drivers/md/md-cluster.c

   1 // SPDX-License-Identifier: GPL-2.0-or-later
   2 /*
   3  * Copyright (C) 2015, SUSE
   4  */
   5
   6
   7 #include <linux/module.h>
   8 #include <linux/kthread.h>
   9 #include <linux/dlm.h>
  10 #include <linux/sched.h>
  11 #include <linux/raid/md_p.h>
  12 #include "md.h"
  13 #include "md-bitmap.h"
  14 #include "md-cluster.h"
  15
  16 #define LVB_SIZE        64
  17 #define NEW_DEV_TIMEOUT 5000
  18
  19 struct dlm_lock_resource {
  20         dlm_lockspace_t *ls;
  21         struct dlm_lksb lksb;
  22         char *name; /* lock name. */
  23         uint32_t flags; /* flags to pass to dlm_lock() */
  24         wait_queue_head_t sync_locking; /* wait queue for synchronized locking */
  25         bool sync_locking_done;
  26         void (*bast)(void *arg, int mode); /* blocking AST function pointer*/
  27         struct mddev *mddev; /* pointing back to mddev. */
  28         int mode;
  29 };
  30
  31 struct resync_info {
  32         __le64 lo;
  33         __le64 hi;
  34 };
  35
  36 /* md_cluster_info flags */
  37 #define         MD_CLUSTER_WAITING_FOR_NEWDISK          1
  38 #define         MD_CLUSTER_SUSPEND_READ_BALANCING       2
  39 #define         MD_CLUSTER_BEGIN_JOIN_CLUSTER           3
  40
  41 /* Lock the send communication. This is done through
  42  * bit manipulation as opposed to a mutex in order to
  43  * accomodate lock and hold. See next comment.
  44  */
  45 #define         MD_CLUSTER_SEND_LOCK                    4
  46 /* If cluster operations (such as adding a disk) must lock the
  47  * communication channel, so as to perform extra operations
  48  * (update metadata) and no other operation is allowed on the
  49  * MD. Token needs to be locked and held until the operation
  50  * completes witha md_update_sb(), which would eventually release
  51  * the lock.
  52  */
  53 #define         MD_CLUSTER_SEND_LOCKED_ALREADY          5
  54 /* We should receive message after node joined cluster and
  55  * set up all the related infos such as bitmap and personality */
  56 #define         MD_CLUSTER_ALREADY_IN_CLUSTER           6
  57 #define         MD_CLUSTER_PENDING_RECV_EVENT           7
  58 #define         MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD              8
  59
  60 struct md_cluster_info {
  61         struct mddev *mddev; /* the md device which md_cluster_info belongs to */
  62         /* dlm lock space and resources for clustered raid. */
  63         dlm_lockspace_t *lockspace;
  64         int slot_number;
  65         struct completion completion;
  66         struct mutex recv_mutex;
  67         struct dlm_lock_resource *bitmap_lockres;
  68         struct dlm_lock_resource **other_bitmap_lockres;
  69         struct dlm_lock_resource *resync_lockres;
  70         struct list_head suspend_list;
  71
  72         spinlock_t suspend_lock;
  73         /* record the region which write should be suspended */
  74         sector_t suspend_lo;
  75         sector_t suspend_hi;
  76         int suspend_from; /* the slot which broadcast suspend_lo/hi */
  77
  78         struct md_thread *recovery_thread;
  79         unsigned long recovery_map;
  80         /* communication loc resources */
  81         struct dlm_lock_resource *ack_lockres;
  82         struct dlm_lock_resource *message_lockres;
  83         struct dlm_lock_resource *token_lockres;
  84         struct dlm_lock_resource *no_new_dev_lockres;
  85         struct md_thread *recv_thread;
  86         struct completion newdisk_completion;
  87         wait_queue_head_t wait;
  88         unsigned long state;
  89         /* record the region in RESYNCING message */
  90         sector_t sync_low;
  91         sector_t sync_hi;
  92 };
  93
  94 enum msg_type {
  95         METADATA_UPDATED = 0,
  96         RESYNCING,
  97         NEWDISK,
  98         REMOVE,
  99         RE_ADD,
 100         BITMAP_NEEDS_SYNC,
 101         CHANGE_CAPACITY,
 102         BITMAP_RESIZE,
 103 };
 104
 105 struct cluster_msg {
 106         __le32 type;
 107         __le32 slot;
 108         /* TODO: Unionize this for smaller footprint */
 109         __le64 low;
 110         __le64 high;
 111         char uuid[16];
 112         __le32 raid_slot;
 113 };
 114
 115 static void sync_ast(void *arg)
 116 {
 117         struct dlm_lock_resource *res;
 118
 119         res = arg;
 120         res->sync_locking_done = true;
 121         wake_up(&res->sync_locking);
 122 }
 123
 124 static int dlm_lock_sync(struct dlm_lock_resource *res, int mode)
 125 {
 126         int ret = 0;
 127
 128         ret = dlm_lock(res->ls, mode, &res->lksb,
 129                         res->flags, res->name, strlen(res->name),
 130                         0, sync_ast, res, res->bast);
 131         if (ret)
 132                 return ret;
 133         wait_event(res->sync_locking, res->sync_locking_done);
 134         res->sync_locking_done = false;
 135         if (res->lksb.sb_status == 0)
 136                 res->mode = mode;
 137         return res->lksb.sb_status;
 138 }
 139
 140 static int dlm_unlock_sync(struct dlm_lock_resource *res)
 141 {
 142         return dlm_lock_sync(res, DLM_LOCK_NL);
 143 }
 144
 145 /*
 146  * An variation of dlm_lock_sync, which make lock request could
 147  * be interrupted
 148  */
 149 static int dlm_lock_sync_interruptible(struct dlm_lock_resource *res, int mode,
 150                                        struct mddev *mddev)
 151 {
 152         int ret = 0;
 153
 154         ret = dlm_lock(res->ls, mode, &res->lksb,
 155                         res->flags, res->name, strlen(res->name),
 156                         0, sync_ast, res, res->bast);
 157         if (ret)
 158                 return ret;
 159
 160         wait_event(res->sync_locking, res->sync_locking_done
 161                                       || kthread_should_stop()
 162                                       || test_bit(MD_CLOSING, &mddev->flags));
 163         if (!res->sync_locking_done) {
 164                 /*
 165                  * the convert queue contains the lock request when request is
 166                  * interrupted, and sync_ast could still be run, so need to
 167                  * cancel the request and reset completion
 168                  */
 169                 ret = dlm_unlock(res->ls, res->lksb.sb_lkid, DLM_LKF_CANCEL,
 170                         &res->lksb, res);
 171                 res->sync_locking_done = false;
 172                 if (unlikely(ret != 0))
 173                         pr_info("failed to cancel previous lock request "
