo->token_lockres);
 930         /* get sync CR lock on no-new-dev. */
 931         if (dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR))
 932                 pr_err("md-cluster: failed to get a sync CR lock on no-new-dev!(%d)\n", ret);
 933
 934
 935         pr_info("md-cluster: Joined cluster %s slot %d\n", str, cinfo->slot_number);
 936         snprintf(str, 64, "bitmap%04d", cinfo->slot_number - 1);
 937         cinfo->bitmap_lockres = lockres_init(mddev, str, NULL, 1);
 938         if (!cinfo->bitmap_lockres) {
 939                 ret = -ENOMEM;
 940                 goto err;
 941         }
 942         if (dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW)) {
 943                 pr_err("Failed to get bitmap lock\n");
 944                 ret = -EINVAL;
 945                 goto err;
 946         }
 947
 948         cinfo->resync_lockres = lockres_init(mddev, "resync", NULL, 0);
 949         if (!cinfo->resync_lockres) {
 950                 ret = -ENOMEM;
 951                 goto err;
 952         }
 953
 954         return 0;
 955 err:
 956         set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 957         md_unregister_thread(&cinfo->recovery_thread);
 958         md_unregister_thread(&cinfo->recv_thread);
 959         lockres_free(cinfo->message_lockres);
 960         lockres_free(cinfo->token_lockres);
 961         lockres_free(cinfo->ack_lockres);
 962         lockres_free(cinfo->no_new_dev_lockres);
 963         lockres_free(cinfo->resync_lockres);
 964         lockres_free(cinfo->bitmap_lockres);
 965         if (cinfo->lockspace)
 966                 dlm_release_lockspace(cinfo->lockspace, 2);
 967         mddev->cluster_info = NULL;
 968         kfree(cinfo);
 969         return ret;
 970 }
 971
 972 static void load_bitmaps(struct mddev *mddev, int total_slots)
 973 {
 974         struct md_cluster_info *cinfo = mddev->cluster_info;
 975
 976         /* load all the node's bitmap info for resync */
 977         if (gather_all_resync_info(mddev, total_slots))
 978                 pr_err("md-cluster: failed to gather all resyn infos\n");
 979         set_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state);
 980         /* wake up recv thread in case something need to be handled */
 981         if (test_and_clear_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state))
 982                 md_wakeup_thread(cinfo->recv_thread);
 983 }
 984
 985 static void resync_bitmap(struct mddev *mddev)
 986 {
 987         struct md_cluster_info *cinfo = mddev->cluster_info;
 988         struct cluster_msg cmsg = {0};
 989         int err;
 990
 991         cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
 992         err = sendmsg(cinfo, &cmsg, 1);
 993         if (err)
 994                 pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
 995                         __func__, __LINE__, err);
 996 }
 997
 998 static void unlock_all_bitmaps(struct mddev *mddev);
 999 static int leave(struct mddev *mddev)
1000 {
1001         struct md_cluster_info *cinfo = mddev->cluster_info;
1002
1003         if (!cinfo)
1004                 return 0;
1005
1006         /* BITMAP_NEEDS_SYNC message should be sent when node
1007          * is leaving the cluster with dirty bitmap, also we
1008          * can only deliver it when dlm connection is available */
1009         if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector)
1010                 resync_bitmap(mddev);
1011
1012         set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1013         md_unregister_thread(&cinfo->recovery_thread);
1014         md_unregister_thread(&cinfo->recv_thread);
1015         lockres_free(cinfo->message_lockres);
1016         lockres_free(cinfo->token_lockres);
1017         lockres_free(cinfo->ack_lockres);
1018         lockres_free(cinfo->no_new_dev_lockres);
1019         lockres_free(cinfo->resync_lockres);
1020         lockres_free(cinfo->bitmap_lockres);
1021         unlock_all_bitmaps(mddev);
1022         dlm_release_lockspace(cinfo->lockspace, 2);
1023         kfree(cinfo);
1024         return 0;
1025 }
1026
1027 /* slot_number(): Returns the MD slot number to use
1028  * DLM starts the slot numbers from 1, wheras cluster-md
1029  * wants the number to be from zero, so we deduct one
1030  */
1031 static int slot_number(struct mddev *mddev)
1032 {
1033         struct md_cluster_info *cinfo = mddev->cluster_info;
1034
1035         return cinfo->slot_number - 1;
1036 }
1037
1038 /*
1039  * Check if the communication is already locked, else lock the communication
1040  * channel.
1041  * If it is already locked, token is in EX mode, and hence lock_token()
1042  * should not be called.
