Pileus Git - ~andy/linux/blob - fs/dlm/recoverd.c

   1 /******************************************************************************
   2 *******************************************************************************
   3 **
   4 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
   5 **  Copyright (C) 2004-2011 Red Hat, Inc.  All rights reserved.
   6 **
   7 **  This copyrighted material is made available to anyone wishing to use,
   8 **  modify, copy, or redistribute it subject to the terms and conditions
   9 **  of the GNU General Public License v.2.
  10 **
  11 *******************************************************************************
  12 ******************************************************************************/
  13
  14 #include "dlm_internal.h"
  15 #include "lockspace.h"
  16 #include "member.h"
  17 #include "dir.h"
  18 #include "ast.h"
  19 #include "recover.h"
  20 #include "lowcomms.h"
  21 #include "lock.h"
  22 #include "requestqueue.h"
  23 #include "recoverd.h"
  24
  25
  26 /* If the start for which we're re-enabling locking (seq) has been superseded
  27    by a newer stop (ls_recover_seq), we need to leave locking disabled.
  28
  29    We suspend dlm_recv threads here to avoid the race where dlm_recv a) sees
  30    locking stopped and b) adds a message to the requestqueue, but dlm_recoverd
  31    enables locking and clears the requestqueue between a and b. */
  32
  33 static int enable_locking(struct dlm_ls *ls, uint64_t seq)
  34 {
  35         int error = -EINTR;
  36
  37         down_write(&ls->ls_recv_active);
  38
  39         spin_lock(&ls->ls_recover_lock);
  40         if (ls->ls_recover_seq == seq) {
  41                 set_bit(LSFL_RUNNING, &ls->ls_flags);
  42                 /* unblocks processes waiting to enter the dlm */
  43                 up_write(&ls->ls_in_recovery);
  44                 error = 0;
  45         }
  46         spin_unlock(&ls->ls_recover_lock);
  47
  48         up_write(&ls->ls_recv_active);
  49         return error;
  50 }
  51
  52 static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
  53 {
  54         unsigned long start;
  55         int error, neg = 0;
  56
  57         log_debug(ls, "dlm_recover %llu", (unsigned long long)rv->seq);
  58
  59         mutex_lock(&ls->ls_recoverd_active);
  60
  61         dlm_callback_suspend(ls);
  62
  63         /*
  64          * Free non-master tossed rsb's.  Master rsb's are kept on toss
  65          * list and put on root list to be included in resdir recovery.
  66          */
  67
  68         dlm_clear_toss_list(ls);
  69
  70         /*
  71          * This list of root rsb's will be the basis of most of the recovery
  72          * routines.
  73          */
  74
  75         dlm_create_root_list(ls);
  76
  77         /*
  78          * Add or remove nodes from the lockspace's ls_nodes list.
  79          */
  80
  81         error = dlm_recover_members(ls, rv, &neg);
  82         if (error) {
  83                 log_debug(ls, "dlm_recover_members error %d", error);
  84                 goto fail;
  85         }
  86
  87         ls->ls_recover_locks_in = 0;
  88
  89         dlm_set_recover_status(ls, DLM_RS_NODES);
  90
  91         error = dlm_recover_members_wait(ls);
  92         if (error) {
  93                 log_debug(ls, "dlm_recover_members_wait error %d", error);
  94                 goto fail;
  95         }
  96
  97         start = jiffies;
  98
  99         /*
 100          * Rebuild our own share of the directory by collecting from all other
 101          * nodes their master rsb names that hash to us.
 102          */
 103
 104         error = dlm_recover_directory(ls);
 105         if (error) {
 106                 log_debug(ls, "dlm_recover_directory error %d", error);
 107                 goto fail;
 108         }
 109
 110         dlm_set_recover_status(ls, DLM_RS_DIR);
 111
 112         error = dlm_recover_directory_wait(ls);
 113         if (error) {
 114                 log_debug(ls, "dlm_recover_directory_wait error %d", error);
 115                 goto fail;
 116         }
 117
 118         /*
 119          * We may have outstanding operations that are waiting for a reply from
 120          * a failed node.  Mark these to be resent after recovery.  Unlock and
 121          * cancel ops can just be completed.
 122          */
 123
 124         dlm_recover_waiters_pre(ls);
 125
 126         error = dlm_recovery_stopped(ls);
 127         if (error)
 128                 goto fail;
 129
 130         if (neg || dlm_no_directory(ls)) {
 131                 /*
 132                  * Clear lkb's for departed nodes.
 133                  */
 134
 135                 dlm_recover_purge(ls);
 136
 137                 /*
 138                  * Get new master nodeid's for rsb's that were mastered on
 139                  * departed nodes.
 140                  */
 141
 142                 error = dlm_recover_masters(ls);
 143                 if (error) {
 144                         log_debug(ls, "dlm_recover_masters error %d", error);
 145                         goto fail;
 146                 }
 147
 148                 /*
 149                  * Send our locks on remastered rsb's to the new masters.
 150                  */
 151
 152                 error = dlm_recover_locks(ls);
 153                 if (error) {
 154                         log_debug(ls, "dlm_recover_locks error %d", error);
 155                         goto fail;
 156                 }
 157
 158                 dlm_set_recover_status(ls, DLM_RS_LOCKS);
 159
 160                 error = dlm_recover_locks_wait(ls);
 161                 if (error) {
 162                         log_debug(ls, "dlm_recover_locks_wait error %d", error);
 163                         goto fail;
 164                 }
 165
 166                 log_debug(ls, "dlm_recover_locks %u in",
 167                           ls->ls_recover_locks_in);
 168
 169                 /*
 170                  * Finalize state in master rsb's now that all locks can be
 171                  * checked.  This includes conversion resolution and lvb
 172                  * settings.
