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