drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c

   1 /*
   2  * GPL HEADER START
   3  *
   4  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License version 2 only,
   8  * as published by the Free Software Foundation.
   9  *
  10  * This program is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * General Public License version 2 for more details (a copy is included
  14  * in the LICENSE file that accompanied this code).
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * version 2 along with this program; If not, see
  18  * http://www.gnu.org/licenses/gpl-2.0.html
  19  *
  20  * GPL HEADER END
  21  */
  22 /*
  23  * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
  24  * Use is subject to license terms.
  25  *
  26  * Copyright (c) 2010, 2015, Intel Corporation.
  27  */
  28 /*
  29  * This file is part of Lustre, http://www.lustre.org/
  30  * Lustre is a trademark of Sun Microsystems, Inc.
  31  *
  32  * lustre/ldlm/ldlm_lockd.c
  33  *
  34  * Author: Peter Braam <braam@clusterfs.com>
  35  * Author: Phil Schwan <phil@clusterfs.com>
  36  */
  37
  38 #define DEBUG_SUBSYSTEM S_LDLM
  39
  40 #include "../../include/linux/libcfs/libcfs.h"
  41 #include "../include/lustre_dlm.h"
  42 #include "../include/obd_class.h"
  43 #include <linux/list.h>
  44 #include "ldlm_internal.h"
  45
  46 static int ldlm_num_threads;
  47 module_param(ldlm_num_threads, int, 0444);
  48 MODULE_PARM_DESC(ldlm_num_threads, "number of DLM service threads to start");
  49
  50 static char *ldlm_cpts;
  51 module_param(ldlm_cpts, charp, 0444);
  52 MODULE_PARM_DESC(ldlm_cpts, "CPU partitions ldlm threads should run on");
  53
  54 static struct mutex     ldlm_ref_mutex;
  55 static int ldlm_refcount;
  56
  57 static struct kobject *ldlm_kobj;
  58 struct kset *ldlm_ns_kset;
  59 static struct kset *ldlm_svc_kset;
  60
  61 struct ldlm_cb_async_args {
  62         struct ldlm_cb_set_arg *ca_set_arg;
  63         struct ldlm_lock       *ca_lock;
  64 };
  65
  66 /* LDLM state */
  67
  68 static struct ldlm_state *ldlm_state;
  69
  70 #define ELT_STOPPED   0
  71 #define ELT_READY     1
  72 #define ELT_TERMINATE 2
  73
  74 struct ldlm_bl_pool {
  75         spinlock_t              blp_lock;
  76
  77         /*
  78          * blp_prio_list is used for callbacks that should be handled
  79          * as a priority. It is used for LDLM_FL_DISCARD_DATA requests.
  80          * see bug 13843
  81          */
  82         struct list_head              blp_prio_list;
  83
  84         /*
  85          * blp_list is used for all other callbacks which are likely
  86          * to take longer to process.
  87          */
  88         struct list_head              blp_list;
  89
  90         wait_queue_head_t            blp_waitq;
  91         struct completion       blp_comp;
  92         atomic_t            blp_num_threads;
  93         atomic_t            blp_busy_threads;
  94         int                  blp_min_threads;
  95         int                  blp_max_threads;
  96 };
  97
  98 struct ldlm_bl_work_item {
  99         struct list_head              blwi_entry;
 100         struct ldlm_namespace  *blwi_ns;
 101         struct ldlm_lock_desc   blwi_ld;
 102         struct ldlm_lock       *blwi_lock;
 103         struct list_head              blwi_head;
 104         int                  blwi_count;
 105         struct completion       blwi_comp;
 106         enum ldlm_cancel_flags  blwi_flags;
 107         int                  blwi_mem_pressure;
 108 };
 109
 110 /**
 111  * Callback handler for receiving incoming blocking ASTs.
 112  *
 113  * This can only happen on client side.
 114  */
 115 void ldlm_handle_bl_callback(struct ldlm_namespace *ns,
 116                              struct ldlm_lock_desc *ld, struct ldlm_lock *lock)
 117 {
 118         int do_ast;
 119
 120         LDLM_DEBUG(lock, "client blocking AST callback handler");
 121
 122         lock_res_and_lock(lock);
 123         ldlm_set_cbpending(lock);
 124
 125         if (ldlm_is_cancel_on_block(lock))
 126                 ldlm_set_cancel(lock);
 127
 128         do_ast = !lock->l_readers && !lock->l_writers;
 129         unlock_res_and_lock(lock);
 130
 131         if (do_ast) {
 132                 CDEBUG(D_DLMTRACE,
 133                        "Lock %p already unused, calling callback (%p)\n", lock,
 134                        lock->l_blocking_ast);
 135                 if (lock->l_blocking_ast)
 136                         lock->l_blocking_ast(lock, ld, lock->l_ast_data,
 137                                              LDLM_CB_BLOCKING);
 138         } else {
 139                 CDEBUG(D_DLMTRACE,
 140                        "Lock %p is referenced, will be cancelled later\n",
 141                        lock);
 142         }
 143
 144         LDLM_DEBUG(lock, "client blocking callback handler END");
 145         LDLM_LOCK_RELEASE(lock);
 146 }
 147
 148 /**
 149  * Callback handler for receiving incoming completion ASTs.
 150  *
 151  * This only can happen on client side.