 174                                  "%s return %d\n", res->name, ret);
 175                 return -EPERM;
 176         } else
 177                 res->sync_locking_done = false;
 178         if (res->lksb.sb_status == 0)
 179                 res->mode = mode;
 180         return res->lksb.sb_status;
 181 }
 182
 183 static struct dlm_lock_resource *lockres_init(struct mddev *mddev,
 184                 char *name, void (*bastfn)(void *arg, int mode), int with_lvb)
 185 {
 186         struct dlm_lock_resource *res = NULL;
 187         int ret, namelen;
 188         struct md_cluster_info *cinfo = mddev->cluster_info;
 189
 190         res = kzalloc(sizeof(struct dlm_lock_resource), GFP_KERNEL);
 191         if (!res)
 192                 return NULL;
 193         init_waitqueue_head(&res->sync_locking);
 194         res->sync_locking_done = false;
 195         res->ls = cinfo->lockspace;
 196         res->mddev = mddev;
 197         res->mode = DLM_LOCK_IV;
 198         namelen = strlen(name);
 199         res->name = kzalloc(namelen + 1, GFP_KERNEL);
 200         if (!res->name) {
 201                 pr_err("md-cluster: Unable to allocate resource name for resource %s\n", name);
 202                 goto out_err;
 203         }
 204         strlcpy(res->name, name, namelen + 1);
 205         if (with_lvb) {
 206                 res->lksb.sb_lvbptr = kzalloc(LVB_SIZE, GFP_KERNEL);
 207                 if (!res->lksb.sb_lvbptr) {
 208                         pr_err("md-cluster: Unable to allocate LVB for resource %s\n", name);
 209                         goto out_err;
 210                 }
 211                 res->flags = DLM_LKF_VALBLK;
 212         }
 213
 214         if (bastfn)
 215                 res->bast = bastfn;
 216
 217         res->flags |= DLM_LKF_EXPEDITE;
 218
 219         ret = dlm_lock_sync(res, DLM_LOCK_NL);
 220         if (ret) {
 221                 pr_err("md-cluster: Unable to lock NL on new lock resource %s\n", name);
 222                 goto out_err;
 223         }
 224         res->flags &= ~DLM_LKF_EXPEDITE;
 225         res->flags |= DLM_LKF_CONVERT;
 226
 227         return res;
 228 out_err:
 229         kfree(res->lksb.sb_lvbptr);
 230         kfree(res->name);
 231         kfree(res);
 232         return NULL;
 233 }
 234
 235 static void lockres_free(struct dlm_lock_resource *res)
 236 {
 237         int ret = 0;
 238
 239         if (!res)
 240                 return;
 241
 242         /*
 243          * use FORCEUNLOCK flag, so we can unlock even the lock is on the
 244          * waiting or convert queue
 245          */
 246         ret = dlm_unlock(res->ls, res->lksb.sb_lkid, DLM_LKF_FORCEUNLOCK,
 247                 &res->lksb, res);
 248         if (unlikely(ret != 0))
 249                 pr_err("failed to unlock %s return %d\n", res->name, ret);
 250         else
 251                 wait_event(res->sync_locking, res->sync_locking_done);
 252
 253         kfree(res->name);
 254         kfree(res->lksb.sb_lvbptr);
 255         kfree(res);
 256 }
 257
 258 static void add_resync_info(struct dlm_lock_resource *lockres,
 259                             sector_t lo, sector_t hi)
 260 {
 261         struct resync_info *ri;
 262
 263         ri = (struct resync_info *)lockres->lksb.sb_lvbptr;
 264         ri->lo = cpu_to_le64(lo);
 265         ri->hi = cpu_to_le64(hi);
 266 }
 267
 268 static int read_resync_info(struct mddev *mddev,
 269                             struct dlm_lock_resource *lockres)
 270 {
 271         struct resync_info ri;
 272         struct md_cluster_info *cinfo = mddev->cluster_info;
 273         int ret = 0;
 274
 275         dlm_lock_sync(lockres, DLM_LOCK_CR);
 276         memcpy(&ri, lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
 277         if (le64_to_cpu(ri.hi) > 0) {
 278                 cinfo->suspend_hi = le64_to_cpu(ri.hi);
 279                 cinfo->suspend_lo = le64_to_cpu(ri.lo);
 280                 ret = 1;
 281         }
 282         dlm_unlock_sync(lockres);
 283         return ret;
 284 }
 285
 286 static void recover_bitmaps(struct md_thread *thread)
 287 {
 288         struct mddev *mddev = thread->mddev;
 289         struct md_cluster_info *cinfo = mddev->cluster_info;
 290         struct dlm_lock_resource *bm_lockres;
 291         char str[64];
 292         int slot, ret;
 293         sector_t lo, hi;
 294
 295         while (cinfo->recovery_map) {
 296                 slot = fls64((u64)cinfo->recovery_map) - 1;
 297
 298                 snprintf(str, 64, "bitmap%04d", slot);
 299                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 300                 if (!bm_lockres) {
 301                         pr_err("md-cluster: Cannot initialize bitmaps\n");
 302                         goto clear_bit;
 303                 }
 304
 305                 ret = dlm_lock_sync_interruptible(bm_lockres, DLM_LOCK_PW, mddev);
 306                 if (ret) {
 307                         pr_err("md-cluster: Could not DLM lock %s: %d\n",
 308                                         str, ret);
 309                         goto clear_bit;
 310                 }
 311                 ret = md_bitmap_copy_from_slot(mddev, slot, &lo, &hi, true);
 312                 if (ret) {
 313                         pr_err("md-cluster: Could not copy data from bitmap %d\n", slot);
 314                         goto clear_bit;
 315                 }
 316
 317                 /* Clear suspend_area associated with the bitmap */
 318                 spin_lock_irq(&cinfo->suspend_lock);
 319                 cinfo->suspend_hi = 0;
 320                 cinfo->suspend_lo = 0;
 321                 cinfo->suspend_from = -1;
 322                 spin_unlock_irq(&cinfo->suspend_lock);
 323
 324                 /* Kick off a reshape if needed */
 325                 if (test_bit(MD_RESYNCING_REMOTE, &mddev->recovery) &&
 326                     test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery) &&
 327                     mddev->reshape_position != MaxSector)
 328                         md_wakeup_thread(mddev->sync_thread);
 329
 330                 if (hi > 0) {
 331                         if (lo < mddev->recovery_cp)
 332                                 mddev->recovery_cp = lo;
 333                         /* wake up thread to continue resync in case resync
 334                          * is not finished */
 335                         if (mddev->recovery_cp != MaxSector) {
 336                                 /*
 337                                  * clear the REMOTE flag since we will launch
 338                                  * resync thread in current node.
