lustre/osc/osc_request.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) 2011, 2017, Intel Corporation.
  27  */
  28 /*
  29  * This file is part of Lustre, http://www.lustre.org/
  30  */
  31
  32 #define DEBUG_SUBSYSTEM S_OSC
  33
  34 #include <linux/workqueue.h>
  35 #include <libcfs/libcfs.h>
  36 #include <linux/falloc.h>
  37 #include <lprocfs_status.h>
  38 #include <lustre_dlm.h>
  39 #include <lustre_fid.h>
  40 #include <lustre_ha.h>
  41 #include <uapi/linux/lustre/lustre_ioctl.h>
  42 #include <lustre_net.h>
  43 #include <lustre_obdo.h>
  44 #include <obd.h>
  45 #include <obd_cksum.h>
  46 #include <obd_class.h>
  47 #include <lustre_osc.h>
  48 #include <linux/falloc.h>
  49
  50 #include "osc_internal.h"
  51 #include <lnet/lnet_rdma.h>
  52
  53 atomic_t osc_pool_req_count;
  54 unsigned int osc_reqpool_maxreqcount;
  55 struct ptlrpc_request_pool *osc_rq_pool;
  56
  57 /* max memory used for request pool, unit is MB */
  58 static unsigned int osc_reqpool_mem_max = 5;
  59 module_param(osc_reqpool_mem_max, uint, 0444);
  60
  61 static int osc_idle_timeout = 20;
  62 module_param(osc_idle_timeout, uint, 0644);
  63
  64 #define osc_grant_args osc_brw_async_args
  65
  66 struct osc_setattr_args {
  67         struct obdo             *sa_oa;
  68         obd_enqueue_update_f     sa_upcall;
  69         void                    *sa_cookie;
  70 };
  71
  72 struct osc_fsync_args {
  73         struct osc_object       *fa_obj;
  74         struct obdo             *fa_oa;
  75         obd_enqueue_update_f    fa_upcall;
  76         void                    *fa_cookie;
  77 };
  78
  79 struct osc_ladvise_args {
  80         struct obdo             *la_oa;
  81         obd_enqueue_update_f     la_upcall;
  82         void                    *la_cookie;
  83 };
  84
  85 static void osc_release_ppga(struct brw_page **ppga, size_t count);
  86 static int brw_interpret(const struct lu_env *env, struct ptlrpc_request *req,
  87                          void *data, int rc);
  88
  89 void osc_pack_req_body(struct ptlrpc_request *req, struct obdo *oa)
  90 {
  91         struct ost_body *body;
  92
  93         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
  94         LASSERT(body);
  95
  96         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
  97 }
  98
  99 static int osc_getattr(const struct lu_env *env, struct obd_export *exp,
 100                        struct obdo *oa)
 101 {
 102         struct ptlrpc_request   *req;
 103         struct ost_body         *body;
 104         int                      rc;
 105
 106         ENTRY;
 107         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
 108         if (req == NULL)
 109                 RETURN(-ENOMEM);
 110
 111         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
 112         if (rc) {
 113                 ptlrpc_request_free(req);
 114                 RETURN(rc);
 115         }
 116
 117         osc_pack_req_body(req, oa);
 118
 119         ptlrpc_request_set_replen(req);
 120
 121         rc = ptlrpc_queue_wait(req);
 122         if (rc)
 123                 GOTO(out, rc);
 124
 125         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 126         if (body == NULL)
 127                 GOTO(out, rc = -EPROTO);
 128
 129         CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
 130         lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
 131
 132         oa->o_blksize = cli_brw_size(exp->exp_obd);
 133         oa->o_valid |= OBD_MD_FLBLKSZ;
 134
 135         EXIT;
 136 out:
 137         ptlrpc_req_finished(req);
 138
 139         return rc;
 140 }
 141
 142 static int osc_setattr(const struct lu_env *env, struct obd_export *exp,
 143                        struct obdo *oa)
 144 {
 145         struct ptlrpc_request   *req;
 146         struct ost_body         *body;
 147         int                      rc;
 148
 149         ENTRY;
 150         LASSERT(oa->o_valid & OBD_MD_FLGROUP);
 151
 152         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
 153         if (req == NULL)
 154                 RETURN(-ENOMEM);
 155
 156         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
 157         if (rc) {
 158                 ptlrpc_request_free(req);
 159                 RETURN(rc);
 160         }
 161
 162         osc_pack_req_body(req, oa);
 163
 164         ptlrpc_request_set_replen(req);
 165
 166         rc = ptlrpc_queue_wait(req);
 167         if (rc)
 168                 GOTO(out, rc);
 169
 170         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 171         if (body == NULL)
 172                 GOTO(out, rc = -EPROTO);
 173
 174         lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
 175
 176         EXIT;
 177 out:
 178         ptlrpc_req_finished(req);
 179
 180         RETURN(rc);
 181 }
 182
 183 static int osc_setattr_interpret(const struct lu_env *env,
 184                                  struct ptlrpc_request *req, void *args, int rc)
 185 {
 186         struct osc_setattr_args *sa = args;
 187         struct ost_body *body;
 188
 189         ENTRY;
 190
 191         if (rc != 0)
 192                 GOTO(out, rc);
 193
 194         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 195         if (body == NULL)
 196                 GOTO(out, rc = -EPROTO);
 197
 198         lustre_get_wire_obdo(&req->rq_import->imp_connect_data, sa->sa_oa,
 199                              &body->oa);
 200 out:
 201         rc = sa->sa_upcall(sa->sa_cookie, rc);
 202         RETURN(rc);
 203 }
 204
 205 int osc_setattr_async(struct obd_export *exp, struct obdo *oa,
 206                       obd_enqueue_update_f upcall, void *cookie,
 207                       struct ptlrpc_request_set *rqset)
 208 {
 209         struct ptlrpc_request   *req;
 210         struct osc_setattr_args *sa;
 211         int                      rc;
 212
 213         ENTRY;
 214
 215         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
 216         if (req == NULL)
 217                 RETURN(-ENOMEM);
 218
 219         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
 220         if (rc) {
 221                 ptlrpc_request_free(req);
 222                 RETURN(rc);
 223         }
 224
 225         osc_pack_req_body(req, oa);
 226
 227         ptlrpc_request_set_replen(req);
 228
 229         /* do mds to ost setattr asynchronously */
 230         if (!rqset) {
 231                 /* Do not wait for response. */
 232                 ptlrpcd_add_req(req);
 233         } else {
 234                 req->rq_interpret_reply = osc_setattr_interpret;
 235
 236                 sa = ptlrpc_req_async_args(sa, req);
 237                 sa->sa_oa = oa;
 238                 sa->sa_upcall = upcall;
 239                 sa->sa_cookie = cookie;
 240
 241                 ptlrpc_set_add_req(rqset, req);
 242         }
 243
 244         RETURN(0);
 245 }
 246
 247 static int osc_ladvise_interpret(const struct lu_env *env,
 248                                  struct ptlrpc_request *req,
 249                                  void *arg, int rc)
 250 {
 251         struct osc_ladvise_args *la = arg;
 252         struct ost_body *body;
 253         ENTRY;
 254
 255         if (rc != 0)
 256                 GOTO(out, rc);
 257
 258         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 259         if (body == NULL)
 260                 GOTO(out, rc = -EPROTO);
 261
 262         *la->la_oa = body->oa;
 263 out:
 264         rc = la->la_upcall(la->la_cookie, rc);
 265         RETURN(rc);
 266 }
 267
 268 /**
 269  * If rqset is NULL, do not wait for response. Upcall and cookie could also
 270  * be NULL in this case
 271  */
 272 int osc_ladvise_base(struct obd_export *exp, struct obdo *oa,
 273                      struct ladvise_hdr *ladvise_hdr,
 274                      obd_enqueue_update_f upcall, void *cookie,
 275                      struct ptlrpc_request_set *rqset)
 276 {
 277         struct ptlrpc_request   *req;
 278         struct ost_body         *body;
 279         struct osc_ladvise_args *la;
 280         int                      rc;
 281         struct lu_ladvise       *req_ladvise;
 282         struct lu_ladvise       *ladvise = ladvise_hdr->lah_advise;
 283         int                      num_advise = ladvise_hdr->lah_count;
 284         struct ladvise_hdr      *req_ladvise_hdr;
 285         ENTRY;
 286
 287         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_LADVISE);
 288         if (req == NULL)
 289                 RETURN(-ENOMEM);
 290
 291         req_capsule_set_size(&req->rq_pill, &RMF_OST_LADVISE, RCL_CLIENT,
 292                              num_advise * sizeof(*ladvise));
 293         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_LADVISE);
 294         if (rc != 0) {
 295                 ptlrpc_request_free(req);
 296                 RETURN(rc);
 297         }
 298         req->rq_request_portal = OST_IO_PORTAL;
 299         ptlrpc_at_set_req_timeout(req);
 300
 301         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 302         LASSERT(body);
 303         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa,
 304                              oa);
 305
 306         req_ladvise_hdr = req_capsule_client_get(&req->rq_pill,
 307                                                  &RMF_OST_LADVISE_HDR);
 308         memcpy(req_ladvise_hdr, ladvise_hdr, sizeof(*ladvise_hdr));
 309
 310         req_ladvise = req_capsule_client_get(&req->rq_pill, &RMF_OST_LADVISE);
 311         memcpy(req_ladvise, ladvise, sizeof(*ladvise) * num_advise);
 312         ptlrpc_request_set_replen(req);
 313
 314         if (rqset == NULL) {
 315                 /* Do not wait for response. */
 316                 ptlrpcd_add_req(req);
 317                 RETURN(0);
 318         }
 319
 320         req->rq_interpret_reply = osc_ladvise_interpret;
 321         la = ptlrpc_req_async_args(la, req);
 322         la->la_oa = oa;
 323         la->la_upcall = upcall;
 324         la->la_cookie = cookie;
 325
 326         ptlrpc_set_add_req(rqset, req);
 327
 328         RETURN(0);
 329 }
 330
 331 static int osc_create(const struct lu_env *env, struct obd_export *exp,
 332                       struct obdo *oa)
 333 {
 334         struct ptlrpc_request *req;
 335         struct ost_body       *body;
 336         int                    rc;
 337         ENTRY;
 338
 339         LASSERT(oa != NULL);
 340         LASSERT(oa->o_valid & OBD_MD_FLGROUP);
 341         LASSERT(fid_seq_is_echo(ostid_seq(&oa->o_oi)));
 342
 343         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
 344         if (req == NULL)
 345                 GOTO(out, rc = -ENOMEM);
 346
 347         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
 348         if (rc) {
 349                 ptlrpc_request_free(req);
 350                 GOTO(out, rc);
 351         }
 352
 353         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 354         LASSERT(body);
 355
 356         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
 357
 358         ptlrpc_request_set_replen(req);
 359
 360         rc = ptlrpc_queue_wait(req);
 361         if (rc)
 362                 GOTO(out_req, rc);
 363
 364         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 365         if (body == NULL)
 366                 GOTO(out_req, rc = -EPROTO);
 367
 368         CDEBUG(D_INFO, "oa flags %x\n", oa->o_flags);
 369         lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
 370
 371         oa->o_blksize = cli_brw_size(exp->exp_obd);
 372         oa->o_valid |= OBD_MD_FLBLKSZ;
 373
 374         CDEBUG(D_HA, "transno: %lld\n",
 375                lustre_msg_get_transno(req->rq_repmsg));
 376 out_req:
 377         ptlrpc_req_finished(req);
 378 out:
 379         RETURN(rc);
 380 }
 381
 382 int osc_punch_send(struct obd_export *exp, struct obdo *oa,
 383                    obd_enqueue_update_f upcall, void *cookie)
 384 {
 385         struct ptlrpc_request *req;
 386         struct osc_setattr_args *sa;
 387         struct obd_import *imp = class_exp2cliimp(exp);
 388         struct ost_body *body;
 389         int rc;
 390
 391         ENTRY;
 392
 393         req = ptlrpc_request_alloc(imp, &RQF_OST_PUNCH);
 394         if (req == NULL)
 395                 RETURN(-ENOMEM);
 396
 397         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
 398         if (rc < 0) {
 399                 ptlrpc_request_free(req);
 400                 RETURN(rc);
 401         }
 402
 403         osc_set_io_portal(req);
 404
 405         ptlrpc_at_set_req_timeout(req);
 406
 407         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 408
 409         lustre_set_wire_obdo(&imp->imp_connect_data, &body->oa, oa);
 410
 411         ptlrpc_request_set_replen(req);
 412
 413         req->rq_interpret_reply = osc_setattr_interpret;
 414         sa = ptlrpc_req_async_args(sa, req);
 415         sa->sa_oa = oa;
 416         sa->sa_upcall = upcall;
 417         sa->sa_cookie = cookie;
 418
 419         ptlrpcd_add_req(req);
 420
 421         RETURN(0);
 422 }
 423 EXPORT_SYMBOL(osc_punch_send);
 424
 425 /**
 426  * osc_fallocate_base() - Handles fallocate request.
 427  *
 428  * @exp:        Export structure
 429  * @oa:         Attributes passed to OSS from client (obdo structure)
 430  * @upcall:     Primary & supplementary group information
 431  * @cookie:     Exclusive identifier
 432  * @rqset:      Request list.
 433  * @mode:       Operation done on given range.
 434  *
 435  * osc_fallocate_base() - Handles fallocate requests only. Only block
 436  * allocation or standard preallocate operation is supported currently.
 437  * Other mode flags is not supported yet. ftruncate(2) or truncate(2)
 438  * is supported via SETATTR request.
 439  *
 440  * Return: Non-zero on failure and O on success.
 441  */
 442 int osc_fallocate_base(struct obd_export *exp, struct obdo *oa,
 443                        obd_enqueue_update_f upcall, void *cookie, int mode)
 444 {
 445         struct ptlrpc_request *req;
 446         struct osc_setattr_args *sa;
 447         struct ost_body *body;
 448         struct obd_import *imp = class_exp2cliimp(exp);
 449         int rc;
 450         ENTRY;
 451
 452         oa->o_falloc_mode = mode;
 453         req = ptlrpc_request_alloc(class_exp2cliimp(exp),
 454                                    &RQF_OST_FALLOCATE);
 455         if (req == NULL)
 456                 RETURN(-ENOMEM);
 457
 458         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_FALLOCATE);
 459         if (rc != 0) {
 460                 ptlrpc_request_free(req);
 461                 RETURN(rc);
 462         }
 463
 464         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 465         LASSERT(body);
 466
 467         lustre_set_wire_obdo(&imp->imp_connect_data, &body->oa, oa);
 468
 469         ptlrpc_request_set_replen(req);
 470
 471         req->rq_interpret_reply = osc_setattr_interpret;
 472         BUILD_BUG_ON(sizeof(*sa) > sizeof(req->rq_async_args));
 473         sa = ptlrpc_req_async_args(sa, req);
 474         sa->sa_oa = oa;
 475         sa->sa_upcall = upcall;
 476         sa->sa_cookie = cookie;
 477
 478         ptlrpcd_add_req(req);
 479
 480         RETURN(0);
 481 }
 482 EXPORT_SYMBOL(osc_fallocate_base);
 483
 484 static int osc_sync_interpret(const struct lu_env *env,
 485                               struct ptlrpc_request *req, void *args, int rc)
 486 {
 487         struct osc_fsync_args *fa = args;
 488         struct ost_body *body;
 489         struct cl_attr *attr = &osc_env_info(env)->oti_attr;
 490         unsigned long valid = 0;
 491         struct cl_object *obj;
 492         ENTRY;
 493
 494         if (rc != 0)
 495                 GOTO(out, rc);
 496
 497         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 498         if (body == NULL) {
 499                 CERROR("can't unpack ost_body\n");
 500                 GOTO(out, rc = -EPROTO);
 501         }
 502
 503         *fa->fa_oa = body->oa;
 504         obj = osc2cl(fa->fa_obj);
 505
 506         /* Update osc object's blocks attribute */
 507         cl_object_attr_lock(obj);
 508         if (body->oa.o_valid & OBD_MD_FLBLOCKS) {
 509                 attr->cat_blocks = body->oa.o_blocks;
 510                 valid |= CAT_BLOCKS;
 511         }
 512
 513         if (valid != 0)
 514                 cl_object_attr_update(env, obj, attr, valid);
 515         cl_object_attr_unlock(obj);
 516
 517 out:
 518         rc = fa->fa_upcall(fa->fa_cookie, rc);
 519         RETURN(rc);
 520 }
 521
 522 int osc_sync_base(struct osc_object *obj, struct obdo *oa,
 523                   obd_enqueue_update_f upcall, void *cookie,
 524                   struct ptlrpc_request_set *rqset)
 525 {
 526         struct obd_export     *exp = osc_export(obj);
 527         struct ptlrpc_request *req;
 528         struct ost_body       *body;
 529         struct osc_fsync_args *fa;
 530         int                    rc;
 531         ENTRY;
 532
 533         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
 534         if (req == NULL)
 535                 RETURN(-ENOMEM);
 536
 537         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
 538         if (rc) {
 539                 ptlrpc_request_free(req);
 540                 RETURN(rc);
 541         }
 542
 543         /* overload the size and blocks fields in the oa with start/end */
 544         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 545         LASSERT(body);
 546         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
 547
 548         ptlrpc_request_set_replen(req);
 549         req->rq_interpret_reply = osc_sync_interpret;
 550
 551         fa = ptlrpc_req_async_args(fa, req);
 552         fa->fa_obj = obj;
 553         fa->fa_oa = oa;
 554         fa->fa_upcall = upcall;
 555         fa->fa_cookie = cookie;
 556
 557         ptlrpc_set_add_req(rqset, req);
 558
 559         RETURN (0);
 560 }
 561
 562 /* Find and cancel locally locks matched by @mode in the resource found by
 563  * @objid. Found locks are added into @cancel list. Returns the amount of
 564  * locks added to @cancels list. */
 565 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
 566                                    struct list_head *cancels,
 567                                    enum ldlm_mode mode, __u64 lock_flags)
 568 {
 569         struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
 570         struct ldlm_res_id res_id;
 571         struct ldlm_resource *res;
 572         int count;
 573         ENTRY;
 574
 575         /* Return, i.e. cancel nothing, only if ELC is supported (flag in
 576          * export) but disabled through procfs (flag in NS).