 152  */
 153 static void ldlm_handle_cp_callback(struct ptlrpc_request *req,
 154                                     struct ldlm_namespace *ns,
 155                                     struct ldlm_request *dlm_req,
 156                                     struct ldlm_lock *lock)
 157 {
 158         int lvb_len;
 159         LIST_HEAD(ast_list);
 160         int rc = 0;
 161
 162         LDLM_DEBUG(lock, "client completion callback handler START");
 163
 164         if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE)) {
 165                 int to = cfs_time_seconds(1);
 166
 167                 while (to > 0) {
 168                         set_current_state(TASK_INTERRUPTIBLE);
 169                         schedule_timeout(to);
 170                         if (lock->l_granted_mode == lock->l_req_mode ||
 171                             ldlm_is_destroyed(lock))
 172                                 break;
 173                 }
 174         }
 175
 176         lvb_len = req_capsule_get_size(&req->rq_pill, &RMF_DLM_LVB, RCL_CLIENT);
 177         if (lvb_len < 0) {
 178                 LDLM_ERROR(lock, "Fail to get lvb_len, rc = %d", lvb_len);
 179                 rc = lvb_len;
 180                 goto out;
 181         } else if (lvb_len > 0) {
 182                 if (lock->l_lvb_len > 0) {
 183                         /* for extent lock, lvb contains ost_lvb{}. */
 184                         LASSERT(lock->l_lvb_data);
 185
 186                         if (unlikely(lock->l_lvb_len < lvb_len)) {
 187                                 LDLM_ERROR(lock, "Replied LVB is larger than expectation, expected = %d, replied = %d",
 188                                            lock->l_lvb_len, lvb_len);
 189                                 rc = -EINVAL;
 190                                 goto out;
 191                         }
 192                 } else if (ldlm_has_layout(lock)) { /* for layout lock, lvb has
 193                                                      * variable length
 194                                                      */
 195                         void *lvb_data;
 196
 197                         lvb_data = kzalloc(lvb_len, GFP_NOFS);
 198                         if (!lvb_data) {
 199                                 LDLM_ERROR(lock, "No memory: %d.\n", lvb_len);
 200                                 rc = -ENOMEM;
 201                                 goto out;
 202                         }
 203
 204                         lock_res_and_lock(lock);
 205                         LASSERT(!lock->l_lvb_data);
 206                         lock->l_lvb_type = LVB_T_LAYOUT;
 207                         lock->l_lvb_data = lvb_data;
 208                         lock->l_lvb_len = lvb_len;
 209                         unlock_res_and_lock(lock);
 210                 }
 211         }
 212
 213         lock_res_and_lock(lock);
 214         if (ldlm_is_destroyed(lock) ||
 215             lock->l_granted_mode == lock->l_req_mode) {
 216                 /* bug 11300: the lock has already been granted */
 217                 unlock_res_and_lock(lock);
 218                 LDLM_DEBUG(lock, "Double grant race happened");
 219                 rc = 0;
 220                 goto out;
 221         }
 222
 223         /* If we receive the completion AST before the actual enqueue returned,
 224          * then we might need to switch lock modes, resources, or extents.
 225          */
 226         if (dlm_req->lock_desc.l_granted_mode != lock->l_req_mode) {
 227                 lock->l_req_mode = dlm_req->lock_desc.l_granted_mode;
 228                 LDLM_DEBUG(lock, "completion AST, new lock mode");
 229         }
 230
 231         if (lock->l_resource->lr_type != LDLM_PLAIN) {
 232                 ldlm_convert_policy_to_local(req->rq_export,
 233                                           dlm_req->lock_desc.l_resource.lr_type,
 234                                           &dlm_req->lock_desc.l_policy_data,
 235                                           &lock->l_policy_data);
 236                 LDLM_DEBUG(lock, "completion AST, new policy data");
 237         }
 238
 239         ldlm_resource_unlink_lock(lock);
 240         if (memcmp(&dlm_req->lock_desc.l_resource.lr_name,
 241                    &lock->l_resource->lr_name,
 242                    sizeof(lock->l_resource->lr_name)) != 0) {
 243                 unlock_res_and_lock(lock);
 244                 rc = ldlm_lock_change_resource(ns, lock,
 245                                 &dlm_req->lock_desc.l_resource.lr_name);
 246                 if (rc < 0) {
 247                         LDLM_ERROR(lock, "Failed to allocate resource");
 248                         goto out;
 249                 }
 250                 LDLM_DEBUG(lock, "completion AST, new resource");
 251                 CERROR("change resource!\n");
 252                 lock_res_and_lock(lock);
 253         }
 254
 255         if (dlm_req->lock_flags & LDLM_FL_AST_SENT) {
 256                 /* BL_AST locks are not needed in LRU.
 257                  * Let ldlm_cancel_lru() be fast.
 258                  */
 259                 ldlm_lock_remove_from_lru(lock);
 260                 lock->l_flags |= LDLM_FL_CBPENDING | LDLM_FL_BL_AST;
 261                 LDLM_DEBUG(lock, "completion AST includes blocking AST");
 262         }
 263
 264         if (lock->l_lvb_len > 0) {
 265                 rc = ldlm_fill_lvb(lock, &req->rq_pill, RCL_CLIENT,
 266                                    lock->l_lvb_data, lvb_len);
 267                 if (rc < 0) {
 268                         unlock_res_and_lock(lock);
 269                         goto out;
 270                 }
 271         }
 272
 273         ldlm_grant_lock(lock, &ast_list);
 274         unlock_res_and_lock(lock);
 275
 276         LDLM_DEBUG(lock, "callback handler finished, about to run_ast_work");
 277
 278         /* Let Enqueue to call osc_lock_upcall() and initialize l_ast_data */
 279         OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 2);
 280
 281         ldlm_run_ast_work(ns, &ast_list, LDLM_WORK_CP_AST);
 282
 283         LDLM_DEBUG_NOLOCK("client completion callback handler END (lock %p)",
 284                           lock);
 285         goto out;
 286
 287 out:
 288         if (rc < 0) {
 289                 lock_res_and_lock(lock);
 290                 ldlm_set_failed(lock);
 291                 unlock_res_and_lock(lock);
 292                 wake_up(&lock->l_waitq);
 293         }
 294         LDLM_LOCK_RELEASE(lock);
 295 }
 296
 297 /**
 298  * Callback handler for receiving incoming glimpse ASTs.
 299  *
 300  * This only can happen on client side.  After handling the glimpse AST
 301  * we also consider dropping the lock here if it is unused locally for a
 302  * long time.