 339                                  */
 340                                 clear_bit(MD_RESYNCING_REMOTE,
 341                                           &mddev->recovery);
 342                                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 343                                 md_wakeup_thread(mddev->thread);
 344                         }
 345                 }
 346 clear_bit:
 347                 lockres_free(bm_lockres);
 348                 clear_bit(slot, &cinfo->recovery_map);
 349         }
 350 }
 351
 352 static void recover_prep(void *arg)
 353 {
 354         struct mddev *mddev = arg;
 355         struct md_cluster_info *cinfo = mddev->cluster_info;
 356         set_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 357 }
 358
 359 static void __recover_slot(struct mddev *mddev, int slot)
 360 {
 361         struct md_cluster_info *cinfo = mddev->cluster_info;
 362
 363         set_bit(slot, &cinfo->recovery_map);
 364         if (!cinfo->recovery_thread) {
 365                 cinfo->recovery_thread = md_register_thread(recover_bitmaps,
 366                                 mddev, "recover");
 367                 if (!cinfo->recovery_thread) {
 368                         pr_warn("md-cluster: Could not create recovery thread\n");
 369                         return;
 370                 }
 371         }
 372         md_wakeup_thread(cinfo->recovery_thread);
 373 }
 374
 375 static void recover_slot(void *arg, struct dlm_slot *slot)
 376 {
 377         struct mddev *mddev = arg;
 378         struct md_cluster_info *cinfo = mddev->cluster_info;
 379
 380         pr_info("md-cluster: %s Node %d/%d down. My slot: %d. Initiating recovery.\n",
 381                         mddev->bitmap_info.cluster_name,
 382                         slot->nodeid, slot->slot,
 383                         cinfo->slot_number);
 384         /* deduct one since dlm slot starts from one while the num of
 385          * cluster-md begins with 0 */
 386         __recover_slot(mddev, slot->slot - 1);
 387 }
 388
 389 static void recover_done(void *arg, struct dlm_slot *slots,
 390                 int num_slots, int our_slot,
 391                 uint32_t generation)
 392 {
 393         struct mddev *mddev = arg;
 394         struct md_cluster_info *cinfo = mddev->cluster_info;
 395
 396         cinfo->slot_number = our_slot;
 397         /* completion is only need to be complete when node join cluster,
 398          * it doesn't need to run during another node's failure */
 399         if (test_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state)) {
 400                 complete(&cinfo->completion);
 401                 clear_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 402         }
 403         clear_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
 404 }
 405
 406 /* the ops is called when node join the cluster, and do lock recovery
 407  * if node failure occurs */
 408 static const struct dlm_lockspace_ops md_ls_ops = {
 409         .recover_prep = recover_prep,
 410         .recover_slot = recover_slot,
 411         .recover_done = recover_done,
 412 };
 413
 414 /*
 415  * The BAST function for the ack lock resource
 416  * This function wakes up the receive thread in
 417  * order to receive and process the message.
 418  */
 419 static void ack_bast(void *arg, int mode)
 420 {
 421         struct dlm_lock_resource *res = arg;
 422         struct md_cluster_info *cinfo = res->mddev->cluster_info;
 423
 424         if (mode == DLM_LOCK_EX) {
 425                 if (test_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state))
 426                         md_wakeup_thread(cinfo->recv_thread);
 427                 else
 428                         set_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state);
 429         }
 430 }
 431
 432 static void remove_suspend_info(struct mddev *mddev, int slot)
 433 {
 434         struct md_cluster_info *cinfo = mddev->cluster_info;
 435         mddev->pers->quiesce(mddev, 1);
 436         spin_lock_irq(&cinfo->suspend_lock);
 437         cinfo->suspend_hi = 0;
 438         cinfo->suspend_lo = 0;
 439         spin_unlock_irq(&cinfo->suspend_lock);
 440         mddev->pers->quiesce(mddev, 0);
 441 }
 442
 443 static void process_suspend_info(struct mddev *mddev,
 444                 int slot, sector_t lo, sector_t hi)
 445 {
 446         struct md_cluster_info *cinfo = mddev->cluster_info;
 447         struct mdp_superblock_1 *sb = NULL;
 448         struct md_rdev *rdev;
 449
 450         if (!hi) {
 451                 /*
 452                  * clear the REMOTE flag since resync or recovery is finished
 453                  * in remote node.
 454                  */
 455                 clear_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
 456                 remove_suspend_info(mddev, slot);
 457                 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 458                 md_wakeup_thread(mddev->thread);
 459                 return;
 460         }
 461
 462         rdev_for_each(rdev, mddev)
 463                 if (rdev->raid_disk > -1 && !test_bit(Faulty, &rdev->flags)) {
 464                         sb = page_address(rdev->sb_page);
 465                         break;
 466                 }
 467
 468         /*
 469          * The bitmaps are not same for different nodes
 470          * if RESYNCING is happening in one node, then
 471          * the node which received the RESYNCING message
 472          * probably will perform resync with the region
 473          * [lo, hi] again, so we could reduce resync time
 474          * a lot if we can ensure that the bitmaps among
 475          * different nodes are match up well.
 476          *
 477          * sync_low/hi is used to record the region which
 478          * arrived in the previous RESYNCING message,
 479          *
 480          * Call md_bitmap_sync_with_cluster to clear NEEDED_MASK
 481          * and set RESYNC_MASK since  resync thread is running
 482          * in another node, so we don't need to do the resync
 483          * again with the same section.
 484          *
 485          * Skip md_bitmap_sync_with_cluster in case reshape
 486          * happening, because reshaping region is small and
 487          * we don't want to trigger lots of WARN.