1043  */
1044 static int metadata_update_start(struct mddev *mddev)
1045 {
1046         struct md_cluster_info *cinfo = mddev->cluster_info;
1047         int ret;
1048
1049         /*
1050          * metadata_update_start is always called with the protection of
1051          * reconfig_mutex, so set WAITING_FOR_TOKEN here.
1052          */
1053         ret = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
1054                                     &cinfo->state);
1055         WARN_ON_ONCE(ret);
1056         md_wakeup_thread(mddev->thread);
1057
1058         wait_event(cinfo->wait,
1059                    !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state) ||
1060                    test_and_clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state));
1061
1062         /* If token is already locked, return 0 */
1063         if (cinfo->token_lockres->mode == DLM_LOCK_EX) {
1064                 clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1065                 return 0;
1066         }
1067
1068         ret = lock_token(cinfo, 1);
1069         clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
1070         return ret;
1071 }
1072
1073 static int metadata_update_finish(struct mddev *mddev)
1074 {
1075         struct md_cluster_info *cinfo = mddev->cluster_info;
1076         struct cluster_msg cmsg;
1077         struct md_rdev *rdev;
1078         int ret = 0;
1079         int raid_slot = -1;
1080
1081         memset(&cmsg, 0, sizeof(cmsg));
1082         cmsg.type = cpu_to_le32(METADATA_UPDATED);
1083         /* Pick up a good active device number to send.
1084          */
1085         rdev_for_each(rdev, mddev)
1086                 if (rdev->raid_disk > -1 && !test_bit(Faulty, &rdev->flags)) {
1087                         raid_slot = rdev->desc_nr;
1088                         break;
1089                 }
1090         if (raid_slot >= 0) {
1091                 cmsg.raid_slot = cpu_to_le32(raid_slot);
1092                 ret = __sendmsg(cinfo, &cmsg);
1093         } else
1094                 pr_warn("md-cluster: No good device id found to send\n");
1095         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1096         unlock_comm(cinfo);
1097         return ret;
1098 }
1099
1100 static void metadata_update_cancel(struct mddev *mddev)
1101 {
1102         struct md_cluster_info *cinfo = mddev->cluster_info;
1103         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1104         unlock_comm(cinfo);
1105 }
1106
1107 /*
1108  * return 0 if all the bitmaps have the same sync_size
1109  */
1110 static int cluster_check_sync_size(struct mddev *mddev)
1111 {
1112         int i, rv;
1113         bitmap_super_t *sb;
1114         unsigned long my_sync_size, sync_size = 0;
1115         int node_num = mddev->bitmap_info.nodes;
1116         int current_slot = md_cluster_ops->slot_number(mddev);
1117         struct bitmap *bitmap = mddev->bitmap;
1118         char str[64];
1119         struct dlm_lock_resource *bm_lockres;
1120
1121         sb = kmap_atomic(bitmap->storage.sb_page);
1122         my_sync_size = sb->sync_size;
1123         kunmap_atomic(sb);
1124
1125         for (i = 0; i < node_num; i++) {
1126                 if (i == current_slot)
1127                         continue;
1128
1129                 bitmap = get_bitmap_from_slot(mddev, i);
1130                 if (IS_ERR(bitmap)) {
1131                         pr_err("can't get bitmap from slot %d\n", i);
1132                         return -1;
1133                 }
1134
1135                 /*
1136                  * If we can hold the bitmap lock of one node then
1137                  * the slot is not occupied, update the sb.
1138                  */
1139                 snprintf(str, 64, "bitmap%04d", i);
1140                 bm_lockres = lockres_init(mddev, str, NULL, 1);
1141                 if (!bm_lockres) {
1142                         pr_err("md-cluster: Cannot initialize %s\n", str);
1143                         bitmap_free(bitmap);
1144                         return -1;
1145                 }
1146                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
1147                 rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
1148                 if (!rv)
1149                         bitmap_update_sb(bitmap);
1150                 lockres_free(bm_lockres);
1151
1152                 sb = kmap_atomic(bitmap->storage.sb_page);
1153                 if (sync_size == 0)
1154                         sync_size = sb->sync_size;
1155                 else if (sync_size != sb->sync_size) {
1156                         kunmap_atomic(sb);
1157                         bitmap_free(bitmap);
1158                         return -1;
1159                 }
1160                 kunmap_atomic(sb);
1161                 bitmap_free(bitmap);
1162         }
1163
1164         return (my_sync_size == sync_size) ? 0 : -1;
1165 }
1166
1167 /*
1168  * Update the size for cluster raid is a little more complex, we perform it
1169  * by the steps:
1170  * 1. hold token lock and update superblock in initiator node.