 303  */
 304 static void ldlm_handle_gl_callback(struct ptlrpc_request *req,
 305                                     struct ldlm_namespace *ns,
 306                                     struct ldlm_request *dlm_req,
 307                                     struct ldlm_lock *lock)
 308 {
 309         int rc = -ENOSYS;
 310
 311         LDLM_DEBUG(lock, "client glimpse AST callback handler");
 312
 313         if (lock->l_glimpse_ast)
 314                 rc = lock->l_glimpse_ast(lock, req);
 315
 316         if (req->rq_repmsg) {
 317                 ptlrpc_reply(req);
 318         } else {
 319                 req->rq_status = rc;
 320                 ptlrpc_error(req);
 321         }
 322
 323         lock_res_and_lock(lock);
 324         if (lock->l_granted_mode == LCK_PW &&
 325             !lock->l_readers && !lock->l_writers &&
 326             cfs_time_after(cfs_time_current(),
 327                            cfs_time_add(lock->l_last_used,
 328                                         cfs_time_seconds(10)))) {
 329                 unlock_res_and_lock(lock);
 330                 if (ldlm_bl_to_thread_lock(ns, NULL, lock))
 331                         ldlm_handle_bl_callback(ns, NULL, lock);
 332
 333                 return;
 334         }
 335         unlock_res_and_lock(lock);
 336         LDLM_LOCK_RELEASE(lock);
 337 }
 338
 339 static int ldlm_callback_reply(struct ptlrpc_request *req, int rc)
 340 {
 341         if (req->rq_no_reply)
 342                 return 0;
 343
 344         req->rq_status = rc;
 345         if (!req->rq_packed_final) {
 346                 rc = lustre_pack_reply(req, 1, NULL, NULL);
 347                 if (rc)
 348                         return rc;
 349         }
 350         return ptlrpc_reply(req);
 351 }
 352
 353 static int __ldlm_bl_to_thread(struct ldlm_bl_work_item *blwi,
 354                                enum ldlm_cancel_flags cancel_flags)
 355 {
 356         struct ldlm_bl_pool *blp = ldlm_state->ldlm_bl_pool;
 357
 358         spin_lock(&blp->blp_lock);
 359         if (blwi->blwi_lock && ldlm_is_discard_data(blwi->blwi_lock)) {
 360                 /* add LDLM_FL_DISCARD_DATA requests to the priority list */
 361                 list_add_tail(&blwi->blwi_entry, &blp->blp_prio_list);
 362         } else {
 363                 /* other blocking callbacks are added to the regular list */
 364                 list_add_tail(&blwi->blwi_entry, &blp->blp_list);
 365         }
 366         spin_unlock(&blp->blp_lock);
 367
 368         wake_up(&blp->blp_waitq);
 369
 370         /* can not check blwi->blwi_flags as blwi could be already freed in
 371          * LCF_ASYNC mode
 372          */
 373         if (!(cancel_flags & LCF_ASYNC))
 374                 wait_for_completion(&blwi->blwi_comp);
 375
 376         return 0;
 377 }
 378
 379 static inline void init_blwi(struct ldlm_bl_work_item *blwi,
 380                              struct ldlm_namespace *ns,
 381                              struct ldlm_lock_desc *ld,
 382                              struct list_head *cancels, int count,
 383                              struct ldlm_lock *lock,
 384                              enum ldlm_cancel_flags cancel_flags)
 385 {
 386         init_completion(&blwi->blwi_comp);
 387         INIT_LIST_HEAD(&blwi->blwi_head);
 388
 389         if (memory_pressure_get())
 390                 blwi->blwi_mem_pressure = 1;
 391
 392         blwi->blwi_ns = ns;
 393         blwi->blwi_flags = cancel_flags;
 394         if (ld)
 395                 blwi->blwi_ld = *ld;
 396         if (count) {
 397                 list_add(&blwi->blwi_head, cancels);
 398                 list_del_init(cancels);
 399                 blwi->blwi_count = count;
 400         } else {
 401                 blwi->blwi_lock = lock;
 402         }
 403 }
 404
 405 /**
 406  * Queues a list of locks \a cancels containing \a count locks
 407  * for later processing by a blocking thread.  If \a count is zero,
 408  * then the lock referenced as \a lock is queued instead.
 409  *
 410  * The blocking thread would then call ->l_blocking_ast callback in the lock.
 411  * If list addition fails an error is returned and caller is supposed to
 412  * call ->l_blocking_ast itself.