 577          *
 578          * This distinguishes from a case when ELC is not supported originally,
 579          * when we still want to cancel locks in advance and just cancel them
 580          * locally, without sending any RPC. */
 581         if (exp_connect_cancelset(exp) && !ns_connect_cancelset(ns))
 582                 RETURN(0);
 583
 584         ostid_build_res_name(&oa->o_oi, &res_id);
 585         res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
 586         if (IS_ERR(res))
 587                 RETURN(0);
 588
 589         LDLM_RESOURCE_ADDREF(res);
 590         count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
 591                                            lock_flags, 0, NULL);
 592         LDLM_RESOURCE_DELREF(res);
 593         ldlm_resource_putref(res);
 594         RETURN(count);
 595 }
 596
 597 static int osc_destroy_interpret(const struct lu_env *env,
 598                                  struct ptlrpc_request *req, void *args, int rc)
 599 {
 600         struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
 601
 602         atomic_dec(&cli->cl_destroy_in_flight);
 603         wake_up(&cli->cl_destroy_waitq);
 604
 605         return 0;
 606 }
 607
 608 static int osc_can_send_destroy(struct client_obd *cli)
 609 {
 610         if (atomic_inc_return(&cli->cl_destroy_in_flight) <=
 611             cli->cl_max_rpcs_in_flight) {
 612                 /* The destroy request can be sent */
 613                 return 1;
 614         }
 615         if (atomic_dec_return(&cli->cl_destroy_in_flight) <
 616             cli->cl_max_rpcs_in_flight) {
 617                 /*
 618                  * The counter has been modified between the two atomic
 619                  * operations.
 620                  */
 621                 wake_up(&cli->cl_destroy_waitq);
 622         }
 623         return 0;
 624 }
 625
 626 static int osc_destroy(const struct lu_env *env, struct obd_export *exp,
 627                        struct obdo *oa)
 628 {
 629         struct client_obd     *cli = &exp->exp_obd->u.cli;
 630         struct ptlrpc_request *req;
 631         struct ost_body       *body;
 632         LIST_HEAD(cancels);
 633         int rc, count;
 634         ENTRY;
 635
 636         if (!oa) {
 637                 CDEBUG(D_INFO, "oa NULL\n");
 638                 RETURN(-EINVAL);
 639         }
 640
 641         count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
 642                                         LDLM_FL_DISCARD_DATA);
 643
 644         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
 645         if (req == NULL) {
 646                 ldlm_lock_list_put(&cancels, l_bl_ast, count);
 647                 RETURN(-ENOMEM);
 648         }
 649
 650         rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
 651                                0, &cancels, count);
 652         if (rc) {
 653                 ptlrpc_request_free(req);
 654                 RETURN(rc);
 655         }
 656
 657         req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
 658         ptlrpc_at_set_req_timeout(req);
 659
 660         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
 661         LASSERT(body);
 662         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
 663
 664         ptlrpc_request_set_replen(req);
 665
 666         req->rq_interpret_reply = osc_destroy_interpret;
 667         if (!osc_can_send_destroy(cli)) {
 668                 /*
 669                  * Wait until the number of on-going destroy RPCs drops
 670                  * under max_rpc_in_flight
 671                  */
 672                 rc = l_wait_event_abortable_exclusive(
 673                         cli->cl_destroy_waitq,
 674                         osc_can_send_destroy(cli));
 675                 if (rc) {
 676                         ptlrpc_req_finished(req);
 677                         RETURN(-EINTR);
 678                 }
 679         }
 680
 681         /* Do not wait for response */
 682         ptlrpcd_add_req(req);
 683         RETURN(0);
 684 }
 685
 686 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
 687                                 long writing_bytes)
 688 {
 689         u64 bits = OBD_MD_FLBLOCKS | OBD_MD_FLGRANT;
 690
 691         LASSERT(!(oa->o_valid & bits));
 692
 693         oa->o_valid |= bits;
 694         spin_lock(&cli->cl_loi_list_lock);
 695         if (cli->cl_ocd_grant_param)
 696                 oa->o_dirty = cli->cl_dirty_grant;
 697         else
 698                 oa->o_dirty = cli->cl_dirty_pages << PAGE_SHIFT;
 699         if (unlikely(cli->cl_dirty_pages > cli->cl_dirty_max_pages)) {
 700                 CERROR("dirty %lu > dirty_max %lu\n",
 701                        cli->cl_dirty_pages,
 702                        cli->cl_dirty_max_pages);
 703                 oa->o_undirty = 0;
 704         } else if (unlikely(atomic_long_read(&obd_dirty_pages) >
 705                             (long)(obd_max_dirty_pages + 1))) {
 706                 /* The atomic_read() allowing the atomic_inc() are
 707                  * not covered by a lock thus they may safely race and trip
 708                  * this CERROR() unless we add in a small fudge factor (+1). */
 709                 CERROR("%s: dirty %ld > system dirty_max %ld\n",
 710                        cli_name(cli), atomic_long_read(&obd_dirty_pages),
 711                        obd_max_dirty_pages);
 712                 oa->o_undirty = 0;
 713         } else if (unlikely(cli->cl_dirty_max_pages - cli->cl_dirty_pages >
 714                             0x7fffffff)) {
 715                 CERROR("dirty %lu - dirty_max %lu too big???\n",
 716                        cli->cl_dirty_pages, cli->cl_dirty_max_pages);
 717                 oa->o_undirty = 0;
 718         } else {
 719                 unsigned long nrpages;
 720                 unsigned long undirty;
 721
 722                 nrpages = cli->cl_max_pages_per_rpc;
 723                 nrpages *= cli->cl_max_rpcs_in_flight + 1;
 724                 nrpages = max(nrpages, cli->cl_dirty_max_pages);
 725                 undirty = nrpages << PAGE_SHIFT;
 726                 if (cli->cl_ocd_grant_param) {
 727                         int nrextents;
 728
 729                         /* take extent tax into account when asking for more
 730                          * grant space */
 731                         nrextents = (nrpages + cli->cl_max_extent_pages - 1) /
 732                                      cli->cl_max_extent_pages;
 733                         undirty += nrextents * cli->cl_grant_extent_tax;
 734                 }
 735                 /* Do not ask for more than OBD_MAX_GRANT - a margin for server
 736                  * to add extent tax, etc.
 737                  */
 738                 oa->o_undirty = min(undirty, OBD_MAX_GRANT &
 739                                     ~(PTLRPC_MAX_BRW_SIZE * 4UL));
 740         }
 741         oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
 742         /* o_dropped AKA o_misc is 32 bits, but cl_lost_grant is 64 bits */
 743         if (cli->cl_lost_grant > INT_MAX) {
 744                 CDEBUG(D_CACHE,
 745                       "%s: avoided o_dropped overflow: cl_lost_grant %lu\n",
 746                       cli_name(cli), cli->cl_lost_grant);
 747                 oa->o_dropped = INT_MAX;
 748         } else {
 749                 oa->o_dropped = cli->cl_lost_grant;
 750         }
 751         cli->cl_lost_grant -= oa->o_dropped;
 752         spin_unlock(&cli->cl_loi_list_lock);
 753         CDEBUG(D_CACHE, "%s: dirty: %llu undirty: %u dropped %u grant: %llu"
 754                " cl_lost_grant %lu\n", cli_name(cli), oa->o_dirty,
 755                oa->o_undirty, oa->o_dropped, oa->o_grant, cli->cl_lost_grant);
 756 }
 757
 758 void osc_update_next_shrink(struct client_obd *cli)
 759 {
 760         cli->cl_next_shrink_grant = ktime_get_seconds() +
 761                                     cli->cl_grant_shrink_interval;
 762
 763         CDEBUG(D_CACHE, "next time %lld to shrink grant\n",
 764                cli->cl_next_shrink_grant);
 765 }
 766 EXPORT_SYMBOL(osc_update_next_shrink);
 767
 768 static void __osc_update_grant(struct client_obd *cli, u64 grant)
 769 {
 770         spin_lock(&cli->cl_loi_list_lock);
 771         cli->cl_avail_grant += grant;
 772         spin_unlock(&cli->cl_loi_list_lock);
 773 }
 774
 775 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
 776 {
 777         if (body->oa.o_valid & OBD_MD_FLGRANT) {
 778                 CDEBUG(D_CACHE, "got %llu extra grant\n", body->oa.o_grant);
 779                 __osc_update_grant(cli, body->oa.o_grant);
 780         }
 781 }
 782
 783 /**
 784  * grant thread data for shrinking space.
 785  */
 786 struct grant_thread_data {
 787         struct list_head        gtd_clients;
 788         struct mutex            gtd_mutex;
 789         unsigned long           gtd_stopped:1;
 790 };
 791 static struct grant_thread_data client_gtd;
 792
 793 static int osc_shrink_grant_interpret(const struct lu_env *env,
 794                                       struct ptlrpc_request *req,
 795                                       void *args, int rc)
 796 {
 797         struct osc_grant_args *aa = args;
 798         struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
 799         struct ost_body *body;
 800
 801         if (rc != 0) {
 802                 __osc_update_grant(cli, aa->aa_oa->o_grant);
 803                 GOTO(out, rc);
 804         }
 805
 806         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
 807         LASSERT(body);
 808         osc_update_grant(cli, body);
 809 out:
 810         OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
 811         aa->aa_oa = NULL;
 812
 813         return rc;
 814 }
 815
 816 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
 817 {
 818         spin_lock(&cli->cl_loi_list_lock);
 819         oa->o_grant = cli->cl_avail_grant / 4;
 820         cli->cl_avail_grant -= oa->o_grant;
 821         spin_unlock(&cli->cl_loi_list_lock);
 822         if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
 823                 oa->o_valid |= OBD_MD_FLFLAGS;
 824                 oa->o_flags = 0;
 825         }
 826         oa->o_flags |= OBD_FL_SHRINK_GRANT;
 827         osc_update_next_shrink(cli);
 828 }
 829
 830 /* Shrink the current grant, either from some large amount to enough for a
 831  * full set of in-flight RPCs, or if we have already shrunk to that limit
 832  * then to enough for a single RPC.  This avoids keeping more grant than
 833  * needed, and avoids shrinking the grant piecemeal. */
 834 static int osc_shrink_grant(struct client_obd *cli)
 835 {
 836         __u64 target_bytes = (cli->cl_max_rpcs_in_flight + 1) *
 837                              (cli->cl_max_pages_per_rpc << PAGE_SHIFT);
 838
 839         spin_lock(&cli->cl_loi_list_lock);
 840         if (cli->cl_avail_grant <= target_bytes)
 841                 target_bytes = cli->cl_max_pages_per_rpc << PAGE_SHIFT;
 842         spin_unlock(&cli->cl_loi_list_lock);
 843
 844         return osc_shrink_grant_to_target(cli, target_bytes);
 845 }
 846
 847 int osc_shrink_grant_to_target(struct client_obd *cli, __u64 target_bytes)
 848 {
 849         int                     rc = 0;
 850         struct ost_body        *body;
 851         ENTRY;
 852
 853         spin_lock(&cli->cl_loi_list_lock);
 854         /* Don't shrink if we are already above or below the desired limit
 855          * We don't want to shrink below a single RPC, as that will negatively
 856          * impact block allocation and long-term performance. */
 857         if (target_bytes < cli->cl_max_pages_per_rpc << PAGE_SHIFT)
 858                 target_bytes = cli->cl_max_pages_per_rpc << PAGE_SHIFT;
 859
 860         if (target_bytes >= cli->cl_avail_grant) {
 861                 spin_unlock(&cli->cl_loi_list_lock);
 862                 RETURN(0);
 863         }
 864         spin_unlock(&cli->cl_loi_list_lock);
 865
 866         OBD_ALLOC_PTR(body);
 867         if (!body)
 868                 RETURN(-ENOMEM);
 869
 870         osc_announce_cached(cli, &body->oa, 0);
 871
 872         spin_lock(&cli->cl_loi_list_lock);
 873         if (target_bytes >= cli->cl_avail_grant) {
 874                 /* available grant has changed since target calculation */
 875                 spin_unlock(&cli->cl_loi_list_lock);
 876                 GOTO(out_free, rc = 0);
 877         }
 878         body->oa.o_grant = cli->cl_avail_grant - target_bytes;
 879         cli->cl_avail_grant = target_bytes;
 880         spin_unlock(&cli->cl_loi_list_lock);
 881         if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
 882                 body->oa.o_valid |= OBD_MD_FLFLAGS;
 883                 body->oa.o_flags = 0;
 884         }
 885         body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
 886         osc_update_next_shrink(cli);
 887
 888         rc = osc_set_info_async(NULL, cli->cl_import->imp_obd->obd_self_export,
 889                                 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
 890                                 sizeof(*body), body, NULL);
 891         if (rc != 0)
 892                 __osc_update_grant(cli, body->oa.o_grant);
 893 out_free:
 894         OBD_FREE_PTR(body);
 895         RETURN(rc);
 896 }
 897
 898 static int osc_should_shrink_grant(struct client_obd *client)
 899 {
 900         time64_t next_shrink = client->cl_next_shrink_grant;
 901
 902         if (client->cl_import == NULL)
 903                 return 0;
 904
 905         if (!OCD_HAS_FLAG(&client->cl_import->imp_connect_data, GRANT_SHRINK) ||
 906             client->cl_import->imp_grant_shrink_disabled) {
 907                 osc_update_next_shrink(client);
 908                 return 0;
 909         }
 910
 911         if (ktime_get_seconds() >= next_shrink - 5) {
 912                 /* Get the current RPC size directly, instead of going via:
 913                  * cli_brw_size(obd->u.cli.cl_import->imp_obd->obd_self_export)
 914                  * Keep comment here so that it can be found by searching. */
 915                 int brw_size = client->cl_max_pages_per_rpc << PAGE_SHIFT;
 916
 917                 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
 918                     client->cl_avail_grant > brw_size)
 919                         return 1;
 920                 else
 921                         osc_update_next_shrink(client);
 922         }
 923         return 0;
 924 }
 925
 926 #define GRANT_SHRINK_RPC_BATCH  100
 927
 928 static struct delayed_work work;
 929
 930 static void osc_grant_work_handler(struct work_struct *data)
 931 {
 932         struct client_obd *cli;
 933         int rpc_sent;
 934         bool init_next_shrink = true;
 935         time64_t next_shrink = ktime_get_seconds() + GRANT_SHRINK_INTERVAL;
 936
 937         rpc_sent = 0;
 938         mutex_lock(&client_gtd.gtd_mutex);
 939         list_for_each_entry(cli, &client_gtd.gtd_clients,
 940                             cl_grant_chain) {
 941                 if (rpc_sent < GRANT_SHRINK_RPC_BATCH &&
 942                     osc_should_shrink_grant(cli)) {
 943                         osc_shrink_grant(cli);
 944                         rpc_sent++;
 945                 }
 946
 947                 if (!init_next_shrink) {
 948                         if (cli->cl_next_shrink_grant < next_shrink &&
 949                             cli->cl_next_shrink_grant > ktime_get_seconds())
 950                                 next_shrink = cli->cl_next_shrink_grant;
 951                 } else {
 952                         init_next_shrink = false;
 953                         next_shrink = cli->cl_next_shrink_grant;
 954                 }
 955         }
 956         mutex_unlock(&client_gtd.gtd_mutex);
 957
 958         if (client_gtd.gtd_stopped == 1)
 959                 return;
 960
 961         if (next_shrink > ktime_get_seconds()) {
 962                 time64_t delay = next_shrink - ktime_get_seconds();
 963
 964                 schedule_delayed_work(&work, cfs_time_seconds(delay));
 965         } else {
 966                 schedule_work(&work.work);
 967         }
 968 }
 969
 970 void osc_schedule_grant_work(void)
 971 {
 972         cancel_delayed_work_sync(&work);
 973         schedule_work(&work.work);
 974 }
 975 EXPORT_SYMBOL(osc_schedule_grant_work);
 976
 977 /**
 978  * Start grant thread for returing grant to server for idle clients.
 979  */
 980 static int osc_start_grant_work(void)
 981 {
 982         client_gtd.gtd_stopped = 0;
 983         mutex_init(&client_gtd.gtd_mutex);
 984         INIT_LIST_HEAD(&client_gtd.gtd_clients);
 985
 986         INIT_DELAYED_WORK(&work, osc_grant_work_handler);
 987         schedule_work(&work.work);
 988
 989         return 0;
 990 }
 991
 992 static void osc_stop_grant_work(void)
 993 {
 994         client_gtd.gtd_stopped = 1;
 995         cancel_delayed_work_sync(&work);
 996 }
 997
 998 static void osc_add_grant_list(struct client_obd *client)
 999 {
1000         mutex_lock(&client_gtd.gtd_mutex);
1001         list_add(&client->cl_grant_chain, &client_gtd.gtd_clients);
1002         mutex_unlock(&client_gtd.gtd_mutex);
1003 }
1004
1005 static void osc_del_grant_list(struct client_obd *client)
1006 {
1007         if (list_empty(&client->cl_grant_chain))
1008                 return;
1009
1010         mutex_lock(&client_gtd.gtd_mutex);
1011         list_del_init(&client->cl_grant_chain);
1012         mutex_unlock(&client_gtd.gtd_mutex);
1013 }
1014
1015 void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1016 {
1017         /*
1018          * ocd_grant is the total grant amount we're expect to hold: if we've
1019          * been evicted, it's the new avail_grant amount, cl_dirty_pages will
1020          * drop to 0 as inflight RPCs fail out; otherwise, it's avail_grant +
1021          * dirty.
1022          *
1023          * race is tolerable here: if we're evicted, but imp_state already
1024          * left EVICTED state, then cl_dirty_pages must be 0 already.