 488          */
 489         if (sb && !(le32_to_cpu(sb->feature_map) & MD_FEATURE_RESHAPE_ACTIVE))
 490                 md_bitmap_sync_with_cluster(mddev, cinfo->sync_low,
 491                                             cinfo->sync_hi, lo, hi);
 492         cinfo->sync_low = lo;
 493         cinfo->sync_hi = hi;
 494
 495         mddev->pers->quiesce(mddev, 1);
 496         spin_lock_irq(&cinfo->suspend_lock);
 497         cinfo->suspend_from = slot;
 498         cinfo->suspend_lo = lo;
 499         cinfo->suspend_hi = hi;
 500         spin_unlock_irq(&cinfo->suspend_lock);
 501         mddev->pers->quiesce(mddev, 0);
 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_and_notify(mddev->gendisk, mddev->array_sectors);
 585                 break;
 586         case RESYNCING:
 587                 set_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
 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         case BITMAP_RESIZE:
 605                 if (le64_to_cpu(msg->high) != mddev->pers->size(mddev, 0, 0))
 606                         ret = md_bitmap_resize(mddev->bitmap,
 607                                             le64_to_cpu(msg->high), 0, 0);
 608                 break;
 609         default:
 610                 ret = -1;
 611                 pr_warn("%s:%d Received unknown message from %d\n",
 612                         __func__, __LINE__, msg->slot);
 613         }
 614         return ret;
 615 }
 616
 617 /*
 618  * thread for receiving message
 619  */
 620 static void recv_daemon(struct md_thread *thread)
 621 {
 622         struct md_cluster_info *cinfo = thread->mddev->cluster_info;
 623         struct dlm_lock_resource *ack_lockres = cinfo->ack_lockres;
 624         struct dlm_lock_resource *message_lockres = cinfo->message_lockres;
 625         struct cluster_msg msg;
 626         int ret;
 627
 628         mutex_lock(&cinfo->recv_mutex);
 629         /*get CR on Message*/
 630         if (dlm_lock_sync(message_lockres, DLM_LOCK_CR)) {
 631                 pr_err("md/raid1:failed to get CR on MESSAGE\n");
 632                 mutex_unlock(&cinfo->recv_mutex);
 633                 return;
 634         }
 635
 636         /* read lvb and wake up thread to process this message_lockres */
 637         memcpy(&msg, message_lockres->lksb.sb_lvbptr, sizeof(struct cluster_msg));
 638         ret = process_recvd_msg(thread->mddev, &msg);
 639         if (ret)
 640                 goto out;
 641
 642         /*release CR on ack_lockres*/
 643         ret = dlm_unlock_sync(ack_lockres);
 644         if (unlikely(ret != 0))
 645                 pr_info("unlock ack failed return %d\n", ret);
 646         /*up-convert to PR on message_lockres*/
 647         ret = dlm_lock_sync(message_lockres, DLM_LOCK_PR);
 648         if (unlikely(ret != 0))
 649                 pr_info("lock PR on msg failed return %d\n", ret);
 650         /*get CR on ack_lockres again*/
 651         ret = dlm_lock_sync(ack_lockres, DLM_LOCK_CR);
 652         if (unlikely(ret != 0))
 653                 pr_info("lock CR on ack failed return %d\n", ret);
 654 out:
 655         /*release CR on message_lockres*/
 656         ret = dlm_unlock_sync(message_lockres);
 657         if (unlikely(ret != 0))
 658                 pr_info("unlock msg failed return %d\n", ret);
 659         mutex_unlock(&cinfo->recv_mutex);
 660 }
 661
 662 /* lock_token()
 663  * Takes the lock on the TOKEN lock resource so no other
 664  * node can communicate while the operation is underway.
 665  */
 666 static int lock_token(struct md_cluster_info *cinfo)
 667 {
 668         int error;
 669
 670         error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 671         if (error) {
 672                 pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
 673                                 __func__, __LINE__, error);
 674         } else {
 675                 /* Lock the receive sequence */
 676                 mutex_lock(&cinfo->recv_mutex);
 677         }
 678         return error;
 679 }
 680
 681 /* lock_comm()
 682  * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
 683  */
 684 static int lock_comm(struct md_cluster_info *cinfo, bool mddev_locked)
 685 {
 686         int rv, set_bit = 0;
 687         struct mddev *mddev = cinfo->mddev;
 688
 689         /*
 690          * If resync thread run after raid1d thread, then process_metadata_update
 691          * could not continue if raid1d held reconfig_mutex (and raid1d is blocked
 692          * since another node already got EX on Token and waitting the EX of Ack),
 693          * so let resync wake up thread in case flag is set.
 694          */
 695         if (mddev_locked && !test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 696                                       &cinfo->state)) {
 697                 rv = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
 698                                               &cinfo->state);
 699                 WARN_ON_ONCE(rv);
 700                 md_wakeup_thread(mddev->thread);
 701                 set_bit = 1;
 702         }
 703
 704         wait_event(cinfo->wait,
 705                    !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));
 706         rv = lock_token(cinfo);
 707         if (set_bit)
 708                 clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 709         return rv;
 710 }
 711
 712 static void unlock_comm(struct md_cluster_info *cinfo)
 713 {
 714         WARN_ON(cinfo->token_lockres->mode != DLM_LOCK_EX);
 715         mutex_unlock(&cinfo->recv_mutex);
 716         dlm_unlock_sync(cinfo->token_lockres);
 717         clear_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state);
 718         wake_up(&cinfo->wait);
 719 }
 720
 721 /* __sendmsg()
 722  * This function performs the actual sending of the message. This function is
 723  * usually called after performing the encompassing operation
 724  * The function:
 725  * 1. Grabs the message lockresource in EX mode
 726  * 2. Copies the message to the message LVB
 727  * 3. Downconverts message lockresource to CW
 728  * 4. Upconverts ack lock resource from CR to EX. This forces the BAST on other nodes
 729  *    and the other nodes read the message. The thread will wait here until all other
 730  *    nodes have released ack lock resource.
 731  * 5. Downconvert ack lockresource to CR
 732  */
 733 static int __sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
 734 {
 735         int error;
 736         int slot = cinfo->slot_number - 1;
 737
 738         cmsg->slot = cpu_to_le32(slot);
 739         /*get EX on Message*/
 740         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_EX);
 741         if (error) {
 742                 pr_err("md-cluster: failed to get EX on MESSAGE (%d)\n", error);
 743                 goto failed_message;
 744         }
 745
 746         memcpy(cinfo->message_lockres->lksb.sb_lvbptr, (void *)cmsg,
 747                         sizeof(struct cluster_msg));
 748         /*down-convert EX to CW on Message*/
 749         error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_CW);
 750         if (error) {
 751                 pr_err("md-cluster: failed to convert EX to CW on MESSAGE(%d)\n",
 752                                 error);
 753                 goto failed_ack;
 754         }
 755
 756         /*up-convert CR to EX on Ack*/
 757         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_EX);
 758         if (error) {
 759                 pr_err("md-cluster: failed to convert CR to EX on ACK(%d)\n",
 760                                 error);
 761                 goto failed_ack;
 762         }
 763
 764         /*down-convert EX to CR on Ack*/
 765         error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR);
 766         if (error) {
 767                 pr_err("md-cluster: failed to convert EX to CR on ACK(%d)\n",
 768                                 error);
 769                 goto failed_ack;
 770         }
 771
 772 failed_ack:
 773         error = dlm_unlock_sync(cinfo->message_lockres);
 774         if (unlikely(error != 0)) {
 775                 pr_err("md-cluster: failed convert to NL on MESSAGE(%d)\n",
 776                         error);
 777                 /* in case the message can't be released due to some reason */
 778                 goto failed_ack;
 779         }
 780 failed_message:
 781         return error;
 782 }
 783
 784 static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg,
 785                    bool mddev_locked)
 786 {
 787         int ret;
 788
 789         ret = lock_comm(cinfo, mddev_locked);
 790         if (!ret) {
 791                 ret = __sendmsg(cinfo, cmsg);
 792                 unlock_comm(cinfo);
 793         }
 794         return ret;
 795 }
 796
 797 static int gather_all_resync_info(struct mddev *mddev, int total_slots)