 413  */
 414 static int ldlm_bl_to_thread(struct ldlm_namespace *ns,
 415                              struct ldlm_lock_desc *ld,
 416                              struct ldlm_lock *lock,
 417                              struct list_head *cancels, int count,
 418                              enum ldlm_cancel_flags cancel_flags)
 419 {
 420         if (cancels && count == 0)
 421                 return 0;
 422
 423         if (cancel_flags & LCF_ASYNC) {
 424                 struct ldlm_bl_work_item *blwi;
 425
 426                 blwi = kzalloc(sizeof(*blwi), GFP_NOFS);
 427                 if (!blwi)
 428                         return -ENOMEM;
 429                 init_blwi(blwi, ns, ld, cancels, count, lock, cancel_flags);
 430
 431                 return __ldlm_bl_to_thread(blwi, cancel_flags);
 432         } else {
 433                 /* if it is synchronous call do minimum mem alloc, as it could
 434                  * be triggered from kernel shrinker
 435                  */
 436                 struct ldlm_bl_work_item blwi;
 437
 438                 memset(&blwi, 0, sizeof(blwi));
 439                 init_blwi(&blwi, ns, ld, cancels, count, lock, cancel_flags);
 440                 return __ldlm_bl_to_thread(&blwi, cancel_flags);
 441         }
 442 }
 443
 444 int ldlm_bl_to_thread_lock(struct ldlm_namespace *ns, struct ldlm_lock_desc *ld,
 445                            struct ldlm_lock *lock)
 446 {
 447         return ldlm_bl_to_thread(ns, ld, lock, NULL, 0, LCF_ASYNC);
 448 }
 449
 450 int ldlm_bl_to_thread_list(struct ldlm_namespace *ns, struct ldlm_lock_desc *ld,
 451                            struct list_head *cancels, int count,
 452                            enum ldlm_cancel_flags cancel_flags)
 453 {
 454         return ldlm_bl_to_thread(ns, ld, NULL, cancels, count, cancel_flags);
 455 }
 456
 457 /* Setinfo coming from Server (eg MDT) to Client (eg MDC)! */
 458 static int ldlm_handle_setinfo(struct ptlrpc_request *req)
 459 {
 460         struct obd_device *obd = req->rq_export->exp_obd;
 461         char *key;
 462         void *val;
 463         int keylen, vallen;
 464         int rc = -ENOSYS;
 465
 466         DEBUG_REQ(D_HSM, req, "%s: handle setinfo\n", obd->obd_name);
 467
 468         req_capsule_set(&req->rq_pill, &RQF_OBD_SET_INFO);
 469
 470         key = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
 471         if (!key) {
 472                 DEBUG_REQ(D_IOCTL, req, "no set_info key");
 473                 return -EFAULT;
 474         }
 475         keylen = req_capsule_get_size(&req->rq_pill, &RMF_SETINFO_KEY,
 476                                       RCL_CLIENT);
 477         val = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_VAL);
 478         if (!val) {
 479                 DEBUG_REQ(D_IOCTL, req, "no set_info val");
 480                 return -EFAULT;
 481         }
 482         vallen = req_capsule_get_size(&req->rq_pill, &RMF_SETINFO_VAL,
 483                                       RCL_CLIENT);
 484
 485         /* We are responsible for swabbing contents of val */
 486
 487         if (KEY_IS(KEY_HSM_COPYTOOL_SEND))
 488                 /* Pass it on to mdc (the "export" in this case) */
 489                 rc = obd_set_info_async(req->rq_svc_thread->t_env,
 490                                         req->rq_export,
 491                                         sizeof(KEY_HSM_COPYTOOL_SEND),
 492                                         KEY_HSM_COPYTOOL_SEND,
 493                                         vallen, val, NULL);
 494         else
 495                 DEBUG_REQ(D_WARNING, req, "ignoring unknown key %s", key);
 496
 497         return rc;
 498 }
 499
 500 static inline void ldlm_callback_errmsg(struct ptlrpc_request *req,
 501                                         const char *msg, int rc,
 502                                         const struct lustre_handle *handle)
 503 {
 504         DEBUG_REQ((req->rq_no_reply || rc) ? D_WARNING : D_DLMTRACE, req,
 505                   "%s: [nid %s] [rc %d] [lock %#llx]",
 506                   msg, libcfs_id2str(req->rq_peer), rc,
 507                   handle ? handle->cookie : 0);
 508         if (req->rq_no_reply)
 509                 CWARN("No reply was sent, maybe cause bug 21636.\n");
 510         else if (rc)
 511                 CWARN("Send reply failed, maybe cause bug 21636.\n");
 512 }
 513
 514 static int ldlm_handle_qc_callback(struct ptlrpc_request *req)
 515 {
 516         struct obd_quotactl *oqctl;
 517         struct client_obd *cli = &req->rq_export->exp_obd->u.cli;
 518
 519         oqctl = req_capsule_client_get(&req->rq_pill, &RMF_OBD_QUOTACTL);
 520         if (!oqctl) {
 521                 CERROR("Can't unpack obd_quotactl\n");
 522                 return -EPROTO;
 523         }
 524
 525         oqctl->qc_stat = ptlrpc_status_ntoh(oqctl->qc_stat);
 526
 527         cli->cl_qchk_stat = oqctl->qc_stat;
 528         return 0;
 529 }
 530
 531 /* TODO: handle requests in a similar way as MDT: see mdt_handle_common() */
 532 static int ldlm_callback_handler(struct ptlrpc_request *req)
 533 {
 534         struct ldlm_namespace *ns;
 535         struct ldlm_request *dlm_req;
 536         struct ldlm_lock *lock;
 537         int rc;
 538
 539         /* Requests arrive in sender's byte order.  The ptlrpc service
 540          * handler has already checked and, if necessary, byte-swapped the
 541          * incoming request message body, but I am responsible for the
 542          * message buffers.
 543          */
 544
 545         /* do nothing for sec context finalize */
 546         if (lustre_msg_get_opc(req->rq_reqmsg) == SEC_CTX_FINI)
 547                 return 0;
 548
 549         req_capsule_init(&req->rq_pill, req, RCL_SERVER);
 550
 551         if (!req->rq_export) {
 552                 rc = ldlm_callback_reply(req, -ENOTCONN);
 553                 ldlm_callback_errmsg(req, "Operate on unconnected server",
 554                                      rc, NULL);
 555                 return 0;
 556         }
 557
 558         LASSERT(req->rq_export->exp_obd);
 559
 560         switch (lustre_msg_get_opc(req->rq_reqmsg)) {
 561         case LDLM_BL_CALLBACK:
 562                 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_BL_CALLBACK_NET))
 563                         return 0;
 564                 break;
 565         case LDLM_CP_CALLBACK:
 566                 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CP_CALLBACK_NET))
 567                         return 0;
 568                 break;
 569         case LDLM_GL_CALLBACK:
 570                 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_GL_CALLBACK_NET))
 571                         return 0;
 572                 break;
 573         case LDLM_SET_INFO:
 574                 rc = ldlm_handle_setinfo(req);
 575                 ldlm_callback_reply(req, rc);
 576                 return 0;
 577         case OBD_QC_CALLBACK:
 578                 req_capsule_set(&req->rq_pill, &RQF_QC_CALLBACK);
 579                 if (OBD_FAIL_CHECK(OBD_FAIL_OBD_QC_CALLBACK_NET))
 580                         return 0;
 581                 rc = ldlm_handle_qc_callback(req);
 582                 ldlm_callback_reply(req, rc);
 583                 return 0;
 584         default:
 585                 CERROR("unknown opcode %u\n",
 586                        lustre_msg_get_opc(req->rq_reqmsg));
 587                 ldlm_callback_reply(req, -EPROTO);
 588                 return 0;
 589         }
 590
 591         ns = req->rq_export->exp_obd->obd_namespace;
 592         LASSERT(ns);
 593
 594         req_capsule_set(&req->rq_pill, &RQF_LDLM_CALLBACK);
 595
 596         dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
 597         if (!dlm_req) {
 598                 rc = ldlm_callback_reply(req, -EPROTO);
 599                 ldlm_callback_errmsg(req, "Operate without parameter", rc,
 600                                      NULL);
 601                 return 0;
 602         }
 603
 604         /* Force a known safe race, send a cancel to the server for a lock
 605          * which the server has already started a blocking callback on.