1025          */
1026         spin_lock(&cli->cl_loi_list_lock);
1027         cli->cl_avail_grant = ocd->ocd_grant;
1028         if (cli->cl_import->imp_state != LUSTRE_IMP_EVICTED) {
1029                 unsigned long consumed = cli->cl_reserved_grant;
1030
1031                 if (OCD_HAS_FLAG(ocd, GRANT_PARAM))
1032                         consumed += cli->cl_dirty_grant;
1033                 else
1034                         consumed += cli->cl_dirty_pages << PAGE_SHIFT;
1035                 if (cli->cl_avail_grant < consumed) {
1036                         CERROR("%s: granted %ld but already consumed %ld\n",
1037                                cli_name(cli), cli->cl_avail_grant, consumed);
1038                         cli->cl_avail_grant = 0;
1039                 } else {
1040                         cli->cl_avail_grant -= consumed;
1041                 }
1042         }
1043
1044         if (OCD_HAS_FLAG(ocd, GRANT_PARAM)) {
1045                 u64 size;
1046                 int chunk_mask;
1047
1048                 /* overhead for each extent insertion */
1049                 cli->cl_grant_extent_tax = ocd->ocd_grant_tax_kb << 10;
1050                 /* determine the appropriate chunk size used by osc_extent. */
1051                 cli->cl_chunkbits = max_t(int, PAGE_SHIFT,
1052                                           ocd->ocd_grant_blkbits);
1053                 /* max_pages_per_rpc must be chunk aligned */
1054                 chunk_mask = ~((1 << (cli->cl_chunkbits - PAGE_SHIFT)) - 1);
1055                 cli->cl_max_pages_per_rpc = (cli->cl_max_pages_per_rpc +
1056                                              ~chunk_mask) & chunk_mask;
1057                 /* determine maximum extent size, in #pages */
1058                 size = (u64)ocd->ocd_grant_max_blks << ocd->ocd_grant_blkbits;
1059                 cli->cl_max_extent_pages = (size >> PAGE_SHIFT) ?: 1;
1060                 cli->cl_ocd_grant_param = 1;
1061         } else {
1062                 cli->cl_ocd_grant_param = 0;
1063                 cli->cl_grant_extent_tax = 0;
1064                 cli->cl_chunkbits = PAGE_SHIFT;
1065                 cli->cl_max_extent_pages = DT_MAX_BRW_PAGES;
1066         }
1067         spin_unlock(&cli->cl_loi_list_lock);
1068
1069         CDEBUG(D_CACHE,
1070                "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld. chunk bits: %d cl_max_extent_pages: %d\n",
1071                cli_name(cli),
1072                cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits,
1073                cli->cl_max_extent_pages);
1074
1075         if (OCD_HAS_FLAG(ocd, GRANT_SHRINK) && list_empty(&cli->cl_grant_chain))
1076                 osc_add_grant_list(cli);
1077 }
1078 EXPORT_SYMBOL(osc_init_grant);
1079
1080 /* We assume that the reason this OSC got a short read is because it read
1081  * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1082  * via the LOV, and it _knows_ it's reading inside the file, it's just that
1083  * this stripe never got written at or beyond this stripe offset yet. */
1084 static void handle_short_read(int nob_read, size_t page_count,
1085                               struct brw_page **pga)
1086 {
1087         char *ptr;
1088         int i = 0;
1089
1090         /* skip bytes read OK */
1091         while (nob_read > 0) {
1092                 LASSERT (page_count > 0);
1093
1094                 if (pga[i]->count > nob_read) {
1095                         /* EOF inside this page */
1096                         ptr = kmap(pga[i]->pg) +
1097                                 (pga[i]->off & ~PAGE_MASK);
1098                         memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1099                         kunmap(pga[i]->pg);
1100                         page_count--;
1101                         i++;
1102                         break;
1103                 }
1104
1105                 nob_read -= pga[i]->count;
1106                 page_count--;
1107                 i++;
1108         }
1109
1110         /* zero remaining pages */
1111         while (page_count-- > 0) {
1112                 ptr = kmap(pga[i]->pg) + (pga[i]->off & ~PAGE_MASK);
1113                 memset(ptr, 0, pga[i]->count);
1114                 kunmap(pga[i]->pg);
1115                 i++;
1116         }
1117 }
1118
1119 static int check_write_rcs(struct ptlrpc_request *req,
1120                            int requested_nob, int niocount,
1121                            size_t page_count, struct brw_page **pga)
1122 {
1123         int     i;
1124         __u32   *remote_rcs;
1125
1126         remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1127                                                   sizeof(*remote_rcs) *
1128                                                   niocount);
1129         if (remote_rcs == NULL) {
1130                 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1131                 return(-EPROTO);
1132         }
1133
1134         /* return error if any niobuf was in error */
1135         for (i = 0; i < niocount; i++) {
1136                 if ((int)remote_rcs[i] < 0) {
1137                         CDEBUG(D_INFO, "rc[%d]: %d req %p\n",
1138                                i, remote_rcs[i], req);
1139                         return remote_rcs[i];
1140                 }
1141
1142                 if (remote_rcs[i] != 0) {
1143                         CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1144                                 i, remote_rcs[i], req);
1145                         return(-EPROTO);
1146                 }
1147         }
1148         if (req->rq_bulk != NULL &&
1149             req->rq_bulk->bd_nob_transferred != requested_nob) {
1150                 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1151                        req->rq_bulk->bd_nob_transferred, requested_nob);
1152                 return(-EPROTO);
1153         }
1154
1155         return (0);
1156 }
1157
1158 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1159 {
1160         if (p1->flag != p2->flag) {
1161                 unsigned mask = ~(OBD_BRW_FROM_GRANT | OBD_BRW_NOCACHE |
1162                                   OBD_BRW_SYNC       | OBD_BRW_ASYNC   |
1163                                   OBD_BRW_NOQUOTA    | OBD_BRW_SOFT_SYNC |
1164                                   OBD_BRW_SYS_RESOURCE);
1165
1166                 /* warn if we try to combine flags that we don't know to be
1167                  * safe to combine */
1168                 if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
1169                         CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1170                               "report this at https://jira.whamcloud.com/\n",
1171                               p1->flag, p2->flag);
1172                 }
1173                 return 0;
1174         }
1175
1176         return (p1->off + p1->count == p2->off);
1177 }
1178
1179 #if IS_ENABLED(CONFIG_CRC_T10DIF)
1180 static int osc_checksum_bulk_t10pi(const char *obd_name, int nob,
1181                                    size_t pg_count, struct brw_page **pga,
1182                                    int opc, obd_dif_csum_fn *fn,
1183                                    int sector_size,
1184                                    u32 *check_sum, bool resend)
1185 {
1186         struct ahash_request *req;
1187         /* Used Adler as the default checksum type on top of DIF tags */
1188         unsigned char cfs_alg = cksum_obd2cfs(OBD_CKSUM_T10_TOP);
1189         struct page *__page;
1190         unsigned char *buffer;
1191         __u16 *guard_start;
1192         unsigned int bufsize;
1193         int guard_number;
1194         int used_number = 0;
1195         int used;
1196         u32 cksum;
1197         int rc = 0;
1198         int i = 0;
1199
1200         LASSERT(pg_count > 0);
1201
1202         __page = alloc_page(GFP_KERNEL);
1203         if (__page == NULL)
1204                 return -ENOMEM;
1205
1206         req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1207         if (IS_ERR(req)) {
1208                 rc = PTR_ERR(req);
1209                 CERROR("%s: unable to initialize checksum hash %s: rc = %d\n",
1210                        obd_name, cfs_crypto_hash_name(cfs_alg), rc);
1211                 GOTO(out, rc);
1212         }
1213
1214         buffer = kmap(__page);
1215         guard_start = (__u16 *)buffer;
1216         guard_number = PAGE_SIZE / sizeof(*guard_start);
1217         CDEBUG(D_PAGE | (resend ? D_HA : 0),
1218                "GRD tags per page=%u, resend=%u, bytes=%u, pages=%zu\n",
1219                guard_number, resend, nob, pg_count);
1220
1221         while (nob > 0 && pg_count > 0) {
1222                 unsigned int count = pga[i]->count > nob ? nob : pga[i]->count;
1223
1224                 /* corrupt the data before we compute the checksum, to
1225                  * simulate an OST->client data error */
1226                 if (unlikely(i == 0 && opc == OST_READ &&
1227                              OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE))) {
1228                         unsigned char *ptr = kmap(pga[i]->pg);
1229                         int off = pga[i]->off & ~PAGE_MASK;
1230
1231                         memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
1232                         kunmap(pga[i]->pg);
1233                 }
1234
1235                 /*
1236                  * The left guard number should be able to hold checksums of a
1237                  * whole page
1238                  */
1239                 rc = obd_page_dif_generate_buffer(obd_name, pga[i]->pg,
1240                                                   pga[i]->off & ~PAGE_MASK,
1241                                                   count,
1242                                                   guard_start + used_number,
1243                                                   guard_number - used_number,
1244                                                   &used, sector_size,
1245                                                   fn);
1246                 if (unlikely(resend))
1247                         CDEBUG(D_PAGE | D_HA,
1248                                "pga[%u]: used %u off %llu+%u gen checksum: %*phN\n",
1249                                i, used, pga[i]->off & ~PAGE_MASK, count,
1250                                (int)(used * sizeof(*guard_start)),
1251                                guard_start + used_number);
1252                 if (rc)
1253                         break;
1254
1255                 used_number += used;
1256                 if (used_number == guard_number) {
1257                         cfs_crypto_hash_update_page(req, __page, 0,
1258                                 used_number * sizeof(*guard_start));
1259                         used_number = 0;
1260                 }
1261
1262                 nob -= pga[i]->count;
1263                 pg_count--;
1264                 i++;
1265         }
1266         kunmap(__page);
1267         if (rc)
1268                 GOTO(out, rc);
1269
1270         if (used_number != 0)
1271                 cfs_crypto_hash_update_page(req, __page, 0,
1272                         used_number * sizeof(*guard_start));
1273
1274         bufsize = sizeof(cksum);
1275         cfs_crypto_hash_final(req, (unsigned char *)&cksum, &bufsize);
1276
1277         /* For sending we only compute the wrong checksum instead
1278          * of corrupting the data so it is still correct on a redo */
1279         if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1280                 cksum++;
1281
1282         *check_sum = cksum;
1283 out:
1284         __free_page(__page);
1285         return rc;
1286 }
1287 #else /* !CONFIG_CRC_T10DIF */
1288 #define obd_dif_ip_fn NULL
1289 #define obd_dif_crc_fn NULL
1290 #define osc_checksum_bulk_t10pi(name, nob, pgc, pga, opc, fn, ssize, csum, re) \
1291         -EOPNOTSUPP
1292 #endif /* CONFIG_CRC_T10DIF */
1293
1294 static int osc_checksum_bulk(int nob, size_t pg_count,
1295                              struct brw_page **pga, int opc,
1296                              enum cksum_types cksum_type,
1297                              u32 *cksum)
1298 {
1299         int                             i = 0;
1300         struct ahash_request           *req;
1301         unsigned int                    bufsize;
1302         unsigned char                   cfs_alg = cksum_obd2cfs(cksum_type);
1303
1304         LASSERT(pg_count > 0);
1305
1306         req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1307         if (IS_ERR(req)) {
1308                 CERROR("Unable to initialize checksum hash %s\n",
1309                        cfs_crypto_hash_name(cfs_alg));
1310                 return PTR_ERR(req);
1311         }
1312
1313         while (nob > 0 && pg_count > 0) {
1314                 unsigned int count = pga[i]->count > nob ? nob : pga[i]->count;
1315
1316                 /* corrupt the data before we compute the checksum, to
1317                  * simulate an OST->client data error */
1318                 if (i == 0 && opc == OST_READ &&
1319                     OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE)) {
1320                         unsigned char *ptr = kmap(pga[i]->pg);
1321                         int off = pga[i]->off & ~PAGE_MASK;
1322
1323                         memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
1324                         kunmap(pga[i]->pg);
1325                 }
1326                 cfs_crypto_hash_update_page(req, pga[i]->pg,
1327                                             pga[i]->off & ~PAGE_MASK,
1328                                             count);
1329                 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d\n",
1330                                (int)(pga[i]->off & ~PAGE_MASK));
1331
1332                 nob -= pga[i]->count;
1333                 pg_count--;
1334                 i++;
1335         }
1336
1337         bufsize = sizeof(*cksum);
1338         cfs_crypto_hash_final(req, (unsigned char *)cksum, &bufsize);
1339
1340         /* For sending we only compute the wrong checksum instead
1341          * of corrupting the data so it is still correct on a redo */
1342         if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1343                 (*cksum)++;
1344
1345         return 0;
1346 }
1347
1348 static int osc_checksum_bulk_rw(const char *obd_name,
1349                                 enum cksum_types cksum_type,
1350                                 int nob, size_t pg_count,
1351                                 struct brw_page **pga, int opc,
1352                                 u32 *check_sum, bool resend)
1353 {
1354         obd_dif_csum_fn *fn = NULL;
1355         int sector_size = 0;
1356         int rc;
1357
1358         ENTRY;
1359         obd_t10_cksum2dif(cksum_type, &fn, &sector_size);
1360
1361         if (fn)
1362                 rc = osc_checksum_bulk_t10pi(obd_name, nob, pg_count, pga,
1363                                              opc, fn, sector_size, check_sum,
1364                                              resend);
1365         else
1366                 rc = osc_checksum_bulk(nob, pg_count, pga, opc, cksum_type,
1367                                        check_sum);
1368
1369         RETURN(rc);
1370 }
1371
1372 static inline void osc_release_bounce_pages(struct brw_page **pga,
1373                                             u32 page_count)
1374 {
1375 #ifdef HAVE_LUSTRE_CRYPTO
1376         int i;
1377
1378         for (i = 0; i < page_count; i++) {
1379                 /* Bounce pages allocated by a call to
1380                  * llcrypt_encrypt_pagecache_blocks() in osc_brw_prep_request()
1381                  * are identified thanks to the PageChecked flag.
1382                  */
1383                 if (PageChecked(pga[i]->pg))
1384                         llcrypt_finalize_bounce_page(&pga[i]->pg);
1385                 pga[i]->count -= pga[i]->bp_count_diff;
1386                 pga[i]->off += pga[i]->bp_off_diff;
1387         }
1388 #endif
1389 }
1390
1391 static int
1392 osc_brw_prep_request(int cmd, struct client_obd *cli, struct obdo *oa,
1393                      u32 page_count, struct brw_page **pga,
1394                      struct ptlrpc_request **reqp, int resend)
1395 {
1396         struct ptlrpc_request *req;
1397         struct ptlrpc_bulk_desc *desc;
1398         struct ost_body *body;
1399         struct obd_ioobj *ioobj;
1400         struct niobuf_remote *niobuf;
1401         int niocount, i, requested_nob, opc, rc, short_io_size = 0;
1402         struct osc_brw_async_args *aa;
1403         struct req_capsule *pill;
1404         struct brw_page *pg_prev;
1405         void *short_io_buf;
1406         const char *obd_name = cli->cl_import->imp_obd->obd_name;
1407         struct inode *inode = NULL;
1408         bool directio = false;
1409         bool enable_checksum = true;
1410
1411         ENTRY;
1412         if (pga[0]->pg) {
1413                 inode = page2inode(pga[0]->pg);
1414                 if (inode == NULL) {
1415                         /* Try to get reference to inode from cl_page if we are
1416                          * dealing with direct IO, as handled pages are not
1417                          * actual page cache pages.
1418                          */
1419                         struct osc_async_page *oap = brw_page2oap(pga[0]);
1420                         struct cl_page *clpage = oap2cl_page(oap);
1421
1422                         inode = clpage->cp_inode;
1423                         if (inode)
1424                                 directio = true;
1425                 }
1426         }
1427         if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1428                 RETURN(-ENOMEM); /* Recoverable */
1429         if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1430                 RETURN(-EINVAL); /* Fatal */
1431
1432         if ((cmd & OBD_BRW_WRITE) != 0) {
1433                 opc = OST_WRITE;
1434                 req = ptlrpc_request_alloc_pool(cli->cl_import,
1435                                                 osc_rq_pool,
1436                                                 &RQF_OST_BRW_WRITE);
1437         } else {
1438                 opc = OST_READ;
1439                 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1440         }
1441         if (req == NULL)
1442                 RETURN(-ENOMEM);
1443
1444         if (opc == OST_WRITE && inode && IS_ENCRYPTED(inode) &&
1445             llcrypt_has_encryption_key(inode)) {
1446                 for (i = 0; i < page_count; i++) {
1447                         struct brw_page *pg = pga[i];
1448                         struct page *data_page = NULL;
1449                         bool retried = false;
1450                         bool lockedbymyself;
1451                         u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
1452                         struct address_space *map_orig = NULL;
1453                         pgoff_t index_orig;
1454
1455 retry_encrypt:
1456                         nunits = round_up(nunits, LUSTRE_ENCRYPTION_UNIT_SIZE);
1457                         /* The page can already be locked when we arrive here.
1458                          * This is possible when cl_page_assume/vvp_page_assume
1459                          * is stuck on wait_on_page_writeback with page lock
1460                          * held. In this case there is no risk for the lock to
1461                          * be released while we are doing our encryption
1462                          * processing, because writeback against that page will
1463                          * end in vvp_page_completion_write/cl_page_completion,
1464                          * which means only once the page is fully processed.
1465                          */
1466                         lockedbymyself = trylock_page(pg->pg);
1467                         if (directio) {
1468                                 map_orig = pg->pg->mapping;
1469                                 pg->pg->mapping = inode->i_mapping;
1470                                 index_orig = pg->pg->index;
1471                                 pg->pg->index = pg->off >> PAGE_SHIFT;
1472                         }
1473                         data_page =
1474                                 llcrypt_encrypt_pagecache_blocks(pg->pg,
1475                                                                  nunits, 0,
1476                                                                  GFP_NOFS);
1477                         if (directio) {
1478                                 pg->pg->mapping = map_orig;
1479                                 pg->pg->index = index_orig;
1480                         }
1481                         if (lockedbymyself)
1482                                 unlock_page(pg->pg);
1483                         if (IS_ERR(data_page)) {
1484                                 rc = PTR_ERR(data_page);
1485                                 if (rc == -ENOMEM && !retried) {
1486                                         retried = true;
1487                                         rc = 0;
1488                                         goto retry_encrypt;
1489                                 }
1490                                 ptlrpc_request_free(req);
1491                                 RETURN(rc);
1492                         }
1493                         /* Set PageChecked flag on bounce page for
1494                          * disambiguation in osc_release_bounce_pages().