1171  * 2. send METADATA_UPDATED msg to other nodes.
1172  * 3. The initiator node continues to check each bitmap's sync_size, if all
1173  *    bitmaps have the same value of sync_size, then we can set capacity and
1174  *    let other nodes to perform it. If one node can't update sync_size
1175  *    accordingly, we need to revert to previous value.
1176  */
1177 static void update_size(struct mddev *mddev, sector_t old_dev_sectors)
1178 {
1179         struct md_cluster_info *cinfo = mddev->cluster_info;
1180         struct cluster_msg cmsg;
1181         struct md_rdev *rdev;
1182         int ret = 0;
1183         int raid_slot = -1;
1184
1185         md_update_sb(mddev, 1);
1186         lock_comm(cinfo, 1);
1187
1188         memset(&cmsg, 0, sizeof(cmsg));
1189         cmsg.type = cpu_to_le32(METADATA_UPDATED);
1190         rdev_for_each(rdev, mddev)
1191                 if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) {
1192                         raid_slot = rdev->desc_nr;
1193                         break;
1194                 }
1195         if (raid_slot >= 0) {
1196                 cmsg.raid_slot = cpu_to_le32(raid_slot);
1197                 /*
1198                  * We can only change capiticy after all the nodes can do it,
1199                  * so need to wait after other nodes already received the msg
1200                  * and handled the change
1201                  */
1202                 ret = __sendmsg(cinfo, &cmsg);
1203                 if (ret) {
1204                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1205                                __func__, __LINE__);
1206                         unlock_comm(cinfo);
1207                         return;
1208                 }
1209         } else {
1210                 pr_err("md-cluster: No good device id found to send\n");
1211                 unlock_comm(cinfo);
1212                 return;
1213         }
1214
1215         /*
1216          * check the sync_size from other node's bitmap, if sync_size
1217          * have already updated in other nodes as expected, send an
1218          * empty metadata msg to permit the change of capacity
1219          */
1220         if (cluster_check_sync_size(mddev) == 0) {
1221                 memset(&cmsg, 0, sizeof(cmsg));
1222                 cmsg.type = cpu_to_le32(CHANGE_CAPACITY);
1223                 ret = __sendmsg(cinfo, &cmsg);
1224                 if (ret)
1225                         pr_err("%s:%d: failed to send CHANGE_CAPACITY msg\n",
1226                                __func__, __LINE__);
1227                 set_capacity(mddev->gendisk, mddev->array_sectors);
1228                 revalidate_disk(mddev->gendisk);
1229         } else {
1230                 /* revert to previous sectors */
1231                 ret = mddev->pers->resize(mddev, old_dev_sectors);
1232                 if (!ret)
1233                         revalidate_disk(mddev->gendisk);
1234                 ret = __sendmsg(cinfo, &cmsg);
1235                 if (ret)
1236                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1237                                __func__, __LINE__);
1238         }
1239         unlock_comm(cinfo);
1240 }
1241
1242 static int resync_start(struct mddev *mddev)
1243 {
1244         struct md_cluster_info *cinfo = mddev->cluster_info;
1245         return dlm_lock_sync_interruptible(cinfo->resync_lockres, DLM_LOCK_EX, mddev);
1246 }
1247
1248 static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
1249 {
1250         struct md_cluster_info *cinfo = mddev->cluster_info;
1251         struct resync_info ri;
1252         struct cluster_msg cmsg = {0};
1253
1254         /* do not send zero again, if we have sent before */
1255         if (hi == 0) {
1256                 memcpy(&ri, cinfo->bitmap_lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
1257                 if (le64_to_cpu(ri.hi) == 0)
1258                         return 0;
1259         }
1260
1261         add_resync_info(cinfo->bitmap_lockres, lo, hi);
1262         /* Re-acquire the lock to refresh LVB */
1263         dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW);
1264         cmsg.type = cpu_to_le32(RESYNCING);
1265         cmsg.low = cpu_to_le64(lo);
1266         cmsg.high = cpu_to_le64(hi);
1267
1268         /*
1269          * mddev_lock is held if resync_info_update is called from
1270          * resync_finish (md_reap_sync_thread -> resync_finish)
1271          */
1272         if (lo == 0 && hi == 0)
1273                 return sendmsg(cinfo, &cmsg, 1);
1274         else
1275                 return sendmsg(cinfo, &cmsg, 0);
1276 }
1277
1278 static int resync_finish(struct mddev *mddev)
1279 {
1280         struct md_cluster_info *cinfo = mddev->cluster_info;
1281
1282         clear_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
1283         dlm_unlock_sync(cinfo->resync_lockres);
1284
1285         /*
1286          * If resync thread is interrupted so we can't say resync is finished,
1287          * another node will launch resync thread to continue.