 606          */
 607         if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE) &&
 608             lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
 609                 rc = ldlm_cli_cancel(&dlm_req->lock_handle[0], 0);
 610                 if (rc < 0)
 611                         CERROR("ldlm_cli_cancel: %d\n", rc);
 612         }
 613
 614         lock = ldlm_handle2lock_long(&dlm_req->lock_handle[0], 0);
 615         if (!lock) {
 616                 CDEBUG(D_DLMTRACE, "callback on lock %#llx - lock disappeared\n",
 617                        dlm_req->lock_handle[0].cookie);
 618                 rc = ldlm_callback_reply(req, -EINVAL);
 619                 ldlm_callback_errmsg(req, "Operate with invalid parameter", rc,
 620                                      &dlm_req->lock_handle[0]);
 621                 return 0;
 622         }
 623
 624         if (ldlm_is_fail_loc(lock) &&
 625             lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK)
 626                 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
 627
 628         /* Copy hints/flags (e.g. LDLM_FL_DISCARD_DATA) from AST. */
 629         lock_res_and_lock(lock);
 630         lock->l_flags |= ldlm_flags_from_wire(dlm_req->lock_flags &
 631                                               LDLM_FL_AST_MASK);
 632         if (lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
 633                 /* If somebody cancels lock and cache is already dropped,
 634                  * or lock is failed before cp_ast received on client,
 635                  * we can tell the server we have no lock. Otherwise, we
 636                  * should send cancel after dropping the cache.
 637                  */
 638                 if ((ldlm_is_canceling(lock) && ldlm_is_bl_done(lock)) ||
 639                     ldlm_is_failed(lock)) {
 640                         LDLM_DEBUG(lock,
 641                                    "callback on lock %#llx - lock disappeared",
 642                                    dlm_req->lock_handle[0].cookie);
 643                         unlock_res_and_lock(lock);
 644                         LDLM_LOCK_RELEASE(lock);
 645                         rc = ldlm_callback_reply(req, -EINVAL);
 646                         ldlm_callback_errmsg(req, "Operate on stale lock", rc,
 647                                              &dlm_req->lock_handle[0]);
 648                         return 0;
 649                 }
 650                 /* BL_AST locks are not needed in LRU.
 651                  * Let ldlm_cancel_lru() be fast.
 652                  */
 653                 ldlm_lock_remove_from_lru(lock);
 654                 ldlm_set_bl_ast(lock);
 655         }
 656         unlock_res_and_lock(lock);
 657
 658         /* We want the ost thread to get this reply so that it can respond
 659          * to ost requests (write cache writeback) that might be triggered
 660          * in the callback.
 661          *
 662          * But we'd also like to be able to indicate in the reply that we're
 663          * cancelling right now, because it's unused, or have an intent result
 664          * in the reply, so we might have to push the responsibility for sending
 665          * the reply down into the AST handlers, alas.
 666          */
 667
 668         switch (lustre_msg_get_opc(req->rq_reqmsg)) {
 669         case LDLM_BL_CALLBACK:
 670                 CDEBUG(D_INODE, "blocking ast\n");
 671                 req_capsule_extend(&req->rq_pill, &RQF_LDLM_BL_CALLBACK);
 672                 if (!ldlm_is_cancel_on_block(lock)) {
 673                         rc = ldlm_callback_reply(req, 0);
 674                         if (req->rq_no_reply || rc)
 675                                 ldlm_callback_errmsg(req, "Normal process", rc,
 676                                                      &dlm_req->lock_handle[0]);
 677                 }
 678                 if (ldlm_bl_to_thread_lock(ns, &dlm_req->lock_desc, lock))
 679                         ldlm_handle_bl_callback(ns, &dlm_req->lock_desc, lock);
 680                 break;
 681         case LDLM_CP_CALLBACK:
 682                 CDEBUG(D_INODE, "completion ast\n");
 683                 req_capsule_extend(&req->rq_pill, &RQF_LDLM_CP_CALLBACK);
 684                 ldlm_callback_reply(req, 0);
 685                 ldlm_handle_cp_callback(req, ns, dlm_req, lock);
 686                 break;
 687         case LDLM_GL_CALLBACK:
 688                 CDEBUG(D_INODE, "glimpse ast\n");
 689                 req_capsule_extend(&req->rq_pill, &RQF_LDLM_GL_CALLBACK);
 690                 ldlm_handle_gl_callback(req, ns, dlm_req, lock);
 691                 break;
 692         default:
 693                 LBUG();                  /* checked above */
 694         }
 695
 696         return 0;
 697 }
 698