1495                          */
1496                         SetPageChecked(data_page);
1497                         pg->pg = data_page;
1498                         /* there should be no gap in the middle of page array */
1499                         if (i == page_count - 1) {
1500                                 struct osc_async_page *oap = brw_page2oap(pg);
1501
1502                                 oa->o_size = oap->oap_count +
1503                                         oap->oap_obj_off + oap->oap_page_off;
1504                         }
1505                         /* len is forced to nunits, and relative offset to 0
1506                          * so store the old, clear text info
1507                          */
1508                         pg->bp_count_diff = nunits - pg->count;
1509                         pg->count = nunits;
1510                         pg->bp_off_diff = pg->off & ~PAGE_MASK;
1511                         pg->off = pg->off & PAGE_MASK;
1512                 }
1513         } else if (opc == OST_WRITE && inode && IS_ENCRYPTED(inode)) {
1514                 struct osc_async_page *oap = brw_page2oap(pga[0]);
1515                 struct cl_page *clpage = oap2cl_page(oap);
1516                 struct cl_object *clobj = clpage->cp_obj;
1517                 struct cl_attr attr = { 0 };
1518                 struct lu_env *env;
1519                 __u16 refcheck;
1520
1521                 env = cl_env_get(&refcheck);
1522                 if (IS_ERR(env)) {
1523                         rc = PTR_ERR(env);
1524                         ptlrpc_request_free(req);
1525                         RETURN(rc);
1526                 }
1527
1528                 cl_object_attr_lock(clobj);
1529                 rc = cl_object_attr_get(env, clobj, &attr);
1530                 cl_object_attr_unlock(clobj);
1531                 cl_env_put(env, &refcheck);
1532                 if (rc != 0) {
1533                         ptlrpc_request_free(req);
1534                         RETURN(rc);
1535                 }
1536                 if (attr.cat_size)
1537                         oa->o_size = attr.cat_size;
1538         } else if (opc == OST_READ && inode && IS_ENCRYPTED(inode) &&
1539                    llcrypt_has_encryption_key(inode)) {
1540                 for (i = 0; i < page_count; i++) {
1541                         struct brw_page *pg = pga[i];
1542                         u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
1543
1544                         nunits = round_up(nunits, LUSTRE_ENCRYPTION_UNIT_SIZE);
1545                         /* count/off are forced to cover the whole encryption
1546                          * unit size so that all encrypted data is stored on the
1547                          * OST, so adjust bp_{count,off}_diff for the size of
1548                          * the clear text.
1549                          */
1550                         pg->bp_count_diff = nunits - pg->count;
1551                         pg->count = nunits;
1552                         pg->bp_off_diff = pg->off & ~PAGE_MASK;
1553                         pg->off = pg->off & PAGE_MASK;
1554                 }
1555         }
1556
1557         for (niocount = i = 1; i < page_count; i++) {
1558                 if (!can_merge_pages(pga[i - 1], pga[i]))
1559                         niocount++;
1560         }
1561
1562         pill = &req->rq_pill;
1563         req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1564                              sizeof(*ioobj));
1565         req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1566                              niocount * sizeof(*niobuf));
1567
1568         for (i = 0; i < page_count; i++) {
1569                 short_io_size += pga[i]->count;
1570                 if (!inode || !IS_ENCRYPTED(inode) ||
1571                     !llcrypt_has_encryption_key(inode)) {
1572                         pga[i]->bp_count_diff = 0;
1573                         pga[i]->bp_off_diff = 0;
1574                 }
1575         }
1576
1577         if (brw_page2oap(pga[0])->oap_brw_flags & OBD_BRW_RDMA_ONLY) {
1578                 enable_checksum = false;
1579                 short_io_size = 0;
1580         }
1581
1582         /* Check if read/write is small enough to be a short io. */
1583         if (short_io_size > cli->cl_max_short_io_bytes || niocount > 1 ||
1584             !imp_connect_shortio(cli->cl_import))
1585                 short_io_size = 0;
1586
1587         /* If this is an empty RPC to old server, just ignore it */
1588         if (!short_io_size && !pga[0]->pg) {
1589                 ptlrpc_request_free(req);
1590                 RETURN(-ENODATA);
1591         }
1592
1593         req_capsule_set_size(pill, &RMF_SHORT_IO, RCL_CLIENT,
1594                              opc == OST_READ ? 0 : short_io_size);
1595         if (opc == OST_READ)
1596                 req_capsule_set_size(pill, &RMF_SHORT_IO, RCL_SERVER,
1597                                      short_io_size);
1598
1599         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1600         if (rc) {
1601                 ptlrpc_request_free(req);
1602                 RETURN(rc);
1603         }
1604         osc_set_io_portal(req);
1605
1606         ptlrpc_at_set_req_timeout(req);
1607         /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1608          * retry logic */
1609         req->rq_no_retry_einprogress = 1;
1610
1611         if (short_io_size != 0) {
1612                 desc = NULL;
1613                 short_io_buf = NULL;
1614                 goto no_bulk;
1615         }
1616
1617         desc = ptlrpc_prep_bulk_imp(req, page_count,
1618                 cli->cl_import->imp_connect_data.ocd_brw_size >> LNET_MTU_BITS,
1619                 (opc == OST_WRITE ? PTLRPC_BULK_GET_SOURCE :
1620                         PTLRPC_BULK_PUT_SINK),
1621                 OST_BULK_PORTAL,
1622                 &ptlrpc_bulk_kiov_pin_ops);
1623
1624         if (desc == NULL)
1625                 GOTO(out, rc = -ENOMEM);
1626         /* NB request now owns desc and will free it when it gets freed */
1627 no_bulk:
1628         body = req_capsule_client_get(pill, &RMF_OST_BODY);
1629         ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1630         niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1631         LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1632
1633         lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
1634
1635         /* For READ and WRITE, we can't fill o_uid and o_gid using from_kuid()
1636          * and from_kgid(), because they are asynchronous. Fortunately, variable
1637          * oa contains valid o_uid and o_gid in these two operations.
1638          * Besides, filling o_uid and o_gid is enough for nrs-tbf, see LU-9658.
1639          * OBD_MD_FLUID and OBD_MD_FLUID is not set in order to avoid breaking
1640          * other process logic */
1641         body->oa.o_uid = oa->o_uid;
1642         body->oa.o_gid = oa->o_gid;
1643
1644         obdo_to_ioobj(oa, ioobj);
1645         ioobj->ioo_bufcnt = niocount;
1646         /* The high bits of ioo_max_brw tells server _maximum_ number of bulks
1647          * that might be send for this request.  The actual number is decided
1648          * when the RPC is finally sent in ptlrpc_register_bulk(). It sends
1649          * "max - 1" for old client compatibility sending "0", and also so the
1650          * the actual maximum is a power-of-two number, not one less. LU-1431 */
1651         if (desc != NULL)
1652                 ioobj_max_brw_set(ioobj, desc->bd_md_max_brw);
1653         else /* short io */
1654                 ioobj_max_brw_set(ioobj, 0);
1655
1656         if (short_io_size != 0) {
1657                 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1658                         body->oa.o_valid |= OBD_MD_FLFLAGS;
1659                         body->oa.o_flags = 0;
1660                 }
1661                 body->oa.o_flags |= OBD_FL_SHORT_IO;
1662                 CDEBUG(D_CACHE, "Using short io for data transfer, size = %d\n",
1663                        short_io_size);
1664                 if (opc == OST_WRITE) {
1665                         short_io_buf = req_capsule_client_get(pill,
1666                                                               &RMF_SHORT_IO);
1667                         LASSERT(short_io_buf != NULL);
1668                 }
1669         }
1670
1671         LASSERT(page_count > 0);
1672         pg_prev = pga[0];
1673         for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1674                 struct brw_page *pg = pga[i];
1675                 int poff = pg->off & ~PAGE_MASK;
1676
1677                 LASSERT(pg->count > 0);
1678                 /* make sure there is no gap in the middle of page array */
1679                 LASSERTF(page_count == 1 ||
1680                          (ergo(i == 0, poff + pg->count == PAGE_SIZE) &&
1681                           ergo(i > 0 && i < page_count - 1,
1682                                poff == 0 && pg->count == PAGE_SIZE)   &&
1683                           ergo(i == page_count - 1, poff == 0)),
1684                          "i: %d/%d pg: %p off: %llu, count: %u\n",
1685                          i, page_count, pg, pg->off, pg->count);
1686                 LASSERTF(i == 0 || pg->off > pg_prev->off,
1687                          "i %d p_c %u pg %p [pri %lu ind %lu] off %llu"
1688                          " prev_pg %p [pri %lu ind %lu] off %llu\n",
1689                          i, page_count,
1690                          pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1691                          pg_prev->pg, page_private(pg_prev->pg),
1692                          pg_prev->pg->index, pg_prev->off);
1693                 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1694                         (pg->flag & OBD_BRW_SRVLOCK));
1695                 if (short_io_size != 0 && opc == OST_WRITE) {
1696                         unsigned char *ptr = kmap_atomic(pg->pg);
1697
1698                         LASSERT(short_io_size >= requested_nob + pg->count);
1699                         memcpy(short_io_buf + requested_nob,
1700                                ptr + poff,
1701                                pg->count);
1702                         kunmap_atomic(ptr);
1703                 } else if (short_io_size == 0) {
1704                         desc->bd_frag_ops->add_kiov_frag(desc, pg->pg, poff,
1705                                                          pg->count);
1706                 }
1707                 requested_nob += pg->count;
1708
1709                 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1710                         niobuf--;
1711                         niobuf->rnb_len += pg->count;
1712                 } else {
1713                         niobuf->rnb_offset = pg->off;
1714                         niobuf->rnb_len    = pg->count;
1715                         niobuf->rnb_flags  = pg->flag;
1716                 }
1717                 pg_prev = pg;
1718         }
1719
1720         LASSERTF((void *)(niobuf - niocount) ==
1721                 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1722                 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1723                 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1724
1725         osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1726         if (resend) {
1727                 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1728                         body->oa.o_valid |= OBD_MD_FLFLAGS;
1729                         body->oa.o_flags = 0;
1730                 }
1731                 body->oa.o_flags |= OBD_FL_RECOV_RESEND;
1732         }
1733
1734         if (osc_should_shrink_grant(cli))
1735                 osc_shrink_grant_local(cli, &body->oa);
1736
1737         if (!cli->cl_checksum || sptlrpc_flavor_has_bulk(&req->rq_flvr))
1738                 enable_checksum = false;
1739
1740         /* size[REQ_REC_OFF] still sizeof (*body) */
1741         if (opc == OST_WRITE) {
1742                 if (enable_checksum) {
1743                         /* store cl_cksum_type in a local variable since
1744                          * it can be changed via lprocfs */
1745                         enum cksum_types cksum_type = cli->cl_cksum_type;
1746
1747                         if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1748                                 body->oa.o_flags = 0;
1749
1750                         body->oa.o_flags |= obd_cksum_type_pack(obd_name,
1751                                                                 cksum_type);
1752                         body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1753
1754                         rc = osc_checksum_bulk_rw(obd_name, cksum_type,
1755                                                   requested_nob, page_count,
1756                                                   pga, OST_WRITE,
1757                                                   &body->oa.o_cksum, resend);
1758                         if (rc < 0) {
1759                                 CDEBUG(D_PAGE, "failed to checksum: rc = %d\n",
1760                                        rc);
1761                                 GOTO(out, rc);
1762                         }
1763                         CDEBUG(D_PAGE | (resend ? D_HA : 0),
1764                                "checksum at write origin: %x (%x)\n",
1765                                body->oa.o_cksum, cksum_type);
1766
1767                         /* save this in 'oa', too, for later checking */
1768                         oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1769                         oa->o_flags |= obd_cksum_type_pack(obd_name,
1770                                                            cksum_type);
1771                 } else {
1772                         /* clear out the checksum flag, in case this is a
1773                          * resend but cl_checksum is no longer set. b=11238 */
1774                         oa->o_valid &= ~OBD_MD_FLCKSUM;
1775                 }
1776                 oa->o_cksum = body->oa.o_cksum;
1777                 /* 1 RC per niobuf */
1778                 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1779                                      sizeof(__u32) * niocount);
1780         } else {
1781                 if (enable_checksum) {
1782                         if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1783                                 body->oa.o_flags = 0;
1784                         body->oa.o_flags |= obd_cksum_type_pack(obd_name,
1785                                 cli->cl_cksum_type);
1786                         body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1787                 }
1788
1789                 /* Client cksum has been already copied to wire obdo in previous
1790                  * lustre_set_wire_obdo(), and in the case a bulk-read is being
1791                  * resent due to cksum error, this will allow Server to
1792                  * check+dump pages on its side */
1793         }
1794         ptlrpc_request_set_replen(req);
1795
1796         aa = ptlrpc_req_async_args(aa, req);
1797         aa->aa_oa = oa;
1798         aa->aa_requested_nob = requested_nob;
1799         aa->aa_nio_count = niocount;
1800         aa->aa_page_count = page_count;
1801         aa->aa_resends = 0;
1802         aa->aa_ppga = pga;
1803         aa->aa_cli = cli;
1804         INIT_LIST_HEAD(&aa->aa_oaps);
1805
1806         *reqp = req;
1807         niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1808         CDEBUG(D_RPCTRACE, "brw rpc %p - object "DOSTID" offset %lld<>%lld\n",
1809                 req, POSTID(&oa->o_oi), niobuf[0].rnb_offset,
1810                 niobuf[niocount - 1].rnb_offset + niobuf[niocount - 1].rnb_len);
1811         RETURN(0);
1812
1813  out:
1814         ptlrpc_req_finished(req);
1815         RETURN(rc);
1816 }
1817
1818 char dbgcksum_file_name[PATH_MAX];
1819
1820 static void dump_all_bulk_pages(struct obdo *oa, __u32 page_count,
1821                                 struct brw_page **pga, __u32 server_cksum,
1822                                 __u32 client_cksum)
1823 {
1824         struct file *filp;
1825         int rc, i;
1826         unsigned int len;
1827         char *buf;
1828
1829         /* will only keep dump of pages on first error for the same range in
1830          * file/fid, not during the resends/retries. */
1831         snprintf(dbgcksum_file_name, sizeof(dbgcksum_file_name),
1832                  "%s-checksum_dump-osc-"DFID":[%llu-%llu]-%x-%x",
1833                  (strncmp(libcfs_debug_file_path, "NONE", 4) != 0 ?