 173                  */
 174
 175                 dlm_recover_rsbs(ls);
 176         } else {
 177                 /*
 178                  * Other lockspace members may be going through the "neg" steps
 179                  * while also adding us to the lockspace, in which case they'll
 180                  * be doing the recover_locks (RS_LOCKS) barrier.
 181                  */
 182                 dlm_set_recover_status(ls, DLM_RS_LOCKS);
 183
 184                 error = dlm_recover_locks_wait(ls);
 185                 if (error) {
 186                         log_debug(ls, "dlm_recover_locks_wait error %d", error);
 187                         goto fail;
 188                 }
 189         }
 190
 191         dlm_release_root_list(ls);
 192
 193         /*
 194          * Purge directory-related requests that are saved in requestqueue.
 195          * All dir requests from before recovery are invalid now due to the dir
 196          * rebuild and will be resent by the requesting nodes.
 197          */
 198
 199         dlm_purge_requestqueue(ls);
 200
 201         dlm_set_recover_status(ls, DLM_RS_DONE);
 202
 203         error = dlm_recover_done_wait(ls);
 204         if (error) {
 205                 log_debug(ls, "dlm_recover_done_wait error %d", error);
 206                 goto fail;
 207         }
 208
 209         dlm_clear_members_gone(ls);
 210
 211         dlm_adjust_timeouts(ls);
 212
 213         dlm_callback_resume(ls);
 214
 215         error = enable_locking(ls, rv->seq);
 216         if (error) {
 217                 log_debug(ls, "enable_locking error %d", error);
 218                 goto fail;
 219         }
 220
 221         error = dlm_process_requestqueue(ls);
 222         if (error) {
 223                 log_debug(ls, "dlm_process_requestqueue error %d", error);
 224                 goto fail;
 225         }
 226
 227         error = dlm_recover_waiters_post(ls);
 228         if (error) {
 229                 log_debug(ls, "dlm_recover_waiters_post error %d", error);
 230                 goto fail;
 231         }
 232
 233         dlm_recover_grant(ls);
 234
 235         log_debug(ls, "dlm_recover %llu generation %u done: %u ms",
 236                   (unsigned long long)rv->seq, ls->ls_generation,
 237                   jiffies_to_msecs(jiffies - start));
 238         mutex_unlock(&ls->ls_recoverd_active);
 239
 240         dlm_lsop_recover_done(ls);
 241         return 0;
 242
 243  fail:
 244         dlm_release_root_list(ls);
 245         log_debug(ls, "dlm_recover %llu error %d",
 246                   (unsigned long long)rv->seq, error);
 247         mutex_unlock(&ls->ls_recoverd_active);
 248         return error;
 249 }
 250
 251 /* The dlm_ls_start() that created the rv we take here may already have been
 252    stopped via dlm_ls_stop(); in that case we need to leave the RECOVERY_STOP
 253    flag set. */
 254
 255 static void do_ls_recovery(struct dlm_ls *ls)
 256 {
 257         struct dlm_recover *rv = NULL;
 258
 259         spin_lock(&ls->ls_recover_lock);
 260         rv = ls->ls_recover_args;
 261         ls->ls_recover_args = NULL;
 262         if (rv && ls->ls_recover_seq == rv->seq)
 263                 clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags);
 264         spin_unlock(&ls->ls_recover_lock);
 265
 266         if (rv) {
 267                 ls_recover(ls, rv);
 268                 kfree(rv->nodes);
 269                 kfree(rv);
 270         }
 271 }
 272
 273 static int dlm_recoverd(void *arg)
 274 {
 275         struct dlm_ls *ls;
 276
 277         ls = dlm_find_lockspace_local(arg);
 278         if (!ls) {
 279                 log_print("dlm_recoverd: no lockspace %p", arg);
 280                 return -1;
 281         }
 282
 283         while (!kthread_should_stop()) {
 284                 set_current_state(TASK_INTERRUPTIBLE);
 285                 if (!test_bit(LSFL_WORK, &ls->ls_flags))
 286                         schedule();
 287                 set_current_state(TASK_RUNNING);
 288
 289                 if (test_and_clear_bit(LSFL_WORK, &ls->ls_flags))
 290                         do_ls_recovery(ls);
 291         }
 292
 293         dlm_put_lockspace(ls);
 294         return 0;
 295 }
 296
 297 void dlm_recoverd_kick(struct dlm_ls *ls)
 298 {
 299         set_bit(LSFL_WORK, &ls->ls_flags);
 300         wake_up_process(ls->ls_recoverd_task);
 301 }
 302
 303 int dlm_recoverd_start(struct dlm_ls *ls)
 304 {
 305         struct task_struct *p;
 306         int error = 0;
 307
 308         p = kthread_run(dlm_recoverd, ls, "dlm_recoverd");
 309         if (IS_ERR(p))
 310                 error = PTR_ERR(p);
 311         else
 312                 ls->ls_recoverd_task = p;
 313         return error;
 314 }
 315
 316 void dlm_recoverd_stop(struct dlm_ls *ls)
 317 {
 318         kthread_stop(ls->ls_recoverd_task);
 319 }
 320
 321 void dlm_recoverd_suspend(struct dlm_ls *ls)
 322 {
 323         wake_up(&ls->ls_wait_general);
 324         mutex_lock(&ls->ls_recoverd_active);
 325 }
 326
 327 void dlm_recoverd_resume(struct dlm_ls *ls)
 328 {
 329         mutex_unlock(&ls->ls_recoverd_active);
 330 }
 331