 699 static struct ldlm_bl_work_item *ldlm_bl_get_work(struct ldlm_bl_pool *blp)
 700 {
 701         struct ldlm_bl_work_item *blwi = NULL;
 702         static unsigned int num_bl;
 703
 704         spin_lock(&blp->blp_lock);
 705         /* process a request from the blp_list at least every blp_num_threads */
 706         if (!list_empty(&blp->blp_list) &&
 707             (list_empty(&blp->blp_prio_list) || num_bl == 0))
 708                 blwi = list_entry(blp->blp_list.next,
 709                                       struct ldlm_bl_work_item, blwi_entry);
 710         else
 711                 if (!list_empty(&blp->blp_prio_list))
 712                         blwi = list_entry(blp->blp_prio_list.next,
 713                                               struct ldlm_bl_work_item,
 714                                               blwi_entry);
 715
 716         if (blwi) {
 717                 if (++num_bl >= atomic_read(&blp->blp_num_threads))
 718                         num_bl = 0;
 719                 list_del(&blwi->blwi_entry);
 720         }
 721         spin_unlock(&blp->blp_lock);
 722
 723         return blwi;
 724 }
 725
 726 /* This only contains temporary data until the thread starts */
 727 struct ldlm_bl_thread_data {
 728         char                    bltd_name[CFS_CURPROC_COMM_MAX];
 729         struct ldlm_bl_pool     *bltd_blp;
 730         struct completion       bltd_comp;
 731         int                     bltd_num;
 732 };
 733
 734 static int ldlm_bl_thread_main(void *arg);
 735
 736 static int ldlm_bl_thread_start(struct ldlm_bl_pool *blp)
 737 {
 738         struct ldlm_bl_thread_data bltd = { .bltd_blp = blp };
 739         struct task_struct *task;
 740
 741         init_completion(&bltd.bltd_comp);
 742         bltd.bltd_num = atomic_read(&blp->blp_num_threads);
 743         snprintf(bltd.bltd_name, sizeof(bltd.bltd_name),
 744                 "ldlm_bl_%02d", bltd.bltd_num);
 745         task = kthread_run(ldlm_bl_thread_main, &bltd, "%s", bltd.bltd_name);
 746         if (IS_ERR(task)) {
 747                 CERROR("cannot start LDLM thread ldlm_bl_%02d: rc %ld\n",
 748                        atomic_read(&blp->blp_num_threads), PTR_ERR(task));
 749                 return PTR_ERR(task);
 750         }
 751         wait_for_completion(&bltd.bltd_comp);
 752
 753         return 0;
 754 }
 755
 756 /**
 757  * Main blocking requests processing thread.
 758  *
 759  * Callers put locks into its queue by calling ldlm_bl_to_thread.
 760  * This thread in the end ends up doing actual call to ->l_blocking_ast
 761  * for queued locks.
 762  */
 763 static int ldlm_bl_thread_main(void *arg)
 764 {
 765         struct ldlm_bl_pool *blp;
 766
 767         {
 768                 struct ldlm_bl_thread_data *bltd = arg;
 769
 770                 blp = bltd->bltd_blp;
 771
 772                 atomic_inc(&blp->blp_num_threads);
 773                 atomic_inc(&blp->blp_busy_threads);
 774
 775                 complete(&bltd->bltd_comp);
 776                 /* cannot use bltd after this, it is only on caller's stack */
 777         }
 778
 779         while (1) {
 780                 struct l_wait_info lwi = { 0 };
 781                 struct ldlm_bl_work_item *blwi = NULL;
 782                 int busy;
 783
 784                 blwi = ldlm_bl_get_work(blp);
 785
 786                 if (!blwi) {
 787                         atomic_dec(&blp->blp_busy_threads);
 788                         l_wait_event_exclusive(blp->blp_waitq,
 789                                          (blwi = ldlm_bl_get_work(blp)),
 790                                          &lwi);
 791                         busy = atomic_inc_return(&blp->blp_busy_threads);
 792                 } else {
 793                         busy = atomic_read(&blp->blp_busy_threads);
 794                 }
 795
 796                 if (!blwi->blwi_ns)
 797                         /* added by ldlm_cleanup() */
 798                         break;
 799
 800                 /* Not fatal if racy and have a few too many threads */
 801                 if (unlikely(busy < blp->blp_max_threads &&
 802                              busy >= atomic_read(&blp->blp_num_threads) &&
 803                              !blwi->blwi_mem_pressure))
 804                         /* discard the return value, we tried */
 805                         ldlm_bl_thread_start(blp);
 806
 807                 if (blwi->blwi_mem_pressure)
 808                         memory_pressure_set();
 809
 810                 if (blwi->blwi_count) {
 811                         int count;
 812                         /* The special case when we cancel locks in LRU
 813                          * asynchronously, we pass the list of locks here.
 814                          * Thus locks are marked LDLM_FL_CANCELING, but NOT
 815                          * canceled locally yet.