 798 {
 799         struct md_cluster_info *cinfo = mddev->cluster_info;
 800         int i, ret = 0;
 801         struct dlm_lock_resource *bm_lockres;
 802         char str[64];
 803         sector_t lo, hi;
 804
 805
 806         for (i = 0; i < total_slots; i++) {
 807                 memset(str, '\0', 64);
 808                 snprintf(str, 64, "bitmap%04d", i);
 809                 bm_lockres = lockres_init(mddev, str, NULL, 1);
 810                 if (!bm_lockres)
 811                         return -ENOMEM;
 812                 if (i == (cinfo->slot_number - 1)) {
 813                         lockres_free(bm_lockres);
 814                         continue;
 815                 }
 816
 817                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
 818                 ret = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
 819                 if (ret == -EAGAIN) {
 820                         if (read_resync_info(mddev, bm_lockres)) {
 821                                 pr_info("%s:%d Resync[%llu..%llu] in progress on %d\n",
 822                                                 __func__, __LINE__,
 823                                         (unsigned long long) cinfo->suspend_lo,
 824                                         (unsigned long long) cinfo->suspend_hi,
 825                                         i);
 826                                 cinfo->suspend_from = i;
 827                         }
 828                         ret = 0;
 829                         lockres_free(bm_lockres);
 830                         continue;
 831                 }
 832                 if (ret) {
 833                         lockres_free(bm_lockres);
 834                         goto out;
 835                 }
 836
 837                 /* Read the disk bitmap sb and check if it needs recovery */
 838                 ret = md_bitmap_copy_from_slot(mddev, i, &lo, &hi, false);
 839                 if (ret) {
 840                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", i);
 841                         lockres_free(bm_lockres);
 842                         continue;
 843                 }
 844                 if ((hi > 0) && (lo < mddev->recovery_cp)) {
 845                         set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 846                         mddev->recovery_cp = lo;
 847                         md_check_recovery(mddev);
 848                 }
 849
 850                 lockres_free(bm_lockres);
 851         }
 852 out:
 853         return ret;
 854 }
 855
 856 static int join(struct mddev *mddev, int nodes)
 857 {
 858         struct md_cluster_info *cinfo;
 859         int ret, ops_rv;
 860         char str[64];
 861
 862         cinfo = kzalloc(sizeof(struct md_cluster_info), GFP_KERNEL);
 863         if (!cinfo)
 864                 return -ENOMEM;
 865
 866         INIT_LIST_HEAD(&cinfo->suspend_list);
 867         spin_lock_init(&cinfo->suspend_lock);
 868         init_completion(&cinfo->completion);
 869         set_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
 870         init_waitqueue_head(&cinfo->wait);
 871         mutex_init(&cinfo->recv_mutex);
 872
 873         mddev->cluster_info = cinfo;
 874         cinfo->mddev = mddev;
 875
 876         memset(str, 0, 64);
 877         sprintf(str, "%pU", mddev->uuid);
 878         ret = dlm_new_lockspace(str, mddev->bitmap_info.cluster_name,
 879                                 DLM_LSFL_FS, LVB_SIZE,
 880                                 &md_ls_ops, mddev, &ops_rv, &cinfo->lockspace);
 881         if (ret)
 882                 goto err;
 883         wait_for_completion(&cinfo->completion);
 884         if (nodes < cinfo->slot_number) {
 885                 pr_err("md-cluster: Slot allotted(%d) is greater than available slots(%d).",
 886                         cinfo->slot_number, nodes);
 887                 ret = -ERANGE;
 888                 goto err;
 889         }
 890         /* Initiate the communication resources */
 891         ret = -ENOMEM;
 892         cinfo->recv_thread = md_register_thread(recv_daemon, mddev, "cluster_recv");
 893         if (!cinfo->recv_thread) {
 894                 pr_err("md-cluster: cannot allocate memory for recv_thread!\n");
 895                 goto err;
 896         }
 897         cinfo->message_lockres = lockres_init(mddev, "message", NULL, 1);
 898         if (!cinfo->message_lockres)
 899                 goto err;
 900         cinfo->token_lockres = lockres_init(mddev, "token", NULL, 0);
 901         if (!cinfo->token_lockres)
 902                 goto err;
 903         cinfo->no_new_dev_lockres = lockres_init(mddev, "no-new-dev", NULL, 0);
 904         if (!cinfo->no_new_dev_lockres)
 905                 goto err;
 906
 907         ret = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
 908         if (ret) {
 909                 ret = -EAGAIN;
 910                 pr_err("md-cluster: can't join cluster to avoid lock issue\n");
 911                 goto err;
 912         }
 913         cinfo->ack_lockres = lockres_init(mddev, "ack", ack_bast, 0);
 914         if (!cinfo->ack_lockres) {
 915                 ret = -ENOMEM;
 916                 goto err;
 917         }
 918         /* get sync CR lock on ACK. */
 919         if (dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR))
 920                 pr_err("md-cluster: failed to get a sync CR lock on ACK!(%d)\n",
 921                                 ret);
 922         dlm_unlock_sync(cinfo->token_lockres);
 923         /* get sync CR lock on no-new-dev. */
 924         if (dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR))
 925                 pr_err("md-cluster: failed to get a sync CR lock on no-new-dev!(%d)\n", ret);
 926
 927
 928         pr_info("md-cluster: Joined cluster %s slot %d\n", str, cinfo->slot_number);
 929         snprintf(str, 64, "bitmap%04d", cinfo->slot_number - 1);
 930         cinfo->bitmap_lockres = lockres_init(mddev, str, NULL, 1);
 931         if (!cinfo->bitmap_lockres) {
 932                 ret = -ENOMEM;
 933                 goto err;
 934         }
 935         if (dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW)) {
 936                 pr_err("Failed to get bitmap lock\n");
 937                 ret = -EINVAL;
 938                 goto err;
 939         }
 940
 941         cinfo->resync_lockres = lockres_init(mddev, "resync", NULL, 0);
 942         if (!cinfo->resync_lockres) {
 943                 ret = -ENOMEM;
 944                 goto err;
 945         }
 946
 947         return 0;
 948 err:
 949         set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 950         md_unregister_thread(&cinfo->recovery_thread);
 951         md_unregister_thread(&cinfo->recv_thread);
 952         lockres_free(cinfo->message_lockres);
 953         lockres_free(cinfo->token_lockres);
 954         lockres_free(cinfo->ack_lockres);
 955         lockres_free(cinfo->no_new_dev_lockres);
 956         lockres_free(cinfo->resync_lockres);
 957         lockres_free(cinfo->bitmap_lockres);
 958         if (cinfo->lockspace)
 959                 dlm_release_lockspace(cinfo->lockspace, 2);
 960         mddev->cluster_info = NULL;
 961         kfree(cinfo);
 962         return ret;
 963 }
 964
 965 static void load_bitmaps(struct mddev *mddev, int total_slots)
 966 {
 967         struct md_cluster_info *cinfo = mddev->cluster_info;
 968
 969         /* load all the node's bitmap info for resync */
 970         if (gather_all_resync_info(mddev, total_slots))
 971                 pr_err("md-cluster: failed to gather all resyn infos\n");
 972         set_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state);
 973         /* wake up recv thread in case something need to be handled */
 974         if (test_and_clear_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state))
 975                 md_wakeup_thread(cinfo->recv_thread);
 976 }
 977
 978 static void resync_bitmap(struct mddev *mddev)
 979 {
 980         struct md_cluster_info *cinfo = mddev->cluster_info;
 981         struct cluster_msg cmsg = {0};
 982         int err;
 983
 984         cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
 985         err = sendmsg(cinfo, &cmsg, 1);
 986         if (err)
 987                 pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
 988                         __func__, __LINE__, err);
 989 }
 990
 991 static void unlock_all_bitmaps(struct mddev *mddev);
 992 static int leave(struct mddev *mddev)
 993 {
 994         struct md_cluster_info *cinfo = mddev->cluster_info;
 995
 996         if (!cinfo)
 997                 return 0;
 998
 999         /*
1000          * BITMAP_NEEDS_SYNC message should be sent when node
1001          * is leaving the cluster with dirty bitmap, also we
1002          * can only deliver it when dlm connection is available.