1834                   libcfs_debug_file_path : LIBCFS_DEBUG_FILE_PATH_DEFAULT),
1835                  oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : 0ULL,
1836                  oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1837                  oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1838                  pga[0]->off,
1839                  pga[page_count-1]->off + pga[page_count-1]->count - 1,
1840                  client_cksum, server_cksum);
1841         CWARN("dumping checksum data to %s\n", dbgcksum_file_name);
1842         filp = filp_open(dbgcksum_file_name,
1843                          O_CREAT | O_EXCL | O_WRONLY | O_LARGEFILE, 0600);
1844         if (IS_ERR(filp)) {
1845                 rc = PTR_ERR(filp);
1846                 if (rc == -EEXIST)
1847                         CDEBUG(D_INFO, "%s: can't open to dump pages with "
1848                                "checksum error: rc = %d\n", dbgcksum_file_name,
1849                                rc);
1850                 else
1851                         CERROR("%s: can't open to dump pages with checksum "
1852                                "error: rc = %d\n", dbgcksum_file_name, rc);
1853                 return;
1854         }
1855
1856         for (i = 0; i < page_count; i++) {
1857                 len = pga[i]->count;
1858                 buf = kmap(pga[i]->pg);
1859                 while (len != 0) {
1860                         rc = cfs_kernel_write(filp, buf, len, &filp->f_pos);
1861                         if (rc < 0) {
1862                                 CERROR("%s: wanted to write %u but got %d "
1863                                        "error\n", dbgcksum_file_name, len, rc);
1864                                 break;
1865                         }
1866                         len -= rc;
1867                         buf += rc;
1868                 }
1869                 kunmap(pga[i]->pg);
1870         }
1871
1872         rc = vfs_fsync_range(filp, 0, LLONG_MAX, 1);
1873         if (rc)
1874                 CERROR("%s: sync returns %d\n", dbgcksum_file_name, rc);
1875         filp_close(filp, NULL);
1876
1877         libcfs_debug_dumplog();
1878 }
1879
1880 static int
1881 check_write_checksum(struct obdo *oa, const struct lnet_process_id *peer,
1882                      __u32 client_cksum, __u32 server_cksum,
1883                      struct osc_brw_async_args *aa)
1884 {
1885         const char *obd_name = aa->aa_cli->cl_import->imp_obd->obd_name;
1886         enum cksum_types cksum_type;
1887         obd_dif_csum_fn *fn = NULL;
1888         int sector_size = 0;
1889         __u32 new_cksum;
1890         char *msg;
1891         int rc;
1892
1893         if (server_cksum == client_cksum) {
1894                 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1895                 return 0;
1896         }
1897
1898         if (aa->aa_cli->cl_checksum_dump)
1899                 dump_all_bulk_pages(oa, aa->aa_page_count, aa->aa_ppga,
1900                                     server_cksum, client_cksum);
1901
1902         cksum_type = obd_cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
1903                                            oa->o_flags : 0);
1904
1905         switch (cksum_type) {
1906         case OBD_CKSUM_T10IP512:
1907                 fn = obd_dif_ip_fn;
1908                 sector_size = 512;
1909                 break;
1910         case OBD_CKSUM_T10IP4K:
1911                 fn = obd_dif_ip_fn;
1912                 sector_size = 4096;
1913                 break;
1914         case OBD_CKSUM_T10CRC512:
1915                 fn = obd_dif_crc_fn;
1916                 sector_size = 512;
1917                 break;
1918         case OBD_CKSUM_T10CRC4K:
1919                 fn = obd_dif_crc_fn;
1920                 sector_size = 4096;
1921                 break;
1922         default:
1923                 break;
1924         }
1925
1926         if (fn)
1927                 rc = osc_checksum_bulk_t10pi(obd_name, aa->aa_requested_nob,
1928                                              aa->aa_page_count, aa->aa_ppga,
1929                                              OST_WRITE, fn, sector_size,
1930                                              &new_cksum, true);
1931         else
1932                 rc = osc_checksum_bulk(aa->aa_requested_nob, aa->aa_page_count,
1933                                        aa->aa_ppga, OST_WRITE, cksum_type,
1934                                        &new_cksum);
1935
1936         if (rc < 0)
1937                 msg = "failed to calculate the client write checksum";
1938         else if (cksum_type != obd_cksum_type_unpack(aa->aa_oa->o_flags))
1939                 msg = "the server did not use the checksum type specified in "
1940                       "the original request - likely a protocol problem";
1941         else if (new_cksum == server_cksum)
1942                 msg = "changed on the client after we checksummed it - "
1943                       "likely false positive due to mmap IO (bug 11742)";
1944         else if (new_cksum == client_cksum)
1945                 msg = "changed in transit before arrival at OST";
1946         else
1947                 msg = "changed in transit AND doesn't match the original - "
1948                       "likely false positive due to mmap IO (bug 11742)";
1949
1950         LCONSOLE_ERROR_MSG(0x132, "%s: BAD WRITE CHECKSUM: %s: from %s inode "
1951                            DFID " object "DOSTID" extent [%llu-%llu], original "
1952                            "client csum %x (type %x), server csum %x (type %x),"
1953                            " client csum now %x\n",
1954                            obd_name, msg, libcfs_nid2str(peer->nid),
1955                            oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1956                            oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1957                            oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1958                            POSTID(&oa->o_oi), aa->aa_ppga[0]->off,
1959                            aa->aa_ppga[aa->aa_page_count - 1]->off +
1960                                 aa->aa_ppga[aa->aa_page_count-1]->count - 1,
1961                            client_cksum,
1962                            obd_cksum_type_unpack(aa->aa_oa->o_flags),
1963                            server_cksum, cksum_type, new_cksum);
1964         return 1;
1965 }
1966
1967 /* Note rc enters this function as number of bytes transferred */
1968 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1969 {
1970         struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1971         struct client_obd *cli = aa->aa_cli;
1972         const char *obd_name = cli->cl_import->imp_obd->obd_name;
1973         const struct lnet_process_id *peer =
1974                 &req->rq_import->imp_connection->c_peer;
1975         struct ost_body *body;
1976         u32 client_cksum = 0;
1977         struct inode *inode;
1978         unsigned int blockbits = 0, blocksize = 0;
1979
1980         ENTRY;
1981
1982         if (rc < 0 && rc != -EDQUOT) {
1983                 DEBUG_REQ(D_INFO, req, "Failed request: rc = %d", rc);
1984                 RETURN(rc);
1985         }
1986
1987         LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1988         body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1989         if (body == NULL) {
1990                 DEBUG_REQ(D_INFO, req, "cannot unpack body");
1991                 RETURN(-EPROTO);
1992         }
1993
1994         /* set/clear over quota flag for a uid/gid/projid */
1995         if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1996             body->oa.o_valid & (OBD_MD_FLALLQUOTA)) {
1997                 unsigned qid[LL_MAXQUOTAS] = {
1998                                          body->oa.o_uid, body->oa.o_gid,
1999                                          body->oa.o_projid };
2000                 CDEBUG(D_QUOTA,
2001                        "setdq for [%u %u %u] with valid %#llx, flags %x\n",
2002                        body->oa.o_uid, body->oa.o_gid, body->oa.o_projid,
2003                        body->oa.o_valid, body->oa.o_flags);
2004                 osc_quota_setdq(cli, req->rq_xid, qid, body->oa.o_valid,
2005                                 body->oa.o_flags);
2006         }
2007
2008         osc_update_grant(cli, body);
2009
2010         if (rc < 0)
2011                 RETURN(rc);
2012
2013         if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
2014                 client_cksum = aa->aa_oa->o_cksum; /* save for later */
2015
2016         if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
2017                 if (rc > 0) {
2018                         CERROR("%s: unexpected positive size %d\n",
2019                                obd_name, rc);
2020                         RETURN(-EPROTO);
2021                 }
2022
2023                 if (req->rq_bulk != NULL &&
2024                     sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
2025                         RETURN(-EAGAIN);
2026
2027                 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
2028                     check_write_checksum(&body->oa, peer, client_cksum,
2029                                          body->oa.o_cksum, aa))
2030                         RETURN(-EAGAIN);
2031
2032                 rc = check_write_rcs(req, aa->aa_requested_nob,
2033                                      aa->aa_nio_count, aa->aa_page_count,
2034                                      aa->aa_ppga);
2035                 GOTO(out, rc);
2036         }
2037
2038         /* The rest of this function executes only for OST_READs */
2039
2040         if (req->rq_bulk == NULL) {
2041                 rc = req_capsule_get_size(&req->rq_pill, &RMF_SHORT_IO,
2042                                           RCL_SERVER);
2043                 LASSERT(rc == req->rq_status);
2044         } else {
2045                 /* if unwrap_bulk failed, return -EAGAIN to retry */
2046                 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
2047         }
2048         if (rc < 0)
2049                 GOTO(out, rc = -EAGAIN);
2050
2051         if (rc > aa->aa_requested_nob) {
2052                 CERROR("%s: unexpected size %d, requested %d\n", obd_name,
2053                        rc, aa->aa_requested_nob);
2054                 RETURN(-EPROTO);
2055         }
2056
2057         if (req->rq_bulk != NULL && rc != req->rq_bulk->bd_nob_transferred) {
2058                 CERROR("%s: unexpected size %d, transferred %d\n", obd_name,
2059                        rc, req->rq_bulk->bd_nob_transferred);
2060                 RETURN(-EPROTO);
2061         }
2062
2063         if (req->rq_bulk == NULL) {
2064                 /* short io */
2065                 int nob, pg_count, i = 0;
2066                 unsigned char *buf;
2067
2068                 CDEBUG(D_CACHE, "Using short io read, size %d\n", rc);
2069                 pg_count = aa->aa_page_count;
2070                 buf = req_capsule_server_sized_get(&req->rq_pill, &RMF_SHORT_IO,
2071                                                    rc);
2072                 nob = rc;
2073                 while (nob > 0 && pg_count > 0) {
2074                         unsigned char *ptr;
2075                         int count = aa->aa_ppga[i]->count > nob ?
2076                                     nob : aa->aa_ppga[i]->count;
2077
2078                         CDEBUG(D_CACHE, "page %p count %d\n",
2079                                aa->aa_ppga[i]->pg, count);
2080                         ptr = kmap_atomic(aa->aa_ppga[i]->pg);
2081                         memcpy(ptr + (aa->aa_ppga[i]->off & ~PAGE_MASK), buf,
2082                                count);
2083                         kunmap_atomic((void *) ptr);
2084
2085                         buf += count;
2086                         nob -= count;
2087                         i++;
2088                         pg_count--;
2089                 }
2090         }
2091
2092         if (rc < aa->aa_requested_nob)
2093                 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
2094
2095         if (body->oa.o_valid & OBD_MD_FLCKSUM) {
2096                 static int cksum_counter;
2097                 u32 server_cksum = body->oa.o_cksum;
2098                 int nob = rc;
2099                 char *via = "";
2100                 char *router = "";
2101                 enum cksum_types cksum_type;
2102                 u32 o_flags = body->oa.o_valid & OBD_MD_FLFLAGS ?
2103                         body->oa.o_flags : 0;
2104
2105                 cksum_type = obd_cksum_type_unpack(o_flags);
2106                 rc = osc_checksum_bulk_rw(obd_name, cksum_type, nob,
2107                                           aa->aa_page_count, aa->aa_ppga,
2108                                           OST_READ, &client_cksum, false);
2109                 if (rc < 0)
2110                         GOTO(out, rc);
2111
2112                 if (req->rq_bulk != NULL &&
2113                     peer->nid != req->rq_bulk->bd_sender) {
2114                         via = " via ";
2115                         router = libcfs_nid2str(req->rq_bulk->bd_sender);
2116                 }
2117
2118                 if (server_cksum != client_cksum) {
2119                         struct ost_body *clbody;
2120                         __u32 client_cksum2;
2121                         u32 page_count = aa->aa_page_count;
2122
2123                         osc_checksum_bulk_rw(obd_name, cksum_type, nob,
2124                                              page_count, aa->aa_ppga,
2125                                              OST_READ, &client_cksum2, true);
2126                         clbody = req_capsule_client_get(&req->rq_pill,
2127                                                         &RMF_OST_BODY);
2128                         if (cli->cl_checksum_dump)
2129                                 dump_all_bulk_pages(&clbody->oa, page_count,
2130                                                     aa->aa_ppga, server_cksum,
2131                                                     client_cksum);
2132
2133                         LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
2134                                            "%s%s%s inode "DFID" object "DOSTID
2135                                            " extent [%llu-%llu], client %x/%x, "
2136                                            "server %x, cksum_type %x\n",
2137                                            obd_name,
2138                                            libcfs_nid2str(peer->nid),
2139                                            via, router,
2140                                            clbody->oa.o_valid & OBD_MD_FLFID ?
2141                                                 clbody->oa.o_parent_seq : 0ULL,
2142                                            clbody->oa.o_valid & OBD_MD_FLFID ?
2143                                                 clbody->oa.o_parent_oid : 0,
2144                                            clbody->oa.o_valid & OBD_MD_FLFID ?
2145                                                 clbody->oa.o_parent_ver : 0,
2146                                            POSTID(&body->oa.o_oi),
2147                                            aa->aa_ppga[0]->off,
2148                                            aa->aa_ppga[page_count-1]->off +
2149                                            aa->aa_ppga[page_count-1]->count - 1,
2150                                            client_cksum, client_cksum2,
2151                                            server_cksum, cksum_type);
2152                         cksum_counter = 0;
2153                         aa->aa_oa->o_cksum = client_cksum;
2154                         rc = -EAGAIN;
2155                 } else {
2156                         cksum_counter++;
2157                         CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
2158                         rc = 0;
2159                 }
2160         } else if (unlikely(client_cksum)) {
2161                 static int cksum_missed;
2162
2163                 cksum_missed++;
2164                 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
2165                         CERROR("%s: checksum %u requested from %s but not sent\n",
2166                                obd_name, cksum_missed,
2167                                libcfs_nid2str(peer->nid));
2168         } else {
2169                 rc = 0;
2170         }
2171
2172         inode = page2inode(aa->aa_ppga[0]->pg);
2173         if (inode == NULL) {
2174                 /* Try to get reference to inode from cl_page if we are
2175                  * dealing with direct IO, as handled pages are not
2176                  * actual page cache pages.
2177                  */
2178                 struct osc_async_page *oap = brw_page2oap(aa->aa_ppga[0]);
2179
2180                 inode = oap2cl_page(oap)->cp_inode;
2181                 if (inode) {
2182                         blockbits = inode->i_blkbits;
2183                         blocksize = 1 << blockbits;
2184                 }
2185         }
2186         if (inode && IS_ENCRYPTED(inode)) {
2187                 int idx;
2188
2189                 if (!llcrypt_has_encryption_key(inode)) {
2190                         CDEBUG(D_SEC, "no enc key for ino %lu\n", inode->i_ino);
2191                         GOTO(out, rc);
2192                 }
2193                 for (idx = 0; idx < aa->aa_page_count; idx++) {
2194                         struct brw_page *pg = aa->aa_ppga[idx];
2195                         unsigned int offs = 0;
2196
2197                         while (offs < PAGE_SIZE) {
2198                                 /* do not decrypt if page is all 0s */
2199                                 if (memchr_inv(page_address(pg->pg) + offs, 0,
2200                                          LUSTRE_ENCRYPTION_UNIT_SIZE) == NULL) {
2201                                         /* if page is empty forward info to
2202                                          * upper layers (ll_io_zero_page) by
2203                                          * clearing PagePrivate2
2204                                          */
2205                                         if (!offs)
2206                                                 ClearPagePrivate2(pg->pg);
2207                                         break;
2208                                 }
2209
2210                                 if (blockbits) {
2211                                         /* This is direct IO case. Directly call
2212                                          * decrypt function that takes inode as
2213                                          * input parameter. Page does not need
2214                                          * to be locked.
2215                                          */
2216                                         u64 lblk_num =
2217                                                 ((u64)(pg->off >> PAGE_SHIFT) <<
2218                                                      (PAGE_SHIFT - blockbits)) +
2219                                                        (offs >> blockbits);
2220                                         unsigned int i;
2221
2222                                         for (i = offs;
2223                                              i < offs +
2224                                                     LUSTRE_ENCRYPTION_UNIT_SIZE;
2225                                              i += blocksize, lblk_num++) {
2226                                                 rc =
2227                                                   llcrypt_decrypt_block_inplace(
2228                                                           inode, pg->pg,
2229                                                           blocksize, i,
2230                                                           lblk_num);
2231                                                 if (rc)
2232                                                         break;
2233                                         }
2234                                 } else {
2235                                         rc = llcrypt_decrypt_pagecache_blocks(
2236                                                 pg->pg,
2237                                                 LUSTRE_ENCRYPTION_UNIT_SIZE,
2238                                                 offs);
2239                                 }
2240                                 if (rc)
2241                                         GOTO(out, rc);
2242
2243                                 offs += LUSTRE_ENCRYPTION_UNIT_SIZE;
2244                         }
2245                 }
2246         }
2247
2248 out:
2249         if (rc >= 0)
2250                 lustre_get_wire_obdo(&req->rq_import->imp_connect_data,
2251                                      aa->aa_oa, &body->oa);
2252
2253         RETURN(rc);
2254 }
2255
2256 static int osc_brw_redo_request(struct ptlrpc_request *request,
2257                                 struct osc_brw_async_args *aa, int rc)
2258 {
2259         struct ptlrpc_request *new_req;
2260         struct osc_brw_async_args *new_aa;
2261         struct osc_async_page *oap;
2262         ENTRY;
2263
2264         /* The below message is checked in replay-ost-single.sh test_8ae*/
2265         DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
2266                   "redo for recoverable error %d", rc);
2267
2268         rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
2269                                 OST_WRITE ? OBD_BRW_WRITE : OBD_BRW_READ,
2270                                   aa->aa_cli, aa->aa_oa, aa->aa_page_count,
2271                                   aa->aa_ppga, &new_req, 1);
2272         if (rc)
2273                 RETURN(rc);
2274
2275         list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2276                 if (oap->oap_request != NULL) {
2277                         LASSERTF(request == oap->oap_request,
2278                                  "request %p != oap_request %p\n",
2279                                  request, oap->oap_request);
2280                 }
2281         }
2282         /*
2283          * New request takes over pga and oaps from old request.
2284          * Note that copying a list_head doesn't work, need to move it...
2285          */
2286         aa->aa_resends++;
2287         new_req->rq_interpret_reply = request->rq_interpret_reply;
2288         new_req->rq_async_args = request->rq_async_args;
2289         new_req->rq_commit_cb = request->rq_commit_cb;
2290         /* cap resend delay to the current request timeout, this is similar to
2291          * what ptlrpc does (see after_reply()) */
2292         if (aa->aa_resends > new_req->rq_timeout)
2293                 new_req->rq_sent = ktime_get_real_seconds() + new_req->rq_timeout;
2294         else
2295                 new_req->rq_sent = ktime_get_real_seconds() + aa->aa_resends;
2296         new_req->rq_generation_set = 1;
2297         new_req->rq_import_generation = request->rq_import_generation;
2298
2299         new_aa = ptlrpc_req_async_args(new_aa, new_req);
2300
2301         INIT_LIST_HEAD(&new_aa->aa_oaps);
2302         list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
2303         INIT_LIST_HEAD(&new_aa->aa_exts);
2304         list_splice_init(&aa->aa_exts, &new_aa->aa_exts);
2305         new_aa->aa_resends = aa->aa_resends;
2306
2307         list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
2308                 if (oap->oap_request) {
2309                         ptlrpc_req_finished(oap->oap_request);
2310                         oap->oap_request = ptlrpc_request_addref(new_req);
2311                 }
2312         }
2313
2314         /* XXX: This code will run into problem if we're going to support
2315          * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
2316          * and wait for all of them to be finished. We should inherit request
2317          * set from old request. */
2318         ptlrpcd_add_req(new_req);
2319
2320         DEBUG_REQ(D_INFO, new_req, "new request");
2321         RETURN(0);
2322 }
2323
2324 /*
2325  * ugh, we want disk allocation on the target to happen in offset order.  we'll
2326  * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
2327  * fine for our small page arrays and doesn't require allocation.  its an
2328  * insertion sort that swaps elements that are strides apart, shrinking the
2329  * stride down until its '1' and the array is sorted.
2330  */
2331 static void sort_brw_pages(struct brw_page **array, int num)
2332 {
2333         int stride, i, j;
2334         struct brw_page *tmp;
2335
2336         if (num == 1)
2337                 return;
2338         for (stride = 1; stride < num ; stride = (stride * 3) + 1)
2339                 ;
2340
2341         do {
2342                 stride /= 3;
2343                 for (i = stride ; i < num ; i++) {
2344                         tmp = array[i];
2345                         j = i;
2346                         while (j >= stride && array[j - stride]->off > tmp->off) {
2347                                 array[j] = array[j - stride];
2348                                 j -= stride;
2349                         }
2350                         array[j] = tmp;
2351                 }
2352         } while (stride > 1);
2353 }
2354
2355 static void osc_release_ppga(struct brw_page **ppga, size_t count)
2356 {
2357         LASSERT(ppga != NULL);
2358         OBD_FREE_PTR_ARRAY_LARGE(ppga, count);
2359 }
2360
2361 static int brw_interpret(const struct lu_env *env,
2362                          struct ptlrpc_request *req, void *args, int rc)
2363 {
2364         struct osc_brw_async_args *aa = args;
2365         struct osc_extent *ext;
2366         struct osc_extent *tmp;
2367         struct client_obd *cli = aa->aa_cli;
2368         unsigned long transferred = 0;
2369
2370         ENTRY;
2371
2372         rc = osc_brw_fini_request(req, rc);
2373         CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
2374
2375         /* restore clear text pages */
2376         osc_release_bounce_pages(aa->aa_ppga, aa->aa_page_count);
2377
2378         /*
2379          * When server returns -EINPROGRESS, client should always retry
2380          * regardless of the number of times the bulk was resent already.