 816                          */
 817                         count = ldlm_cli_cancel_list_local(&blwi->blwi_head,
 818                                                            blwi->blwi_count,
 819                                                            LCF_BL_AST);
 820                         ldlm_cli_cancel_list(&blwi->blwi_head, count, NULL,
 821                                              blwi->blwi_flags);
 822                 } else {
 823                         ldlm_handle_bl_callback(blwi->blwi_ns, &blwi->blwi_ld,
 824                                                 blwi->blwi_lock);
 825                 }
 826                 if (blwi->blwi_mem_pressure)
 827                         memory_pressure_clr();
 828
 829                 if (blwi->blwi_flags & LCF_ASYNC)
 830                         kfree(blwi);
 831                 else
 832                         complete(&blwi->blwi_comp);
 833         }
 834
 835         atomic_dec(&blp->blp_busy_threads);
 836         atomic_dec(&blp->blp_num_threads);
 837         complete(&blp->blp_comp);
 838         return 0;
 839 }
 840
 841 static int ldlm_setup(void);
 842 static int ldlm_cleanup(void);
 843
 844 int ldlm_get_ref(void)
 845 {
 846         int rc = 0;
 847
 848         mutex_lock(&ldlm_ref_mutex);
 849         if (++ldlm_refcount == 1) {
 850                 rc = ldlm_setup();
 851                 if (rc)
 852                         ldlm_refcount--;
 853         }
 854         mutex_unlock(&ldlm_ref_mutex);
 855
 856         return rc;
 857 }
 858 EXPORT_SYMBOL(ldlm_get_ref);
 859
 860 void ldlm_put_ref(void)
 861 {
 862         mutex_lock(&ldlm_ref_mutex);
 863         if (ldlm_refcount == 1) {
 864                 int rc = ldlm_cleanup();
 865
 866                 if (rc)
 867                         CERROR("ldlm_cleanup failed: %d\n", rc);
 868                 else
 869                         ldlm_refcount--;
 870         } else {
 871                 ldlm_refcount--;
 872         }
 873         mutex_unlock(&ldlm_ref_mutex);
 874 }
 875 EXPORT_SYMBOL(ldlm_put_ref);
 876
 877 extern unsigned int ldlm_cancel_unused_locks_before_replay;
 878
 879 static ssize_t cancel_unused_locks_before_replay_show(struct kobject *kobj,
 880                                                       struct attribute *attr,
 881                                                       char *buf)
 882 {
 883         return sprintf(buf, "%d\n", ldlm_cancel_unused_locks_before_replay);
 884 }
 885
 886 static ssize_t cancel_unused_locks_before_replay_store(struct kobject *kobj,
 887                                                        struct attribute *attr,
 888                                                        const char *buffer,
 889                                                        size_t count)
 890 {
 891         int rc;
 892         unsigned long val;
 893
 894         rc = kstrtoul(buffer, 10, &val);
 895         if (rc)
 896                 return rc;
 897
 898         ldlm_cancel_unused_locks_before_replay = val;
 899
 900         return count;
 901 }
 902 LUSTRE_RW_ATTR(cancel_unused_locks_before_replay);
 903
 904 /* These are for root of /sys/fs/lustre/ldlm */
 905 static struct attribute *ldlm_attrs[] = {
 906         &lustre_attr_cancel_unused_locks_before_replay.attr,
 907         NULL,
 908 };
 909
 910 static struct attribute_group ldlm_attr_group = {
 911         .attrs = ldlm_attrs,
 912 };
 913
 914 static int ldlm_setup(void)
 915 {
 916         static struct ptlrpc_service_conf       conf;
 917         struct ldlm_bl_pool                     *blp = NULL;
 918         int rc = 0;
 919         int i;
 920
 921         if (ldlm_state)
 922                 return -EALREADY;
 923
 924         ldlm_state = kzalloc(sizeof(*ldlm_state), GFP_NOFS);
 925         if (!ldlm_state)
 926                 return -ENOMEM;
 927
 928         ldlm_kobj = kobject_create_and_add("ldlm", lustre_kobj);
 929         if (!ldlm_kobj) {
 930                 rc = -ENOMEM;
 931                 goto out;
 932         }
 933
 934         rc = sysfs_create_group(ldlm_kobj, &ldlm_attr_group);
 935         if (rc)
 936                 goto out;
 937
 938         ldlm_ns_kset = kset_create_and_add("namespaces", NULL, ldlm_kobj);
 939         if (!ldlm_ns_kset) {
 940                 rc = -ENOMEM;
 941                 goto out;
 942         }
 943
 944         ldlm_svc_kset = kset_create_and_add("services", NULL, ldlm_kobj);
 945         if (!ldlm_svc_kset) {
 946                 rc = -ENOMEM;
 947                 goto out;
 948         }
 949
 950         rc = ldlm_debugfs_setup();
 951         if (rc != 0)
 952                 goto out;
 953
 954         memset(&conf, 0, sizeof(conf));
 955         conf = (typeof(conf)) {
 956                 .psc_name               = "ldlm_cbd",
 957                 .psc_watchdog_factor    = 2,
 958                 .psc_buf                = {
 959                         .bc_nbufs               = LDLM_CLIENT_NBUFS,
 960                         .bc_buf_size            = LDLM_BUFSIZE,
 961                         .bc_req_max_size        = LDLM_MAXREQSIZE,
 962                         .bc_rep_max_size        = LDLM_MAXREPSIZE,
 963                         .bc_req_portal          = LDLM_CB_REQUEST_PORTAL,
 964                         .bc_rep_portal          = LDLM_CB_REPLY_PORTAL,
 965                 },
 966                 .psc_thr                = {
 967                         .tc_thr_name            = "ldlm_cb",
 968                         .tc_thr_factor          = LDLM_THR_FACTOR,
 969                         .tc_nthrs_init          = LDLM_NTHRS_INIT,
 970                         .tc_nthrs_base          = LDLM_NTHRS_BASE,
 971                         .tc_nthrs_max           = LDLM_NTHRS_MAX,
 972                         .tc_nthrs_user          = ldlm_num_threads,
 973                         .tc_cpu_affinity        = 1,
 974                         .tc_ctx_tags            = LCT_MD_THREAD | LCT_DT_THREAD,
 975                 },
 976                 .psc_cpt                = {