1003          *
1004          * Also, we should send BITMAP_NEEDS_SYNC message in
1005          * case reshaping is interrupted.
1006          */
1007         if ((cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector) ||
1008             (mddev->reshape_position != MaxSector &&
1009              test_bit(MD_CLOSING, &mddev->flags)))
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);
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 static int update_bitmap_size(struct mddev *mddev, sector_t size)
1108 {
1109         struct md_cluster_info *cinfo = mddev->cluster_info;
1110         struct cluster_msg cmsg = {0};
1111         int ret;
1112
1113         cmsg.type = cpu_to_le32(BITMAP_RESIZE);
1114         cmsg.high = cpu_to_le64(size);
1115         ret = sendmsg(cinfo, &cmsg, 0);
1116         if (ret)
1117                 pr_err("%s:%d: failed to send BITMAP_RESIZE message (%d)\n",
1118                         __func__, __LINE__, ret);
1119         return ret;
1120 }
1121
1122 static int resize_bitmaps(struct mddev *mddev, sector_t newsize, sector_t oldsize)
1123 {
1124         struct bitmap_counts *counts;
1125         char str[64];
1126         struct dlm_lock_resource *bm_lockres;
1127         struct bitmap *bitmap = mddev->bitmap;
1128         unsigned long my_pages = bitmap->counts.pages;
1129         int i, rv;
1130
1131         /*
1132          * We need to ensure all the nodes can grow to a larger
1133          * bitmap size before make the reshaping.
1134          */
1135         rv = update_bitmap_size(mddev, newsize);
1136         if (rv)
1137                 return rv;
1138
1139         for (i = 0; i < mddev->bitmap_info.nodes; i++) {
1140                 if (i == md_cluster_ops->slot_number(mddev))
1141                         continue;
1142
1143                 bitmap = get_bitmap_from_slot(mddev, i);
1144                 if (IS_ERR(bitmap)) {
1145                         pr_err("can't get bitmap from slot %d\n", i);
1146                         bitmap = NULL;
1147                         goto out;
1148                 }
1149                 counts = &bitmap->counts;
1150
1151                 /*
1152                  * If we can hold the bitmap lock of one node then
1153                  * the slot is not occupied, update the pages.
1154                  */
1155                 snprintf(str, 64, "bitmap%04d", i);
1156                 bm_lockres = lockres_init(mddev, str, NULL, 1);
1157                 if (!bm_lockres) {
1158                         pr_err("Cannot initialize %s lock\n", str);
1159                         goto out;
1160                 }
1161                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
1162                 rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
1163                 if (!rv)
1164                         counts->pages = my_pages;
1165                 lockres_free(bm_lockres);
1166
1167                 if (my_pages != counts->pages)
1168                         /*
1169                          * Let's revert the bitmap size if one node
1170                          * can't resize bitmap
1171                          */
1172                         goto out;
1173                 md_bitmap_free(bitmap);
1174         }
1175
1176         return 0;
1177 out:
1178         md_bitmap_free(bitmap);
1179         update_bitmap_size(mddev, oldsize);
1180         return -1;
1181 }
1182
1183 /*
1184  * return 0 if all the bitmaps have the same sync_size
1185  */
1186 static int cluster_check_sync_size(struct mddev *mddev)
1187 {
1188         int i, rv;
1189         bitmap_super_t *sb;
1190         unsigned long my_sync_size, sync_size = 0;
1191         int node_num = mddev->bitmap_info.nodes;
1192         int current_slot = md_cluster_ops->slot_number(mddev);
1193         struct bitmap *bitmap = mddev->bitmap;
1194         char str[64];
1195         struct dlm_lock_resource *bm_lockres;
1196
1197         sb = kmap_atomic(bitmap->storage.sb_page);
1198         my_sync_size = sb->sync_size;
1199         kunmap_atomic(sb);
1200
1201         for (i = 0; i < node_num; i++) {
1202                 if (i == current_slot)
1203                         continue;
1204
1205                 bitmap = get_bitmap_from_slot(mddev, i);
1206                 if (IS_ERR(bitmap)) {
1207                         pr_err("can't get bitmap from slot %d\n", i);
1208                         return -1;
1209                 }
1210
1211                 /*
1212                  * If we can hold the bitmap lock of one node then
1213                  * the slot is not occupied, update the sb.
1214                  */
1215                 snprintf(str, 64, "bitmap%04d", i);
1216                 bm_lockres = lockres_init(mddev, str, NULL, 1);
1217                 if (!bm_lockres) {
1218                         pr_err("md-cluster: Cannot initialize %s\n", str);
1219                         md_bitmap_free(bitmap);
1220                         return -1;
1221                 }
1222                 bm_lockres->flags |= DLM_LKF_NOQUEUE;
1223                 rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
1224                 if (!rv)
1225                         md_bitmap_update_sb(bitmap);
1226                 lockres_free(bm_lockres);
1227
1228                 sb = kmap_atomic(bitmap->storage.sb_page);
1229                 if (sync_size == 0)
1230                         sync_size = sb->sync_size;
1231                 else if (sync_size != sb->sync_size) {
1232                         kunmap_atomic(sb);
1233                         md_bitmap_free(bitmap);
1234                         return -1;
1235                 }
1236                 kunmap_atomic(sb);
1237                 md_bitmap_free(bitmap);
1238         }
1239
1240         return (my_sync_size == sync_size) ? 0 : -1;
1241 }
1242
1243 /*
1244  * Update the size for cluster raid is a little more complex, we perform it
1245  * by the steps:
1246  * 1. hold token lock and update superblock in initiator node.
1247  * 2. send METADATA_UPDATED msg to other nodes.
1248  * 3. The initiator node continues to check each bitmap's sync_size, if all
1249  *    bitmaps have the same value of sync_size, then we can set capacity and
1250  *    let other nodes to perform it. If one node can't update sync_size
1251  *    accordingly, we need to revert to previous value.