2381          */
2382         if (osc_recoverable_error(rc) && !req->rq_no_delay) {
2383                 if (req->rq_import_generation !=
2384                     req->rq_import->imp_generation) {
2385                         CDEBUG(D_HA, "%s: resend cross eviction for object: "
2386                                ""DOSTID", rc = %d.\n",
2387                                req->rq_import->imp_obd->obd_name,
2388                                POSTID(&aa->aa_oa->o_oi), rc);
2389                 } else if (rc == -EINPROGRESS ||
2390                            client_should_resend(aa->aa_resends, aa->aa_cli)) {
2391                         rc = osc_brw_redo_request(req, aa, rc);
2392                 } else {
2393                         CERROR("%s: too many resent retries for object: "
2394                                "%llu:%llu, rc = %d.\n",
2395                                req->rq_import->imp_obd->obd_name,
2396                                POSTID(&aa->aa_oa->o_oi), rc);
2397                 }
2398
2399                 if (rc == 0)
2400                         RETURN(0);
2401                 else if (rc == -EAGAIN || rc == -EINPROGRESS)
2402                         rc = -EIO;
2403         }
2404
2405         if (rc == 0) {
2406                 struct obdo *oa = aa->aa_oa;
2407                 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
2408                 unsigned long valid = 0;
2409                 struct cl_object *obj;
2410                 struct osc_async_page *last;
2411
2412                 last = brw_page2oap(aa->aa_ppga[aa->aa_page_count - 1]);
2413                 obj = osc2cl(last->oap_obj);
2414
2415                 cl_object_attr_lock(obj);
2416                 if (oa->o_valid & OBD_MD_FLBLOCKS) {
2417                         attr->cat_blocks = oa->o_blocks;
2418                         valid |= CAT_BLOCKS;
2419                 }
2420                 if (oa->o_valid & OBD_MD_FLMTIME) {
2421                         attr->cat_mtime = oa->o_mtime;
2422                         valid |= CAT_MTIME;
2423                 }
2424                 if (oa->o_valid & OBD_MD_FLATIME) {
2425                         attr->cat_atime = oa->o_atime;
2426                         valid |= CAT_ATIME;
2427                 }
2428                 if (oa->o_valid & OBD_MD_FLCTIME) {
2429                         attr->cat_ctime = oa->o_ctime;
2430                         valid |= CAT_CTIME;
2431                 }
2432
2433                 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
2434                         struct lov_oinfo *loi = cl2osc(obj)->oo_oinfo;
2435                         loff_t last_off = last->oap_count + last->oap_obj_off +
2436                                 last->oap_page_off;
2437
2438                         /* Change file size if this is an out of quota or
2439                          * direct IO write and it extends the file size */
2440                         if (loi->loi_lvb.lvb_size < last_off) {
2441                                 attr->cat_size = last_off;
2442                                 valid |= CAT_SIZE;
2443                         }
2444                         /* Extend KMS if it's not a lockless write */
2445                         if (loi->loi_kms < last_off &&
2446                             oap2osc_page(last)->ops_srvlock == 0) {
2447                                 attr->cat_kms = last_off;
2448                                 valid |= CAT_KMS;
2449                         }
2450                 }
2451
2452                 if (valid != 0)
2453                         cl_object_attr_update(env, obj, attr, valid);
2454                 cl_object_attr_unlock(obj);
2455         }
2456         OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
2457         aa->aa_oa = NULL;
2458
2459         if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE && rc == 0)
2460                 osc_inc_unstable_pages(req);
2461
2462         list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
2463                 list_del_init(&ext->oe_link);
2464                 osc_extent_finish(env, ext, 1,
2465                                   rc && req->rq_no_delay ? -EAGAIN : rc);
2466         }
2467         LASSERT(list_empty(&aa->aa_exts));
2468         LASSERT(list_empty(&aa->aa_oaps));
2469
2470         transferred = (req->rq_bulk == NULL ? /* short io */
2471                        aa->aa_requested_nob :
2472                        req->rq_bulk->bd_nob_transferred);
2473
2474         osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2475         ptlrpc_lprocfs_brw(req, transferred);
2476
2477         spin_lock(&cli->cl_loi_list_lock);
2478         /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2479          * is called so we know whether to go to sync BRWs or wait for more
2480          * RPCs to complete */
2481         if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2482                 cli->cl_w_in_flight--;
2483         else
2484                 cli->cl_r_in_flight--;
2485         osc_wake_cache_waiters(cli);
2486         spin_unlock(&cli->cl_loi_list_lock);
2487
2488         osc_io_unplug(env, cli, NULL);
2489         RETURN(rc);
2490 }
2491
2492 static void brw_commit(struct ptlrpc_request *req)
2493 {
2494         /* If osc_inc_unstable_pages (via osc_extent_finish) races with
2495          * this called via the rq_commit_cb, I need to ensure
2496          * osc_dec_unstable_pages is still called. Otherwise unstable
2497          * pages may be leaked. */
2498         spin_lock(&req->rq_lock);
2499         if (likely(req->rq_unstable)) {
2500                 req->rq_unstable = 0;
2501                 spin_unlock(&req->rq_lock);
2502
2503                 osc_dec_unstable_pages(req);
2504         } else {
2505                 req->rq_committed = 1;
2506                 spin_unlock(&req->rq_lock);
2507         }
2508 }
2509
2510 /**
2511  * Build an RPC by the list of extent @ext_list. The caller must ensure
2512  * that the total pages in this list are NOT over max pages per RPC.
2513  * Extents in the list must be in OES_RPC state.
2514  */
2515 int osc_build_rpc(const struct lu_env *env, struct client_obd *cli,
2516                   struct list_head *ext_list, int cmd)
2517 {
2518         struct ptlrpc_request           *req = NULL;
2519         struct osc_extent               *ext;
2520         struct brw_page                 **pga = NULL;
2521         struct osc_brw_async_args       *aa = NULL;
2522         struct obdo                     *oa = NULL;
2523         struct osc_async_page           *oap;
2524         struct osc_object               *obj = NULL;
2525         struct cl_req_attr              *crattr = NULL;
2526         loff_t                          starting_offset = OBD_OBJECT_EOF;
2527         loff_t                          ending_offset = 0;
2528         /* '1' for consistency with code that checks !mpflag to restore */
2529         int mpflag = 1;
2530         int                             mem_tight = 0;
2531         int                             page_count = 0;
2532         bool                            soft_sync = false;
2533         bool                            ndelay = false;
2534         int                             i;
2535         int                             grant = 0;
2536         int                             rc;
2537         __u32                           layout_version = 0;
2538         LIST_HEAD(rpc_list);
2539         struct ost_body                 *body;
2540         ENTRY;
2541         LASSERT(!list_empty(ext_list));
2542
2543         /* add pages into rpc_list to build BRW rpc */
2544         list_for_each_entry(ext, ext_list, oe_link) {
2545                 LASSERT(ext->oe_state == OES_RPC);
2546                 mem_tight |= ext->oe_memalloc;
2547                 grant += ext->oe_grants;
2548                 page_count += ext->oe_nr_pages;
2549                 layout_version = max(layout_version, ext->oe_layout_version);
2550                 if (obj == NULL)
2551                         obj = ext->oe_obj;
2552         }
2553
2554         soft_sync = osc_over_unstable_soft_limit(cli);
2555         if (mem_tight)
2556                 mpflag = memalloc_noreclaim_save();
2557
2558         OBD_ALLOC_PTR_ARRAY_LARGE(pga, page_count);
2559         if (pga == NULL)
2560                 GOTO(out, rc = -ENOMEM);
2561
2562         OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
2563         if (oa == NULL)
2564                 GOTO(out, rc = -ENOMEM);
2565
2566         i = 0;
2567         list_for_each_entry(ext, ext_list, oe_link) {
2568                 list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
2569                         if (mem_tight)
2570                                 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
2571                         if (soft_sync)
2572                                 oap->oap_brw_flags |= OBD_BRW_SOFT_SYNC;
2573                         pga[i] = &oap->oap_brw_page;
2574                         pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2575                         i++;
2576
2577                         list_add_tail(&oap->oap_rpc_item, &rpc_list);
2578                         if (starting_offset == OBD_OBJECT_EOF ||
2579                             starting_offset > oap->oap_obj_off)
2580                                 starting_offset = oap->oap_obj_off;
2581                         else
2582                                 LASSERT(oap->oap_page_off == 0);
2583                         if (ending_offset < oap->oap_obj_off + oap->oap_count)
2584                                 ending_offset = oap->oap_obj_off +
2585                                                 oap->oap_count;
2586                         else
2587                                 LASSERT(oap->oap_page_off + oap->oap_count ==
2588                                         PAGE_SIZE);
2589                 }
2590                 if (ext->oe_ndelay)
2591                         ndelay = true;
2592         }
2593
2594         /* first page in the list */
2595         oap = list_first_entry(&rpc_list, typeof(*oap), oap_rpc_item);
2596
2597         crattr = &osc_env_info(env)->oti_req_attr;
2598         memset(crattr, 0, sizeof(*crattr));
2599         crattr->cra_type = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2600         crattr->cra_flags = ~0ULL;
2601         crattr->cra_page = oap2cl_page(oap);
2602         crattr->cra_oa = oa;
2603         cl_req_attr_set(env, osc2cl(obj), crattr);
2604
2605         if (cmd == OBD_BRW_WRITE) {
2606                 oa->o_grant_used = grant;
2607                 if (layout_version > 0) {
2608                         CDEBUG(D_LAYOUT, DFID": write with layout version %u\n",
2609                                PFID(&oa->o_oi.oi_fid), layout_version);
2610
2611                         oa->o_layout_version = layout_version;
2612                         oa->o_valid |= OBD_MD_LAYOUT_VERSION;
2613                 }
2614         }
2615
2616         sort_brw_pages(pga, page_count);
2617         rc = osc_brw_prep_request(cmd, cli, oa, page_count, pga, &req, 0);
2618         if (rc != 0) {
2619                 CERROR("prep_req failed: %d\n", rc);
2620                 GOTO(out, rc);
2621         }
2622
2623         req->rq_commit_cb = brw_commit;
2624         req->rq_interpret_reply = brw_interpret;
2625         req->rq_memalloc = mem_tight != 0;
2626         oap->oap_request = ptlrpc_request_addref(req);
2627         if (ndelay) {
2628                 req->rq_no_resend = req->rq_no_delay = 1;
2629                 /* probably set a shorter timeout value.
2630                  * to handle ETIMEDOUT in brw_interpret() correctly. */
2631                 /* lustre_msg_set_timeout(req, req->rq_timeout / 2); */
2632         }
2633
2634         /* Need to update the timestamps after the request is built in case
2635          * we race with setattr (locally or in queue at OST).  If OST gets
2636          * later setattr before earlier BRW (as determined by the request xid),
2637          * the OST will not use BRW timestamps.  Sadly, there is no obvious
2638          * way to do this in a single call.  bug 10150 */
2639         body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
2640         crattr->cra_oa = &body->oa;
2641         crattr->cra_flags = OBD_MD_FLMTIME | OBD_MD_FLCTIME | OBD_MD_FLATIME;
2642         cl_req_attr_set(env, osc2cl(obj), crattr);
2643         lustre_msg_set_jobid(req->rq_reqmsg, crattr->cra_jobid);
2644
2645         aa = ptlrpc_req_async_args(aa, req);
2646         INIT_LIST_HEAD(&aa->aa_oaps);
2647         list_splice_init(&rpc_list, &aa->aa_oaps);
2648         INIT_LIST_HEAD(&aa->aa_exts);
2649         list_splice_init(ext_list, &aa->aa_exts);
2650
2651         spin_lock(&cli->cl_loi_list_lock);
2652         starting_offset >>= PAGE_SHIFT;
2653         if (cmd == OBD_BRW_READ) {
2654                 cli->cl_r_in_flight++;
2655                 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2656                 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2657                 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2658                                       starting_offset + 1);
2659         } else {
2660                 cli->cl_w_in_flight++;
2661                 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2662                 lprocfs_oh_tally(&cli->cl_write_rpc_hist, cli->cl_w_in_flight);
2663                 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2664                                       starting_offset + 1);
2665         }
2666         spin_unlock(&cli->cl_loi_list_lock);
2667
2668         DEBUG_REQ(D_INODE, req, "%d pages, aa %p, now %ur/%uw in flight",
2669                   page_count, aa, cli->cl_r_in_flight,
2670                   cli->cl_w_in_flight);
2671         OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_DELAY_IO, cfs_fail_val);
2672
2673         ptlrpcd_add_req(req);
2674         rc = 0;
2675         EXIT;
2676
2677 out:
2678         if (mem_tight)
2679                 memalloc_noreclaim_restore(mpflag);
2680
2681         if (rc != 0) {
2682                 LASSERT(req == NULL);
2683
2684                 if (oa)
2685                         OBD_SLAB_FREE_PTR(oa, osc_obdo_kmem);
2686                 if (pga) {
2687                         osc_release_bounce_pages(pga, page_count);
2688                         osc_release_ppga(pga, page_count);
2689                 }
2690                 /* this should happen rarely and is pretty bad, it makes the
2691                  * pending list not follow the dirty order
2692                  */
2693                 while ((ext = list_first_entry_or_null(ext_list,
2694                                                        struct osc_extent,
2695                                                        oe_link)) != NULL) {
2696                         list_del_init(&ext->oe_link);
2697                         osc_extent_finish(env, ext, 0, rc);
2698                 }
2699         }
2700         RETURN(rc);
2701 }
2702
2703 /* This is to refresh our lock in face of no RPCs. */
2704 void osc_send_empty_rpc(struct osc_object *osc, pgoff_t start)
2705 {
2706         struct ptlrpc_request *req;
2707         struct obdo oa;
2708         struct brw_page bpg = { .off = start, .count = 1};
2709         struct brw_page *pga = &bpg;
2710         int rc;
2711
2712         memset(&oa, 0, sizeof(oa));
2713         oa.o_oi = osc->oo_oinfo->loi_oi;
2714         oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP | OBD_MD_FLFLAGS;
2715         /* For updated servers - don't do a read */
2716         oa.o_flags = OBD_FL_NORPC;
2717
2718         rc = osc_brw_prep_request(OBD_BRW_READ, osc_cli(osc), &oa, 1, &pga,
2719                                   &req, 0);
2720
2721         /* If we succeeded we ship it off, if not there's no point in doing
2722          * anything. Also no resends.
2723          * No interpret callback, no commit callback.
2724          */
2725         if (!rc) {
2726                 req->rq_no_resend = 1;
2727                 ptlrpcd_add_req(req);
2728         }
2729 }
2730
2731 static int osc_set_lock_data(struct ldlm_lock *lock, void *data)
2732 {
2733         int set = 0;
2734
2735         LASSERT(lock != NULL);
2736
2737         lock_res_and_lock(lock);
2738
2739         if (lock->l_ast_data == NULL)
2740                 lock->l_ast_data = data;
2741         if (lock->l_ast_data == data)
2742                 set = 1;
2743
2744         unlock_res_and_lock(lock);
2745
2746         return set;
2747 }
2748
2749 int osc_enqueue_fini(struct ptlrpc_request *req, osc_enqueue_upcall_f upcall,
2750                      void *cookie, struct lustre_handle *lockh,
2751                      enum ldlm_mode mode, __u64 *flags, bool speculative,
2752                      int errcode)
2753 {
2754         bool intent = *flags & LDLM_FL_HAS_INTENT;
2755         int rc;
2756         ENTRY;
2757
2758         /* The request was created before ldlm_cli_enqueue call. */
2759         if (intent && errcode == ELDLM_LOCK_ABORTED) {
2760                 struct ldlm_reply *rep;
2761
2762                 rep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
2763                 LASSERT(rep != NULL);
2764
2765                 rep->lock_policy_res1 =
2766                         ptlrpc_status_ntoh(rep->lock_policy_res1);
2767                 if (rep->lock_policy_res1)
2768                         errcode = rep->lock_policy_res1;
2769                 if (!speculative)
2770                         *flags |= LDLM_FL_LVB_READY;
2771         } else if (errcode == ELDLM_OK) {
2772                 *flags |= LDLM_FL_LVB_READY;
2773         }
2774
2775         /* Call the update callback. */
2776         rc = (*upcall)(cookie, lockh, errcode);
2777
2778         /* release the reference taken in ldlm_cli_enqueue() */
2779         if (errcode == ELDLM_LOCK_MATCHED)
2780                 errcode = ELDLM_OK;
2781         if (errcode == ELDLM_OK && lustre_handle_is_used(lockh))
2782                 ldlm_lock_decref(lockh, mode);
2783
2784         RETURN(rc);
2785 }
2786
2787 int osc_enqueue_interpret(const struct lu_env *env, struct ptlrpc_request *req,
2788                           void *args, int rc)
2789 {
2790         struct osc_enqueue_args *aa = args;
2791         struct ldlm_lock *lock;
2792         struct lustre_handle *lockh = &aa->oa_lockh;
2793         enum ldlm_mode mode = aa->oa_mode;
2794         struct ost_lvb *lvb = aa->oa_lvb;
2795         __u32 lvb_len = sizeof(*lvb);
2796         __u64 flags = 0;
2797         struct ldlm_enqueue_info einfo = {
2798                 .ei_type = aa->oa_type,
2799                 .ei_mode = mode,
2800         };
2801
2802         ENTRY;
2803
2804         /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2805          * be valid. */
2806         lock = ldlm_handle2lock(lockh);
2807         LASSERTF(lock != NULL,
2808                  "lockh %#llx, req %p, aa %p - client evicted?\n",
2809                  lockh->cookie, req, aa);
2810
2811         /* Take an additional reference so that a blocking AST that
2812          * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2813          * to arrive after an upcall has been executed by
2814          * osc_enqueue_fini(). */
2815         ldlm_lock_addref(lockh, mode);
2816
2817         /* Let cl_lock_state_wait fail with -ERESTARTSYS to unuse sublocks. */
2818         OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_ENQUEUE_HANG, 2);
2819
2820         /* Let CP AST to grant the lock first. */
2821         OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
2822
2823         if (aa->oa_speculative) {
2824                 LASSERT(aa->oa_lvb == NULL);
2825                 LASSERT(aa->oa_flags == NULL);
2826                 aa->oa_flags = &flags;
2827         }
2828
2829         /* Complete obtaining the lock procedure. */
2830         rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, &einfo, 1, aa->oa_flags,
2831                                    lvb, lvb_len, lockh, rc, false);
2832         /* Complete osc stuff. */
2833         rc = osc_enqueue_fini(req, aa->oa_upcall, aa->oa_cookie, lockh, mode,
2834                               aa->oa_flags, aa->oa_speculative, rc);
2835
2836         OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
2837
2838         ldlm_lock_decref(lockh, mode);
2839         LDLM_LOCK_PUT(lock);
2840         RETURN(rc);
2841 }
2842
2843 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2844  * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2845  * other synchronous requests, however keeping some locks and trying to obtain
2846  * others may take a considerable amount of time in a case of ost failure; and
2847  * when other sync requests do not get released lock from a client, the client
2848  * is evicted from the cluster -- such scenarious make the life difficult, so
2849  * release locks just after they are obtained. */
2850 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2851                      __u64 *flags, union ldlm_policy_data *policy,
2852                      struct ost_lvb *lvb, osc_enqueue_upcall_f upcall,
2853                      void *cookie, struct ldlm_enqueue_info *einfo,
2854                      struct ptlrpc_request_set *rqset, int async,
2855                      bool speculative)
2856 {
2857         struct obd_device *obd = exp->exp_obd;
2858         struct lustre_handle lockh = { 0 };
2859         struct ptlrpc_request *req = NULL;
2860         int intent = *flags & LDLM_FL_HAS_INTENT;
2861         __u64 match_flags = *flags;
2862         enum ldlm_mode mode;
2863         int rc;
2864         ENTRY;
2865
2866         /* Filesystem lock extents are extended to page boundaries so that
2867          * dealing with the page cache is a little smoother.  */
2868         policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
2869         policy->l_extent.end |= ~PAGE_MASK;
2870
2871         /* Next, search for already existing extent locks that will cover us */
2872         /* If we're trying to read, we also search for an existing PW lock.  The
2873          * VFS and page cache already protect us locally, so lots of readers/
2874          * writers can share a single PW lock.