 977                         .cc_pattern             = ldlm_cpts,
 978                 },
 979                 .psc_ops                = {
 980                         .so_req_handler         = ldlm_callback_handler,
 981                 },
 982         };
 983         ldlm_state->ldlm_cb_service =
 984                         ptlrpc_register_service(&conf, ldlm_svc_kset,
 985                                                 ldlm_svc_debugfs_dir);
 986         if (IS_ERR(ldlm_state->ldlm_cb_service)) {
 987                 CERROR("failed to start service\n");
 988                 rc = PTR_ERR(ldlm_state->ldlm_cb_service);
 989                 ldlm_state->ldlm_cb_service = NULL;
 990                 goto out;
 991         }
 992
 993         blp = kzalloc(sizeof(*blp), GFP_NOFS);
 994         if (!blp) {
 995                 rc = -ENOMEM;
 996                 goto out;
 997         }
 998         ldlm_state->ldlm_bl_pool = blp;
 999
1000         spin_lock_init(&blp->blp_lock);
1001         INIT_LIST_HEAD(&blp->blp_list);
1002         INIT_LIST_HEAD(&blp->blp_prio_list);
1003         init_waitqueue_head(&blp->blp_waitq);
1004         atomic_set(&blp->blp_num_threads, 0);
1005         atomic_set(&blp->blp_busy_threads, 0);
1006
1007         if (ldlm_num_threads == 0) {
1008                 blp->blp_min_threads = LDLM_NTHRS_INIT;
1009                 blp->blp_max_threads = LDLM_NTHRS_MAX;
1010         } else {
1011                 blp->blp_min_threads = min_t(int, LDLM_NTHRS_MAX,
1012                                              max_t(int, LDLM_NTHRS_INIT,
1013                                                    ldlm_num_threads));
1014
1015                 blp->blp_max_threads = blp->blp_min_threads;
1016         }
1017
1018         for (i = 0; i < blp->blp_min_threads; i++) {
1019                 rc = ldlm_bl_thread_start(blp);
1020                 if (rc < 0)
1021                         goto out;
1022         }
1023
1024         rc = ldlm_pools_init();
1025         if (rc) {
1026                 CERROR("Failed to initialize LDLM pools: %d\n", rc);
1027                 goto out;
1028         }
1029         return 0;
1030
1031  out:
1032         ldlm_cleanup();
1033         return rc;
1034 }
1035
1036 static int ldlm_cleanup(void)
1037 {
1038         if (!list_empty(ldlm_namespace_list(LDLM_NAMESPACE_SERVER)) ||
1039             !list_empty(ldlm_namespace_list(LDLM_NAMESPACE_CLIENT))) {
1040                 CERROR("ldlm still has namespaces; clean these up first.\n");
1041                 ldlm_dump_all_namespaces(LDLM_NAMESPACE_SERVER, D_DLMTRACE);
1042                 ldlm_dump_all_namespaces(LDLM_NAMESPACE_CLIENT, D_DLMTRACE);
1043                 return -EBUSY;
1044         }
1045
1046         ldlm_pools_fini();
1047
1048         if (ldlm_state->ldlm_bl_pool) {
1049                 struct ldlm_bl_pool *blp = ldlm_state->ldlm_bl_pool;
1050
1051                 while (atomic_read(&blp->blp_num_threads) > 0) {
1052                         struct ldlm_bl_work_item blwi = { .blwi_ns = NULL };
1053
1054                         init_completion(&blp->blp_comp);
1055
1056                         spin_lock(&blp->blp_lock);
1057                         list_add_tail(&blwi.blwi_entry, &blp->blp_list);
1058                         wake_up(&blp->blp_waitq);
1059                         spin_unlock(&blp->blp_lock);
1060
1061                         wait_for_completion(&blp->blp_comp);
1062                 }
1063
1064                 kfree(blp);
1065         }
1066
1067         if (ldlm_state->ldlm_cb_service)
1068                 ptlrpc_unregister_service(ldlm_state->ldlm_cb_service);
1069
1070         if (ldlm_ns_kset)
1071                 kset_unregister(ldlm_ns_kset);
1072         if (ldlm_svc_kset)
1073                 kset_unregister(ldlm_svc_kset);
1074         if (ldlm_kobj)
1075                 kobject_put(ldlm_kobj);
1076
1077         ldlm_debugfs_cleanup();
1078
1079         kfree(ldlm_state);
1080         ldlm_state = NULL;
1081
1082         return 0;
1083 }
1084
1085 int ldlm_init(void)
1086 {
1087         mutex_init(&ldlm_ref_mutex);
1088         mutex_init(ldlm_namespace_lock(LDLM_NAMESPACE_SERVER));
1089         mutex_init(ldlm_namespace_lock(LDLM_NAMESPACE_CLIENT));
1090         ldlm_resource_slab = kmem_cache_create("ldlm_resources",
1091                                                sizeof(struct ldlm_resource), 0,
1092                                                SLAB_HWCACHE_ALIGN, NULL);
1093         if (!ldlm_resource_slab)
1094                 return -ENOMEM;
1095
1096         ldlm_lock_slab = kmem_cache_create("ldlm_locks",
1097                               sizeof(struct ldlm_lock), 0,
1098                               SLAB_HWCACHE_ALIGN | SLAB_DESTROY_BY_RCU, NULL);
1099         if (!ldlm_lock_slab) {
1100                 kmem_cache_destroy(ldlm_resource_slab);
1101                 return -ENOMEM;
1102         }
1103
1104         ldlm_interval_slab = kmem_cache_create("interval_node",
1105                                         sizeof(struct ldlm_interval),
1106                                         0, SLAB_HWCACHE_ALIGN, NULL);
1107         if (!ldlm_interval_slab) {
1108                 kmem_cache_destroy(ldlm_resource_slab);
1109                 kmem_cache_destroy(ldlm_lock_slab);
1110                 return -ENOMEM;
1111         }
1112 #if LUSTRE_TRACKS_LOCK_EXP_REFS
1113         class_export_dump_hook = ldlm_dump_export_locks;
1114 #endif
1115         return 0;
1116 }
1117
1118 void ldlm_exit(void)
1119 {
1120         if (ldlm_refcount)
1121                 CERROR("ldlm_refcount is %d in ldlm_exit!\n", ldlm_refcount);
1122         kmem_cache_destroy(ldlm_resource_slab);
1123         /* ldlm_lock_put() use RCU to call ldlm_lock_free, so need call
1124          * synchronize_rcu() to wait a grace period elapsed, so that
1125          * ldlm_lock_free() get a chance to be called.
1126          */
1127         synchronize_rcu();
1128         kmem_cache_destroy(ldlm_lock_slab);
1129         kmem_cache_destroy(ldlm_interval_slab);
1130 }