1252  */
1253 static void update_size(struct mddev *mddev, sector_t old_dev_sectors)
1254 {
1255         struct md_cluster_info *cinfo = mddev->cluster_info;
1256         struct cluster_msg cmsg;
1257         struct md_rdev *rdev;
1258         int ret = 0;
1259         int raid_slot = -1;
1260
1261         md_update_sb(mddev, 1);
1262         if (lock_comm(cinfo, 1)) {
1263                 pr_err("%s: lock_comm failed\n", __func__);
1264                 return;
1265         }
1266
1267         memset(&cmsg, 0, sizeof(cmsg));
1268         cmsg.type = cpu_to_le32(METADATA_UPDATED);
1269         rdev_for_each(rdev, mddev)
1270                 if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) {
1271                         raid_slot = rdev->desc_nr;
1272                         break;
1273                 }
1274         if (raid_slot >= 0) {
1275                 cmsg.raid_slot = cpu_to_le32(raid_slot);
1276                 /*
1277                  * We can only change capiticy after all the nodes can do it,
1278                  * so need to wait after other nodes already received the msg
1279                  * and handled the change
1280                  */
1281                 ret = __sendmsg(cinfo, &cmsg);
1282                 if (ret) {
1283                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1284                                __func__, __LINE__);
1285                         unlock_comm(cinfo);
1286                         return;
1287                 }
1288         } else {
1289                 pr_err("md-cluster: No good device id found to send\n");
1290                 unlock_comm(cinfo);
1291                 return;
1292         }
1293
1294         /*
1295          * check the sync_size from other node's bitmap, if sync_size
1296          * have already updated in other nodes as expected, send an
1297          * empty metadata msg to permit the change of capacity
1298          */
1299         if (cluster_check_sync_size(mddev) == 0) {
1300                 memset(&cmsg, 0, sizeof(cmsg));
1301                 cmsg.type = cpu_to_le32(CHANGE_CAPACITY);
1302                 ret = __sendmsg(cinfo, &cmsg);
1303                 if (ret)
1304                         pr_err("%s:%d: failed to send CHANGE_CAPACITY msg\n",
1305                                __func__, __LINE__);
1306                 set_capacity_and_notify(mddev->gendisk, mddev->array_sectors);
1307         } else {
1308                 /* revert to previous sectors */
1309                 ret = mddev->pers->resize(mddev, old_dev_sectors);
1310                 ret = __sendmsg(cinfo, &cmsg);
1311                 if (ret)
1312                         pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
1313                                __func__, __LINE__);
1314         }
1315         unlock_comm(cinfo);
1316 }
1317
1318 static int resync_start(struct mddev *mddev)
1319 {
1320         struct md_cluster_info *cinfo = mddev->cluster_info;
1321         return dlm_lock_sync_interruptible(cinfo->resync_lockres, DLM_LOCK_EX, mddev);
1322 }
1323
1324 static void resync_info_get(struct mddev *mddev, sector_t *lo, sector_t *hi)
1325 {
1326         struct md_cluster_info *cinfo = mddev->cluster_info;
1327
1328         spin_lock_irq(&cinfo->suspend_lock);
1329         *lo = cinfo->suspend_lo;
1330         *hi = cinfo->suspend_hi;
1331         spin_unlock_irq(&cinfo->suspend_lock);
1332 }
1333
1334 static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
1335 {
1336         struct md_cluster_info *cinfo = mddev->cluster_info;
1337         struct resync_info ri;
1338         struct cluster_msg cmsg = {0};
1339
1340         /* do not send zero again, if we have sent before */
1341         if (hi == 0) {
1342                 memcpy(&ri, cinfo->bitmap_lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
1343                 if (le64_to_cpu(ri.hi) == 0)
1344                         return 0;
1345         }
1346
1347         add_resync_info(cinfo->bitmap_lockres, lo, hi);
1348         /* Re-acquire the lock to refresh LVB */
1349         dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW);
1350         cmsg.type = cpu_to_le32(RESYNCING);
1351         cmsg.low = cpu_to_le64(lo);
1352         cmsg.high = cpu_to_le64(hi);
1353
1354         /*
1355          * mddev_lock is held if resync_info_update is called from
1356          * resync_finish (md_reap_sync_thread -> resync_finish)
1357          */
1358         if (lo == 0 && hi == 0)
1359                 return sendmsg(cinfo, &cmsg, 1);
1360         else
1361                 return sendmsg(cinfo, &cmsg, 0);
1362 }
1363
1364 static int resync_finish(struct mddev *mddev)
1365 {
1366         struct md_cluster_info *cinfo = mddev->cluster_info;
1367         int ret = 0;
1368
1369         clear_bit(MD_RESYNCING_REMOTE, &mddev->recovery);
1370
1371         /*
1372          * If resync thread is interrupted so we can't say resync is finished,
1373          * another node will launch resync thread to continue.
1374          */
1375         if (!test_bit(MD_CLOSING, &mddev->flags))
1376                 ret = resync_info_update(mddev, 0, 0);
1377         dlm_unlock_sync(cinfo->resync_lockres);
1378         return ret;
1379 }
1380
1381 static int area_resyncing(struct mddev *mddev, int direction,
1382                 sector_t lo, sector_t hi)
1383 {
1384         struct md_cluster_info *cinfo = mddev->cluster_info;
1385         int ret = 0;
1386
1387         if ((direction == READ) &&
1388                 test_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state))
1389                 return 1;
1390
1391         spin_lock_irq(&cinfo->suspend_lock);
1392         if (hi > cinfo->suspend_lo && lo < cinfo->suspend_hi)
1393                 ret = 1;
1394         spin_unlock_irq(&cinfo->suspend_lock);
1395         return ret;
1396 }
1397
1398 /* add_new_disk() - initiates a disk add
1399  * However, if this fails before writing md_update_sb(),
1400  * add_new_disk_cancel() must be called to release token lock
1401  */
1402 static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
1403 {
1404         struct md_cluster_info *cinfo = mddev->cluster_info;
1405         struct cluster_msg cmsg;
1406         int ret = 0;
1407         struct mdp_superblock_1 *sb = page_address(rdev->sb_page);
1408         char *uuid = sb->device_uuid;
1409
1410         memset(&cmsg, 0, sizeof(cmsg));
1411         cmsg.type = cpu_to_le32(NEWDISK);
1412         memcpy(cmsg.uuid, uuid, 16);
1413         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1414         if (lock_comm(cinfo, 1))
1415                 return -EAGAIN;
1416         ret = __sendmsg(cinfo, &cmsg);
1417         if (ret) {
1418                 unlock_comm(cinfo);
1419                 return ret;
1420         }
1421         cinfo->no_new_dev_lockres->flags |= DLM_LKF_NOQUEUE;
1422         ret = dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_EX);
1423         cinfo->no_new_dev_lockres->flags &= ~DLM_LKF_NOQUEUE;
1424         /* Some node does not "see" the device */
1425         if (ret == -EAGAIN)
1426                 ret = -ENOENT;
1427         if (ret)
1428                 unlock_comm(cinfo);
1429         else {
1430                 dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
1431                 /* Since MD_CHANGE_DEVS will be set in add_bound_rdev which
1432                  * will run soon after add_new_disk, the below path will be
1433                  * invoked:
1434                  *   md_wakeup_thread(mddev->thread)
1435                  *      -> conf->thread (raid1d)
1436                  *      -> md_check_recovery -> md_update_sb
1437                  *      -> metadata_update_start/finish
1438                  * MD_CLUSTER_SEND_LOCKED_ALREADY will be cleared eventually.