2875          *
2876          * There are problems with conversion deadlocks, so instead of
2877          * converting a read lock to a write lock, we'll just enqueue a new
2878          * one.
2879          *
2880          * At some point we should cancel the read lock instead of making them
2881          * send us a blocking callback, but there are problems with canceling
2882          * locks out from other users right now, too. */
2883         mode = einfo->ei_mode;
2884         if (einfo->ei_mode == LCK_PR)
2885                 mode |= LCK_PW;
2886         /* Normal lock requests must wait for the LVB to be ready before
2887          * matching a lock; speculative lock requests do not need to,
2888          * because they will not actually use the lock. */
2889         if (!speculative)
2890                 match_flags |= LDLM_FL_LVB_READY;
2891         if (intent != 0)
2892                 match_flags |= LDLM_FL_BLOCK_GRANTED;
2893         mode = ldlm_lock_match(obd->obd_namespace, match_flags, res_id,
2894                                einfo->ei_type, policy, mode, &lockh);
2895         if (mode) {
2896                 struct ldlm_lock *matched;
2897
2898                 if (*flags & LDLM_FL_TEST_LOCK)
2899                         RETURN(ELDLM_OK);
2900
2901                 matched = ldlm_handle2lock(&lockh);
2902                 if (speculative) {
2903                         /* This DLM lock request is speculative, and does not
2904                          * have an associated IO request. Therefore if there
2905                          * is already a DLM lock, it wll just inform the
2906                          * caller to cancel the request for this stripe.*/
2907                         lock_res_and_lock(matched);
2908                         if (ldlm_extent_equal(&policy->l_extent,
2909                             &matched->l_policy_data.l_extent))
2910                                 rc = -EEXIST;
2911                         else
2912                                 rc = -ECANCELED;
2913                         unlock_res_and_lock(matched);
2914
2915                         ldlm_lock_decref(&lockh, mode);
2916                         LDLM_LOCK_PUT(matched);
2917                         RETURN(rc);
2918                 } else if (osc_set_lock_data(matched, einfo->ei_cbdata)) {
2919                         *flags |= LDLM_FL_LVB_READY;
2920
2921                         /* We already have a lock, and it's referenced. */
2922                         (*upcall)(cookie, &lockh, ELDLM_LOCK_MATCHED);
2923
2924                         ldlm_lock_decref(&lockh, mode);
2925                         LDLM_LOCK_PUT(matched);
2926                         RETURN(ELDLM_OK);
2927                 } else {
2928                         ldlm_lock_decref(&lockh, mode);
2929                         LDLM_LOCK_PUT(matched);
2930                 }
2931         }
2932
2933         if (*flags & (LDLM_FL_TEST_LOCK | LDLM_FL_MATCH_LOCK))
2934                 RETURN(-ENOLCK);
2935
2936         /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2937         *flags &= ~LDLM_FL_BLOCK_GRANTED;
2938
2939         rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
2940                               sizeof(*lvb), LVB_T_OST, &lockh, async);
2941         if (async) {
2942                 if (!rc) {
2943                         struct osc_enqueue_args *aa;
2944                         aa = ptlrpc_req_async_args(aa, req);
2945                         aa->oa_exp         = exp;
2946                         aa->oa_mode        = einfo->ei_mode;
2947                         aa->oa_type        = einfo->ei_type;
2948                         lustre_handle_copy(&aa->oa_lockh, &lockh);
2949                         aa->oa_upcall      = upcall;
2950                         aa->oa_cookie      = cookie;
2951                         aa->oa_speculative = speculative;
2952                         if (!speculative) {
2953                                 aa->oa_flags  = flags;
2954                                 aa->oa_lvb    = lvb;
2955                         } else {
2956                                 /* speculative locks are essentially to enqueue
2957                                  * a DLM lock  in advance, so we don't care
2958                                  * about the result of the enqueue. */
2959                                 aa->oa_lvb    = NULL;
2960                                 aa->oa_flags  = NULL;
2961                         }
2962
2963                         req->rq_interpret_reply = osc_enqueue_interpret;
2964                         ptlrpc_set_add_req(rqset, req);
2965                 }
2966                 RETURN(rc);
2967         }
2968
2969         rc = osc_enqueue_fini(req, upcall, cookie, &lockh, einfo->ei_mode,
2970                               flags, speculative, rc);
2971
2972         RETURN(rc);
2973 }
2974
2975 int osc_match_base(const struct lu_env *env, struct obd_export *exp,
2976                    struct ldlm_res_id *res_id, enum ldlm_type type,
2977                    union ldlm_policy_data *policy, enum ldlm_mode mode,
2978                    __u64 *flags, struct osc_object *obj,
2979                    struct lustre_handle *lockh, enum ldlm_match_flags match_flags)
2980 {
2981         struct obd_device *obd = exp->exp_obd;
2982         __u64 lflags = *flags;
2983         enum ldlm_mode rc;
2984         ENTRY;
2985
2986         if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
2987                 RETURN(-EIO);
2988
2989         /* Filesystem lock extents are extended to page boundaries so that
2990          * dealing with the page cache is a little smoother */
2991         policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
2992         policy->l_extent.end |= ~PAGE_MASK;
2993
2994         /* Next, search for already existing extent locks that will cover us */
2995         rc = ldlm_lock_match_with_skip(obd->obd_namespace, lflags, 0,
2996                                         res_id, type, policy, mode, lockh,
2997                                         match_flags);
2998         if (rc == 0 || lflags & LDLM_FL_TEST_LOCK)
2999                 RETURN(rc);
3000
3001         if (obj != NULL) {
3002                 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
3003
3004                 LASSERT(lock != NULL);
3005                 if (osc_set_lock_data(lock, obj)) {
3006                         lock_res_and_lock(lock);
3007                         if (!ldlm_is_lvb_cached(lock)) {
3008                                 LASSERT(lock->l_ast_data == obj);
3009                                 osc_lock_lvb_update(env, obj, lock, NULL);
3010                                 ldlm_set_lvb_cached(lock);
3011                         }
3012                         unlock_res_and_lock(lock);
3013                 } else {
3014                         ldlm_lock_decref(lockh, rc);
3015                         rc = 0;
3016                 }
3017                 LDLM_LOCK_PUT(lock);
3018         }
3019         RETURN(rc);
3020 }
3021
3022 static int osc_statfs_interpret(const struct lu_env *env,
3023                                 struct ptlrpc_request *req, void *args, int rc)
3024 {
3025         struct osc_async_args *aa = args;
3026         struct obd_statfs *msfs;
3027
3028         ENTRY;
3029         if (rc == -EBADR)
3030                 /*
3031                  * The request has in fact never been sent due to issues at
3032                  * a higher level (LOV).  Exit immediately since the caller
3033                  * is aware of the problem and takes care of the clean up.
3034                  */
3035                 RETURN(rc);
3036
3037         if ((rc == -ENOTCONN || rc == -EAGAIN) &&
3038             (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
3039                 GOTO(out, rc = 0);
3040
3041         if (rc != 0)
3042                 GOTO(out, rc);
3043
3044         msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3045         if (msfs == NULL)
3046                 GOTO(out, rc = -EPROTO);
3047
3048         *aa->aa_oi->oi_osfs = *msfs;
3049 out:
3050         rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
3051
3052         RETURN(rc);
3053 }
3054
3055 static int osc_statfs_async(struct obd_export *exp,
3056                             struct obd_info *oinfo, time64_t max_age,
3057                             struct ptlrpc_request_set *rqset)
3058 {
3059         struct obd_device     *obd = class_exp2obd(exp);
3060         struct ptlrpc_request *req;
3061         struct osc_async_args *aa;
3062         int rc;
3063         ENTRY;
3064
3065         if (obd->obd_osfs_age >= max_age) {
3066                 CDEBUG(D_SUPER,
3067                        "%s: use %p cache blocks %llu/%llu objects %llu/%llu\n",
3068                        obd->obd_name, &obd->obd_osfs,
3069                        obd->obd_osfs.os_bavail, obd->obd_osfs.os_blocks,
3070                        obd->obd_osfs.os_ffree, obd->obd_osfs.os_files);
3071                 spin_lock(&obd->obd_osfs_lock);
3072                 memcpy(oinfo->oi_osfs, &obd->obd_osfs, sizeof(*oinfo->oi_osfs));
3073                 spin_unlock(&obd->obd_osfs_lock);
3074                 oinfo->oi_flags |= OBD_STATFS_FROM_CACHE;
3075                 if (oinfo->oi_cb_up)
3076                         oinfo->oi_cb_up(oinfo, 0);
3077
3078                 RETURN(0);
3079         }
3080
3081         /* We could possibly pass max_age in the request (as an absolute
3082          * timestamp or a "seconds.usec ago") so the target can avoid doing
3083          * extra calls into the filesystem if that isn't necessary (e.g.
3084          * during mount that would help a bit).  Having relative timestamps
3085          * is not so great if request processing is slow, while absolute
3086          * timestamps are not ideal because they need time synchronization. */
3087         req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
3088         if (req == NULL)
3089                 RETURN(-ENOMEM);
3090
3091         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3092         if (rc) {
3093                 ptlrpc_request_free(req);
3094                 RETURN(rc);
3095         }
3096         ptlrpc_request_set_replen(req);
3097         req->rq_request_portal = OST_CREATE_PORTAL;
3098         ptlrpc_at_set_req_timeout(req);
3099
3100         if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
3101                 /* procfs requests not want stat in wait for avoid deadlock */
3102                 req->rq_no_resend = 1;
3103                 req->rq_no_delay = 1;
3104         }
3105
3106         req->rq_interpret_reply = osc_statfs_interpret;
3107         aa = ptlrpc_req_async_args(aa, req);
3108         aa->aa_oi = oinfo;
3109
3110         ptlrpc_set_add_req(rqset, req);
3111         RETURN(0);
3112 }
3113
3114 static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
3115                       struct obd_statfs *osfs, time64_t max_age, __u32 flags)
3116 {
3117         struct obd_device     *obd = class_exp2obd(exp);
3118         struct obd_statfs     *msfs;
3119         struct ptlrpc_request *req;
3120         struct obd_import     *imp, *imp0;
3121         int rc;
3122         ENTRY;
3123
3124         /*Since the request might also come from lprocfs, so we need
3125          *sync this with client_disconnect_export Bug15684
3126          */
3127         with_imp_locked(obd, imp0, rc)
3128                 imp = class_import_get(imp0);
3129         if (rc)
3130                 RETURN(rc);
3131
3132         /* We could possibly pass max_age in the request (as an absolute
3133          * timestamp or a "seconds.usec ago") so the target can avoid doing
3134          * extra calls into the filesystem if that isn't necessary (e.g.
3135          * during mount that would help a bit).  Having relative timestamps
3136          * is not so great if request processing is slow, while absolute
3137          * timestamps are not ideal because they need time synchronization. */
3138         req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
3139
3140         class_import_put(imp);
3141
3142         if (req == NULL)
3143                 RETURN(-ENOMEM);
3144
3145         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3146         if (rc) {
3147                 ptlrpc_request_free(req);
3148                 RETURN(rc);
3149         }
3150         ptlrpc_request_set_replen(req);
3151         req->rq_request_portal = OST_CREATE_PORTAL;
3152         ptlrpc_at_set_req_timeout(req);
3153
3154         if (flags & OBD_STATFS_NODELAY) {
3155                 /* procfs requests not want stat in wait for avoid deadlock */
3156                 req->rq_no_resend = 1;
3157                 req->rq_no_delay = 1;
3158         }
3159
3160         rc = ptlrpc_queue_wait(req);
3161         if (rc)
3162                 GOTO(out, rc);
3163
3164         msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3165         if (msfs == NULL)
3166                 GOTO(out, rc = -EPROTO);
3167
3168         *osfs = *msfs;
3169
3170         EXIT;
3171 out:
3172         ptlrpc_req_finished(req);
3173         return rc;
3174 }
3175
3176 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
3177                          void *karg, void __user *uarg)
3178 {
3179         struct obd_device *obd = exp->exp_obd;
3180         struct obd_ioctl_data *data = karg;
3181         int rc = 0;
3182
3183         ENTRY;
3184         if (!try_module_get(THIS_MODULE)) {
3185                 CERROR("%s: cannot get module '%s'\n", obd->obd_name,
3186                        module_name(THIS_MODULE));
3187                 return -EINVAL;
3188         }
3189         switch (cmd) {
3190         case OBD_IOC_CLIENT_RECOVER:
3191                 rc = ptlrpc_recover_import(obd->u.cli.cl_import,
3192                                            data->ioc_inlbuf1, 0);
3193                 if (rc > 0)
3194                         rc = 0;
3195                 break;
3196         case IOC_OSC_SET_ACTIVE:
3197                 rc = ptlrpc_set_import_active(obd->u.cli.cl_import,
3198                                               data->ioc_offset);
3199                 break;
3200         default:
3201                 rc = -ENOTTY;
3202                 CDEBUG(D_INODE, "%s: unrecognised ioctl %#x by %s: rc = %d\n",
3203                        obd->obd_name, cmd, current->comm, rc);
3204                 break;
3205         }
3206
3207         module_put(THIS_MODULE);
3208         return rc;
3209 }
3210
3211 int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
3212                        u32 keylen, void *key, u32 vallen, void *val,
3213                        struct ptlrpc_request_set *set)
3214 {
3215         struct ptlrpc_request *req;
3216         struct obd_device     *obd = exp->exp_obd;
3217         struct obd_import     *imp = class_exp2cliimp(exp);
3218         char                  *tmp;
3219         int                    rc;
3220         ENTRY;
3221
3222         OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3223
3224         if (KEY_IS(KEY_CHECKSUM)) {
3225                 if (vallen != sizeof(int))
3226                         RETURN(-EINVAL);
3227                 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3228                 RETURN(0);
3229         }
3230
3231         if (KEY_IS(KEY_SPTLRPC_CONF)) {
3232                 sptlrpc_conf_client_adapt(obd);
3233                 RETURN(0);
3234         }
3235
3236         if (KEY_IS(KEY_FLUSH_CTX)) {
3237                 sptlrpc_import_flush_my_ctx(imp);
3238                 RETURN(0);
3239         }
3240
3241         if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
3242                 struct client_obd *cli = &obd->u.cli;
3243                 long nr = atomic_long_read(&cli->cl_lru_in_list) >> 1;
3244                 long target = *(long *)val;
3245
3246                 nr = osc_lru_shrink(env, cli, min(nr, target), true);
3247                 *(long *)val -= nr;
3248                 RETURN(0);
3249         }
3250
3251         if (!set && !KEY_IS(KEY_GRANT_SHRINK))
3252                 RETURN(-EINVAL);
3253
3254         /* We pass all other commands directly to OST. Since nobody calls osc
3255            methods directly and everybody is supposed to go through LOV, we
3256            assume lov checked invalid values for us.
3257            The only recognised values so far are evict_by_nid and mds_conn.
3258            Even if something bad goes through, we'd get a -EINVAL from OST
3259            anyway. */
3260
3261         req = ptlrpc_request_alloc(imp, KEY_IS(KEY_GRANT_SHRINK) ?
3262                                                 &RQF_OST_SET_GRANT_INFO :
3263                                                 &RQF_OBD_SET_INFO);
3264         if (req == NULL)
3265                 RETURN(-ENOMEM);
3266
3267         req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3268                              RCL_CLIENT, keylen);
3269         if (!KEY_IS(KEY_GRANT_SHRINK))
3270                 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
3271                                      RCL_CLIENT, vallen);
3272         rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
3273         if (rc) {
3274                 ptlrpc_request_free(req);
3275                 RETURN(rc);
3276         }
3277
3278         tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3279         memcpy(tmp, key, keylen);
3280         tmp = req_capsule_client_get(&req->rq_pill, KEY_IS(KEY_GRANT_SHRINK) ?