1288          */
1289         if (test_bit(MD_CLOSING, &mddev->flags))
1290                 return 0;
1291         else
1292                 return resync_info_update(mddev, 0, 0);
1293 }
1294
1295 static int area_resyncing(struct mddev *mddev, int direction,
1296                 sector_t lo, sector_t hi)
1297 {
1298         struct md_cluster_info *cinfo = mddev->cluster_info;
1299         int ret = 0;
1300         struct suspend_info *s;
1301
1302         if ((direction == READ) &&
1303                 test_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state))
1304                 return 1;
1305
1306         spin_lock_irq(&cinfo->suspend_lock);
1307         if (list_empty(&cinfo->suspend_list))
1308                 goto out;
1309         list_for_each_entry(s, &cinfo->suspend_list, list)
1310                 if (hi > s->lo && lo < s->hi) {
1311                         ret = 1;
1312                         break;
1313                 }
1314 out:
1315         spin_unlock_irq(&cinfo->suspend_lock);
1316         return ret;
1317 }
1318
1319 /* add_new_disk() - initiates a disk add
1320  * However, if this fails before writing md_update_sb(),
1321  * add_new_disk_cancel() must be called to release token lock
1322  */
1323 static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
1324 {
1325         struct md_cluster_info *cinfo = mddev->cluster_info;
1326         struct cluster_msg cmsg;
1327         int ret = 0;
1328         struct mdp_superblock_1 *sb = page_address(rdev->sb_page);
1329         char *uuid = sb->device_uuid;
1330
1331         memset(&cmsg, 0, sizeof(cmsg));
1332         cmsg.type = cpu_to_le32(NEWDISK);
1333         memcpy(cmsg.uuid, uuid, 16);
1334         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1335         lock_comm(cinfo, 1);
1336         ret = __sendmsg(cinfo, &cmsg);
1337         if (ret) {
1338                 unlock_comm(cinfo);
1339                 return ret;
1340         }
1341         cinfo->no_new_dev_lockres->flags |= DLM_LKF_NOQUEUE;
1342         ret = dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_EX);
1343         cinfo->no_new_dev_lockres->flags &= ~DLM_LKF_NOQUEUE;
1344         /* Some node does not "see" the device */
1345         if (ret == -EAGAIN)
1346                 ret = -ENOENT;
1347         if (ret)
1348                 unlock_comm(cinfo);
1349         else {
1350                 dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
1351                 /* Since MD_CHANGE_DEVS will be set in add_bound_rdev which
1352                  * will run soon after add_new_disk, the below path will be
1353                  * invoked:
1354                  *   md_wakeup_thread(mddev->thread)
1355                  *      -> conf->thread (raid1d)
1356                  *      -> md_check_recovery -> md_update_sb
1357                  *      -> metadata_update_start/finish
1358                  * MD_CLUSTER_SEND_LOCKED_ALREADY will be cleared eventually.
1359                  *
1360                  * For other failure cases, metadata_update_cancel and
1361                  * add_new_disk_cancel also clear below bit as well.
1362                  * */
1363                 set_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1364                 wake_up(&cinfo->wait);
1365         }
1366         return ret;
1367 }
1368
1369 static void add_new_disk_cancel(struct mddev *mddev)
1370 {
1371         struct md_cluster_info *cinfo = mddev->cluster_info;
1372         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1373         unlock_comm(cinfo);
1374 }
1375
1376 static int new_disk_ack(struct mddev *mddev, bool ack)
1377 {
1378         struct md_cluster_info *cinfo = mddev->cluster_info;
1379
1380         if (!test_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state)) {
1381                 pr_warn("md-cluster(%s): Spurious cluster confirmation\n", mdname(mddev));
1382                 return -EINVAL;
1383         }
1384
1385         if (ack)
1386                 dlm_unlock_sync(cinfo->no_new_dev_lockres);
1387         complete(&cinfo->newdisk_completion);
1388         return 0;
1389 }
1390
1391 static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
1392 {
1393         struct cluster_msg cmsg = {0};
1394         struct md_cluster_info *cinfo = mddev->cluster_info;
1395         cmsg.type = cpu_to_le32(REMOVE);
1396         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1397         return sendmsg(cinfo, &cmsg, 1);
1398 }
1399
1400 static int lock_all_bitmaps(struct mddev *mddev)
1401 {