1439                  *
1440                  * For other failure cases, metadata_update_cancel and
1441                  * add_new_disk_cancel also clear below bit as well.
1442                  * */
1443                 set_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1444                 wake_up(&cinfo->wait);
1445         }
1446         return ret;
1447 }
1448
1449 static void add_new_disk_cancel(struct mddev *mddev)
1450 {
1451         struct md_cluster_info *cinfo = mddev->cluster_info;
1452         clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1453         unlock_comm(cinfo);
1454 }
1455
1456 static int new_disk_ack(struct mddev *mddev, bool ack)
1457 {
1458         struct md_cluster_info *cinfo = mddev->cluster_info;
1459
1460         if (!test_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state)) {
1461                 pr_warn("md-cluster(%s): Spurious cluster confirmation\n", mdname(mddev));
1462                 return -EINVAL;
1463         }
1464
1465         if (ack)
1466                 dlm_unlock_sync(cinfo->no_new_dev_lockres);
1467         complete(&cinfo->newdisk_completion);
1468         return 0;
1469 }
1470
1471 static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
1472 {
1473         struct cluster_msg cmsg = {0};
1474         struct md_cluster_info *cinfo = mddev->cluster_info;
1475         cmsg.type = cpu_to_le32(REMOVE);
1476         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1477         return sendmsg(cinfo, &cmsg, 1);
1478 }
1479
1480 static int lock_all_bitmaps(struct mddev *mddev)
1481 {
1482         int slot, my_slot, ret, held = 1, i = 0;
1483         char str[64];
1484         struct md_cluster_info *cinfo = mddev->cluster_info;
1485
1486         cinfo->other_bitmap_lockres =
1487                 kcalloc(mddev->bitmap_info.nodes - 1,
1488                         sizeof(struct dlm_lock_resource *), GFP_KERNEL);
1489         if (!cinfo->other_bitmap_lockres) {
1490                 pr_err("md: can't alloc mem for other bitmap locks\n");
1491                 return 0;
1492         }
1493
1494         my_slot = slot_number(mddev);
1495         for (slot = 0; slot < mddev->bitmap_info.nodes; slot++) {
1496                 if (slot == my_slot)
1497                         continue;
1498
1499                 memset(str, '\0', 64);
1500                 snprintf(str, 64, "bitmap%04d", slot);
1501                 cinfo->other_bitmap_lockres[i] = lockres_init(mddev, str, NULL, 1);
1502                 if (!cinfo->other_bitmap_lockres[i])
1503                         return -ENOMEM;
1504
1505                 cinfo->other_bitmap_lockres[i]->flags |= DLM_LKF_NOQUEUE;
1506                 ret = dlm_lock_sync(cinfo->other_bitmap_lockres[i], DLM_LOCK_PW);
1507                 if (ret)
1508                         held = -1;
1509                 i++;
1510         }
1511
1512         return held;
1513 }
1514
1515 static void unlock_all_bitmaps(struct mddev *mddev)
1516 {
1517         struct md_cluster_info *cinfo = mddev->cluster_info;
1518         int i;
1519
1520         /* release other node's bitmap lock if they are existed */
1521         if (cinfo->other_bitmap_lockres) {
1522                 for (i = 0; i < mddev->bitmap_info.nodes - 1; i++) {
1523                         if (cinfo->other_bitmap_lockres[i]) {
1524                                 lockres_free(cinfo->other_bitmap_lockres[i]);
1525                         }
1526                 }
1527                 kfree(cinfo->other_bitmap_lockres);
1528                 cinfo->other_bitmap_lockres = NULL;
1529         }
1530 }
1531
1532 static int gather_bitmaps(struct md_rdev *rdev)
1533 {
1534         int sn, err;
1535         sector_t lo, hi;
1536         struct cluster_msg cmsg = {0};
1537         struct mddev *mddev = rdev->mddev;
1538         struct md_cluster_info *cinfo = mddev->cluster_info;
1539
1540         cmsg.type = cpu_to_le32(RE_ADD);
1541         cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1542         err = sendmsg(cinfo, &cmsg, 1);
1543         if (err)
1544                 goto out;
1545
1546         for (sn = 0; sn < mddev->bitmap_info.nodes; sn++) {
1547                 if (sn == (cinfo->slot_number - 1))
1548                         continue;
1549                 err = md_bitmap_copy_from_slot(mddev, sn, &lo, &hi, false);
1550                 if (err) {
1551                         pr_warn("md-cluster: Could not gather bitmaps from slot %d", sn);
1552                         goto out;
1553                 }
1554                 if ((hi > 0) && (lo < mddev->recovery_cp))
1555                         mddev->recovery_cp = lo;
1556         }
1557 out:
1558         return err;
1559 }
1560
1561 static struct md_cluster_operations cluster_ops = {
1562         .join   = join,
1563         .leave  = leave,
1564         .slot_number = slot_number,
1565         .resync_start = resync_start,
1566         .resync_finish = resync_finish,
1567         .resync_info_update = resync_info_update,
1568         .resync_info_get = resync_info_get,
1569         .metadata_update_start = metadata_update_start,
1570         .metadata_update_finish = metadata_update_finish,
1571         .metadata_update_cancel = metadata_update_cancel,
1572         .area_resyncing = area_resyncing,
1573         .add_new_disk = add_new_disk,
1574         .add_new_disk_cancel = add_new_disk_cancel,
1575         .new_disk_ack = new_disk_ack,
1576         .remove_disk = remove_disk,
1577         .load_bitmaps = load_bitmaps,
1578         .gather_bitmaps = gather_bitmaps,
1579         .resize_bitmaps = resize_bitmaps,
1580         .lock_all_bitmaps = lock_all_bitmaps,
1581         .unlock_all_bitmaps = unlock_all_bitmaps,
1582         .update_size = update_size,
1583 };
1584
1585 static int __init cluster_init(void)
1586 {
1587         pr_warn("md-cluster: support raid1 and raid10 (limited support)\n");
1588         pr_info("Registering Cluster MD functions\n");
1589         register_md_cluster_operations(&cluster_ops, THIS_MODULE);
1590         return 0;
1591 }
1592
1593 static void cluster_exit(void)
1594 {
1595         unregister_md_cluster_operations();
1596 }
1597
1598 module_init(cluster_init);
1599 module_exit(cluster_exit);
1600 MODULE_AUTHOR("SUSE");
1601 MODULE_LICENSE("GPL");
1602 MODULE_DESCRIPTION("Clustering support for MD");