3281                                                         &RMF_OST_BODY :
3282                                                         &RMF_SETINFO_VAL);
3283         memcpy(tmp, val, vallen);
3284
3285         if (KEY_IS(KEY_GRANT_SHRINK)) {
3286                 struct osc_grant_args *aa;
3287                 struct obdo *oa;
3288
3289                 aa = ptlrpc_req_async_args(aa, req);
3290                 OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
3291                 if (!oa) {
3292                         ptlrpc_req_finished(req);
3293                         RETURN(-ENOMEM);
3294                 }
3295                 *oa = ((struct ost_body *)val)->oa;
3296                 aa->aa_oa = oa;
3297                 req->rq_interpret_reply = osc_shrink_grant_interpret;
3298         }
3299
3300         ptlrpc_request_set_replen(req);
3301         if (!KEY_IS(KEY_GRANT_SHRINK)) {
3302                 LASSERT(set != NULL);
3303                 ptlrpc_set_add_req(set, req);
3304                 ptlrpc_check_set(NULL, set);
3305         } else {
3306                 ptlrpcd_add_req(req);
3307         }
3308
3309         RETURN(0);
3310 }
3311 EXPORT_SYMBOL(osc_set_info_async);
3312
3313 int osc_reconnect(const struct lu_env *env, struct obd_export *exp,
3314                   struct obd_device *obd, struct obd_uuid *cluuid,
3315                   struct obd_connect_data *data, void *localdata)
3316 {
3317         struct client_obd *cli = &obd->u.cli;
3318
3319         if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3320                 long lost_grant;
3321                 long grant;
3322
3323                 spin_lock(&cli->cl_loi_list_lock);
3324                 grant = cli->cl_avail_grant + cli->cl_reserved_grant;
3325                 if (data->ocd_connect_flags & OBD_CONNECT_GRANT_PARAM) {
3326                         /* restore ocd_grant_blkbits as client page bits */
3327                         data->ocd_grant_blkbits = PAGE_SHIFT;
3328                         grant += cli->cl_dirty_grant;
3329                 } else {
3330                         grant += cli->cl_dirty_pages << PAGE_SHIFT;
3331                 }
3332                 data->ocd_grant = grant ? : 2 * cli_brw_size(obd);
3333                 lost_grant = cli->cl_lost_grant;
3334                 cli->cl_lost_grant = 0;
3335                 spin_unlock(&cli->cl_loi_list_lock);
3336
3337                 CDEBUG(D_RPCTRACE, "ocd_connect_flags: %#llx ocd_version: %d"
3338                        " ocd_grant: %d, lost: %ld.\n", data->ocd_connect_flags,
3339                        data->ocd_version, data->ocd_grant, lost_grant);
3340         }
3341
3342         RETURN(0);
3343 }
3344 EXPORT_SYMBOL(osc_reconnect);
3345
3346 int osc_disconnect(struct obd_export *exp)
3347 {
3348         struct obd_device *obd = class_exp2obd(exp);
3349         int rc;
3350
3351         rc = client_disconnect_export(exp);
3352         /**
3353          * Initially we put del_shrink_grant before disconnect_export, but it
3354          * causes the following problem if setup (connect) and cleanup
3355          * (disconnect) are tangled together.
3356          *      connect p1                     disconnect p2
3357          *   ptlrpc_connect_import
3358          *     ...............               class_manual_cleanup
3359          *                                     osc_disconnect
3360          *                                     del_shrink_grant
3361          *   ptlrpc_connect_interrupt
3362          *     osc_init_grant
3363          *   add this client to shrink list
3364          *                                      cleanup_osc
3365          * Bang! grant shrink thread trigger the shrink. BUG18662
3366          */
3367         osc_del_grant_list(&obd->u.cli);
3368         return rc;
3369 }
3370 EXPORT_SYMBOL(osc_disconnect);
3371
3372 int osc_ldlm_resource_invalidate(struct cfs_hash *hs, struct cfs_hash_bd *bd,
3373                                  struct hlist_node *hnode, void *arg)
3374 {
3375         struct lu_env *env = arg;
3376         struct ldlm_resource *res = cfs_hash_object(hs, hnode);
3377         struct ldlm_lock *lock;
3378         struct osc_object *osc = NULL;
3379         ENTRY;
3380
3381         lock_res(res);
3382         list_for_each_entry(lock, &res->lr_granted, l_res_link) {
3383                 if (lock->l_ast_data != NULL && osc == NULL) {
3384                         osc = lock->l_ast_data;
3385                         cl_object_get(osc2cl(osc));
3386                 }
3387
3388                 /* clear LDLM_FL_CLEANED flag to make sure it will be canceled
3389                  * by the 2nd round of ldlm_namespace_clean() call in
3390                  * osc_import_event(). */
3391                 ldlm_clear_cleaned(lock);
3392         }
3393         unlock_res(res);
3394
3395         if (osc != NULL) {
3396                 osc_object_invalidate(env, osc);
3397                 cl_object_put(env, osc2cl(osc));
3398         }
3399
3400         RETURN(0);
3401 }
3402 EXPORT_SYMBOL(osc_ldlm_resource_invalidate);
3403
3404 static int osc_import_event(struct obd_device *obd,
3405                             struct obd_import *imp,
3406                             enum obd_import_event event)
3407 {
3408         struct client_obd *cli;
3409         int rc = 0;
3410
3411         ENTRY;
3412         LASSERT(imp->imp_obd == obd);
3413
3414         switch (event) {
3415         case IMP_EVENT_DISCON: {
3416                 cli = &obd->u.cli;
3417                 spin_lock(&cli->cl_loi_list_lock);
3418                 cli->cl_avail_grant = 0;
3419                 cli->cl_lost_grant = 0;
3420                 spin_unlock(&cli->cl_loi_list_lock);
3421                 break;
3422         }
3423         case IMP_EVENT_INACTIVE: {
3424                 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE);
3425                 break;
3426         }
3427         case IMP_EVENT_INVALIDATE: {
3428                 struct ldlm_namespace *ns = obd->obd_namespace;
3429                 struct lu_env         *env;
3430                 __u16                  refcheck;
3431
3432                 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3433
3434                 env = cl_env_get(&refcheck);
3435                 if (!IS_ERR(env)) {
3436                         osc_io_unplug(env, &obd->u.cli, NULL);
3437
3438                         cfs_hash_for_each_nolock(ns->ns_rs_hash,
3439                                                  osc_ldlm_resource_invalidate,
3440                                                  env, 0);
3441                         cl_env_put(env, &refcheck);
3442
3443                         ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3444                 } else
3445                         rc = PTR_ERR(env);
3446                 break;
3447         }
3448         case IMP_EVENT_ACTIVE: {
3449                 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE);
3450                 break;
3451         }
3452         case IMP_EVENT_OCD: {
3453                 struct obd_connect_data *ocd = &imp->imp_connect_data;
3454
3455                 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3456                         osc_init_grant(&obd->u.cli, ocd);
3457
3458                 /* See bug 7198 */
3459                 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3460                         imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3461
3462                 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD);
3463                 break;
3464         }
3465         case IMP_EVENT_DEACTIVATE: {
3466                 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_DEACTIVATE);
3467                 break;
3468         }
3469         case IMP_EVENT_ACTIVATE: {
3470                 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVATE);
3471                 break;
3472         }
3473         default:
3474                 CERROR("Unknown import event %d\n", event);
3475                 LBUG();
3476         }
3477         RETURN(rc);
3478 }
3479
3480 /**
3481  * Determine whether the lock can be canceled before replaying the lock
3482  * during recovery, see bug16774 for detailed information.
3483  *
3484  * \retval zero the lock can't be canceled
3485  * \retval other ok to cancel
3486  */
3487 static int osc_cancel_weight(struct ldlm_lock *lock)
3488 {
3489         /*
3490          * Cancel all unused and granted extent lock.
3491          */
3492         if (lock->l_resource->lr_type == LDLM_EXTENT &&
3493             ldlm_is_granted(lock) &&
3494             osc_ldlm_weigh_ast(lock) == 0)
3495                 RETURN(1);
3496
3497         RETURN(0);
3498 }
3499
3500 static int brw_queue_work(const struct lu_env *env, void *data)
3501 {
3502         struct client_obd *cli = data;
3503
3504         CDEBUG(D_CACHE, "Run writeback work for client obd %p.\n", cli);
3505
3506         osc_io_unplug(env, cli, NULL);
3507         RETURN(0);
3508 }
3509
3510 int osc_setup_common(struct obd_device *obd, struct lustre_cfg *lcfg)
3511 {
3512         struct client_obd *cli = &obd->u.cli;
3513         void *handler;
3514         int rc;
3515
3516         ENTRY;
3517
3518         rc = ptlrpcd_addref();
3519         if (rc)
3520                 RETURN(rc);
3521
3522         rc = client_obd_setup(obd, lcfg);
3523         if (rc)
3524                 GOTO(out_ptlrpcd, rc);
3525
3526
3527         handler = ptlrpcd_alloc_work(cli->cl_import, brw_queue_work, cli);
3528         if (IS_ERR(handler))
3529                 GOTO(out_ptlrpcd_work, rc = PTR_ERR(handler));
3530         cli->cl_writeback_work = handler;
3531
3532         handler = ptlrpcd_alloc_work(cli->cl_import, lru_queue_work, cli);
3533         if (IS_ERR(handler))
3534                 GOTO(out_ptlrpcd_work, rc = PTR_ERR(handler));
3535         cli->cl_lru_work = handler;
3536
3537         rc = osc_quota_setup(obd);
3538         if (rc)
3539                 GOTO(out_ptlrpcd_work, rc);
3540
3541         cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
3542         cli->cl_root_squash = 0;
3543         osc_update_next_shrink(cli);
3544
3545         RETURN(rc);
3546
3547 out_ptlrpcd_work:
3548         if (cli->cl_writeback_work != NULL) {
3549                 ptlrpcd_destroy_work(cli->cl_writeback_work);
3550                 cli->cl_writeback_work = NULL;
3551         }
3552         if (cli->cl_lru_work != NULL) {
3553                 ptlrpcd_destroy_work(cli->cl_lru_work);
3554                 cli->cl_lru_work = NULL;
3555         }
3556         client_obd_cleanup(obd);
3557 out_ptlrpcd:
3558         ptlrpcd_decref();
3559         RETURN(rc);
3560 }
3561 EXPORT_SYMBOL(osc_setup_common);
3562
3563 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3564 {
3565         struct client_obd *cli = &obd->u.cli;
3566         int                adding;
3567         int                added;
3568         int                req_count;
3569         int                rc;
3570
3571         ENTRY;
3572
3573         rc = osc_setup_common(obd, lcfg);
3574         if (rc < 0)
3575                 RETURN(rc);
3576
3577         rc = osc_tunables_init(obd);
3578         if (rc)
3579                 RETURN(rc);
3580
3581         /*
3582          * We try to control the total number of requests with a upper limit
3583          * osc_reqpool_maxreqcount. There might be some race which will cause
3584          * over-limit allocation, but it is fine.
3585          */
3586         req_count = atomic_read(&osc_pool_req_count);
3587         if (req_count < osc_reqpool_maxreqcount) {
3588                 adding = cli->cl_max_rpcs_in_flight + 2;
3589                 if (req_count + adding > osc_reqpool_maxreqcount)
3590                         adding = osc_reqpool_maxreqcount - req_count;
3591
3592                 added = ptlrpc_add_rqs_to_pool(osc_rq_pool, adding);
3593                 atomic_add(added, &osc_pool_req_count);
3594         }
3595
3596         ns_register_cancel(obd->obd_namespace, osc_cancel_weight);
3597
3598         spin_lock(&osc_shrink_lock);
3599         list_add_tail(&cli->cl_shrink_list, &osc_shrink_list);
3600         spin_unlock(&osc_shrink_lock);
3601         cli->cl_import->imp_idle_timeout = osc_idle_timeout;
3602         cli->cl_import->imp_idle_debug = D_HA;
3603
3604         RETURN(0);
3605 }
3606
3607 int osc_precleanup_common(struct obd_device *obd)
3608 {
3609         struct client_obd *cli = &obd->u.cli;
3610         ENTRY;
3611
3612         /* LU-464
3613          * for echo client, export may be on zombie list, wait for
3614          * zombie thread to cull it, because cli.cl_import will be
3615          * cleared in client_disconnect_export():
3616          *   class_export_destroy() -> obd_cleanup() ->
3617          *   echo_device_free() -> echo_client_cleanup() ->
3618          *   obd_disconnect() -> osc_disconnect() ->
3619          *   client_disconnect_export()
3620          */
3621         obd_zombie_barrier();
3622         if (cli->cl_writeback_work) {
3623                 ptlrpcd_destroy_work(cli->cl_writeback_work);
3624                 cli->cl_writeback_work = NULL;
3625         }
3626
3627         if (cli->cl_lru_work) {
3628                 ptlrpcd_destroy_work(cli->cl_lru_work);
3629                 cli->cl_lru_work = NULL;
3630         }
3631
3632         obd_cleanup_client_import(obd);
3633         RETURN(0);
3634 }
3635 EXPORT_SYMBOL(osc_precleanup_common);
3636
3637 static int osc_precleanup(struct obd_device *obd)
3638 {
3639         ENTRY;
3640
3641         osc_precleanup_common(obd);
3642
3643         ptlrpc_lprocfs_unregister_obd(obd);
3644         RETURN(0);
3645 }
3646
3647 int osc_cleanup_common(struct obd_device *obd)
3648 {
3649         struct client_obd *cli = &obd->u.cli;
3650         int rc;
3651
3652         ENTRY;
3653
3654         spin_lock(&osc_shrink_lock);
3655         list_del(&cli->cl_shrink_list);
3656         spin_unlock(&osc_shrink_lock);
3657
3658         /* lru cleanup */
3659         if (cli->cl_cache != NULL) {
3660                 LASSERT(atomic_read(&cli->cl_cache->ccc_users) > 0);
3661                 spin_lock(&cli->cl_cache->ccc_lru_lock);
3662                 list_del_init(&cli->cl_lru_osc);
3663                 spin_unlock(&cli->cl_cache->ccc_lru_lock);
3664                 cli->cl_lru_left = NULL;
3665                 cl_cache_decref(cli->cl_cache);
3666                 cli->cl_cache = NULL;
3667         }
3668
3669         /* free memory of osc quota cache */
3670         osc_quota_cleanup(obd);
3671
3672         rc = client_obd_cleanup(obd);
3673
3674         ptlrpcd_decref();
3675         RETURN(rc);
3676 }
3677 EXPORT_SYMBOL(osc_cleanup_common);
3678
3679 static const struct obd_ops osc_obd_ops = {
3680         .o_owner                = THIS_MODULE,
3681         .o_setup                = osc_setup,
3682         .o_precleanup           = osc_precleanup,
3683         .o_cleanup              = osc_cleanup_common,
3684         .o_add_conn             = client_import_add_conn,
3685         .o_del_conn             = client_import_del_conn,
3686         .o_connect              = client_connect_import,
3687         .o_reconnect            = osc_reconnect,
3688         .o_disconnect           = osc_disconnect,
3689         .o_statfs               = osc_statfs,
3690         .o_statfs_async         = osc_statfs_async,
3691         .o_create               = osc_create,
3692         .o_destroy              = osc_destroy,
3693         .o_getattr              = osc_getattr,
3694         .o_setattr              = osc_setattr,
3695         .o_iocontrol            = osc_iocontrol,
3696         .o_set_info_async       = osc_set_info_async,
3697         .o_import_event         = osc_import_event,
3698         .o_quotactl             = osc_quotactl,
3699 };
3700
3701 LIST_HEAD(osc_shrink_list);
3702 DEFINE_SPINLOCK(osc_shrink_lock);
3703
3704 #ifdef HAVE_SHRINKER_COUNT
3705 static struct shrinker osc_cache_shrinker = {
3706         .count_objects  = osc_cache_shrink_count,
3707         .scan_objects   = osc_cache_shrink_scan,
3708         .seeks          = DEFAULT_SEEKS,
3709 };
3710 #else
3711 static int osc_cache_shrink(struct shrinker *shrinker,
3712                             struct shrink_control *sc)
3713 {
3714         (void)osc_cache_shrink_scan(shrinker, sc);
3715
3716         return osc_cache_shrink_count(shrinker, sc);
3717 }
3718
3719 static struct shrinker osc_cache_shrinker = {
3720         .shrink   = osc_cache_shrink,
3721         .seeks    = DEFAULT_SEEKS,
3722 };
3723 #endif
3724
3725 static int __init osc_init(void)
3726 {
3727         unsigned int reqpool_size;
3728         unsigned int reqsize;
3729         int rc;
3730         ENTRY;
3731
3732         /* print an address of _any_ initialized kernel symbol from this
3733          * module, to allow debugging with gdb that doesn't support data
3734          * symbols from modules.*/
3735         CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
3736
3737         rc = lu_kmem_init(osc_caches);
3738         if (rc)
3739                 RETURN(rc);
3740
3741         rc = class_register_type(&osc_obd_ops, NULL, true,
3742                                  LUSTRE_OSC_NAME, &osc_device_type);
3743         if (rc)
3744                 GOTO(out_kmem, rc);
3745
3746         rc = register_shrinker(&osc_cache_shrinker);
3747         if (rc)
3748                 GOTO(out_type, rc);
3749
3750         /* This is obviously too much memory, only prevent overflow here */
3751         if (osc_reqpool_mem_max >= 1 << 12 || osc_reqpool_mem_max == 0)
3752                 GOTO(out_shrinker, rc = -EINVAL);
3753
3754         reqpool_size = osc_reqpool_mem_max << 20;
3755
3756         reqsize = 1;
3757         while (reqsize < OST_IO_MAXREQSIZE)
3758                 reqsize = reqsize << 1;
3759
3760         /*
3761          * We don't enlarge the request count in OSC pool according to
3762          * cl_max_rpcs_in_flight. The allocation from the pool will only be
3763          * tried after normal allocation failed. So a small OSC pool won't
3764          * cause much performance degression in most of cases.
3765          */
3766         osc_reqpool_maxreqcount = reqpool_size / reqsize;
3767
3768         atomic_set(&osc_pool_req_count, 0);
3769         osc_rq_pool = ptlrpc_init_rq_pool(0, OST_IO_MAXREQSIZE,
3770                                           ptlrpc_add_rqs_to_pool);
3771
3772         if (osc_rq_pool == NULL)
3773                 GOTO(out_shrinker, rc = -ENOMEM);
3774
3775         rc = osc_start_grant_work();
3776         if (rc != 0)
3777                 GOTO(out_req_pool, rc);
3778
3779         RETURN(rc);
3780
3781 out_req_pool:
3782         ptlrpc_free_rq_pool(osc_rq_pool);
3783 out_shrinker:
3784         unregister_shrinker(&osc_cache_shrinker);
3785 out_type:
3786         class_unregister_type(LUSTRE_OSC_NAME);
3787 out_kmem:
3788         lu_kmem_fini(osc_caches);
3789
3790         RETURN(rc);
3791 }
3792
3793 static void __exit osc_exit(void)
3794 {
3795         osc_stop_grant_work();
3796         unregister_shrinker(&osc_cache_shrinker);
3797         class_unregister_type(LUSTRE_OSC_NAME);
3798         lu_kmem_fini(osc_caches);
3799         ptlrpc_free_rq_pool(osc_rq_pool);
3800 }
3801
3802 MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
3803 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3804 MODULE_VERSION(LUSTRE_VERSION_STRING);
3805 MODULE_LICENSE("GPL");
3806
3807 module_init(osc_init);
3808 module_exit(osc_exit);