1402         int slot, my_slot, ret, held = 1, i = 0;
1403         char str[64];
1404         struct md_cluster_info *cinfo = mddev->cluster_info;
1405
1406         cinfo->other_bitmap_lockres =
1407                 kcalloc(mddev->bitmap_info.nodes - 1,
1408                         sizeof(struct dlm_lock_resource *), GFP_KERNEL);
1409         if (!cinfo->other_bitmap_lockres) {
1410                 pr_err("md: can't alloc mem for other bitmap locks\n");
1411                 return 0;
1412         }
1413
1414         my_slot = slot_number(mddev);
1415         for (slot = 0; slot < mddev->bitmap_info.nodes; slot++) {
1416                 if (slot == my_slot)
1417                         continue;
1418
1419                 memset(str, '\0', 64);
1420                 snprintf(str, 64, "bitmap%04d", slot);
1421                 cinfo->other_bitmap_lockres[i] = lockres_init(mddev, str, NULL, 1);
1422                 if (!cinfo->other_bitmap_lockres[i])
1423                         return -ENOMEM;
1424
1425                 cinfo->other_bitmap_lockres[i]->flags |= DLM_LKF_NOQUEUE;
1426                 ret = dlm_lock_sync(cinfo->other_bitmap_lockres[i], DLM_LOCK_PW);
1427                 if (ret)
1428                         held = -1;
1429                 i++;
1430         }
1431
1432         return held;
1433 }
1434
1435 static void unlock_all_bitmaps(struct mddev *mddev)
1436 {
1437         struct md_cluster_info *cinfo = mddev->cluster_info;
1438         int i;
1439
1440         /* release other node's bitmap lock if they are existed */
1441         if (cinfo->other_bitmap_lockres) {
1442                 for (i = 0; i < mddev->bitmap_info.nodes - 1; i++) {
1443                         if (cinfo->other_bitmap_lockres[i]) {
1444                                 lockres_free(cinfo->other_bitmap_lockres[i]);
1445                         }
1446                 }
1447                 kfree(cinfo->other_bitmap_lockres);
1448         }
1449 }
1450
1451 static int gather_bitmaps(struct md_rdev *rdev)
1452 {
1453         int sn, err;
1454         sector_t lo, hi;
1455         struct cluster_msg cmsg = {0};
1456         struct mddev *mddev = rdev->mddev;
1457         struct md_cluster_info *cinfo = mddev->cluster_info;
1458
1459         cmsg.type = cpu_to_le32(RE_ADD);
1460         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1461         err = sendmsg(cinfo, &cmsg, 1);
1462         if (err)
1463                 goto out;
1464
1465         for (sn = 0; sn < mddev->bitmap_info.nodes; sn++) {
1466                 if (sn == (cinfo->slot_number - 1))
1467                         continue;
1468                 err = bitmap_copy_from_slot(mddev, sn, &lo, &hi, false);
1469                 if (err) {
1470                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", sn);
1471                         goto out;
1472                 }
1473                 if ((hi > 0) && (lo < mddev->recovery_cp))
1474                         mddev->recovery_cp = lo;
1475         }
1476 out:
1477         return err;
1478 }
1479
1480 static struct md_cluster_operations cluster_ops = {
1481         .join   = join,
1482         .leave  = leave,
1483         .slot_number = slot_number,
1484         .resync_start = resync_start,
1485         .resync_finish = resync_finish,
1486         .resync_info_update = resync_info_update,
1487         .metadata_update_start = metadata_update_start,
1488         .metadata_update_finish = metadata_update_finish,
1489         .metadata_update_cancel = metadata_update_cancel,
1490         .area_resyncing = area_resyncing,
1491         .add_new_disk = add_new_disk,
1492         .add_new_disk_cancel = add_new_disk_cancel,
1493         .new_disk_ack = new_disk_ack,
1494         .remove_disk = remove_disk,
1495         .load_bitmaps = load_bitmaps,
1496         .gather_bitmaps = gather_bitmaps,
1497         .lock_all_bitmaps = lock_all_bitmaps,
1498         .unlock_all_bitmaps = unlock_all_bitmaps,
1499         .update_size = update_size,
1500 };
1501
1502 static int __init cluster_init(void)
1503 {
1504         pr_warn("md-cluster: support raid1 and raid10 (limited support)\n");
1505         pr_info("Registering Cluster MD functions\n");
1506         register_md_cluster_operations(&cluster_ops, THIS_MODULE);
1507         return 0;
1508 }
1509
1510 static void cluster_exit(void)
1511 {
1512         unregister_md_cluster_operations();
1513 }
1514
1515 module_init(cluster_init);
1516 module_exit(cluster_exit);
1517 MODULE_AUTHOR("SUSE");
1518 MODULE_LICENSE("GPL");
1519 MODULE_DESCRIPTION("Clustering support for MD");