lustre/mdc/mdc_locks.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) 2003, 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_MDC
  34
  35 #include <linux/module.h>
  36
  37 #include <obd.h>
  38 #include <obd_class.h>
  39 #include <lustre_dlm.h>
  40 #include <lustre_fid.h>
  41 #include <lustre_intent.h>
  42 #include <lustre_mdc.h>
  43 #include <lustre_net.h>
  44 #include <lustre_req_layout.h>
  45 #include <lustre_swab.h>
  46 #include <lustre_acl.h>
  47
  48 #include "mdc_internal.h"
  49
  50 struct mdc_getattr_args {
  51         struct obd_export               *ga_exp;
  52         struct md_enqueue_info          *ga_minfo;
  53 };
  54
  55 int it_open_error(int phase, struct lookup_intent *it)
  56 {
  57         if (it_disposition(it, DISP_OPEN_LEASE)) {
  58                 if (phase >= DISP_OPEN_LEASE)
  59                         return it->it_status;
  60                 else
  61                         return 0;
  62         }
  63         if (it_disposition(it, DISP_OPEN_OPEN)) {
  64                 if (phase >= DISP_OPEN_OPEN)
  65                         return it->it_status;
  66                 else
  67                         return 0;
  68         }
  69
  70         if (it_disposition(it, DISP_OPEN_CREATE)) {
  71                 if (phase >= DISP_OPEN_CREATE)
  72                         return it->it_status;
  73                 else
  74                         return 0;
  75         }
  76
  77         if (it_disposition(it, DISP_LOOKUP_EXECD)) {
  78                 if (phase >= DISP_LOOKUP_EXECD)
  79                         return it->it_status;
  80                 else
  81                         return 0;
  82         }
  83
  84         if (it_disposition(it, DISP_IT_EXECD)) {
  85                 if (phase >= DISP_IT_EXECD)
  86                         return it->it_status;
  87                 else
  88                         return 0;
  89         }
  90
  91         CERROR("it disp: %X, status: %d\n", it->it_disposition, it->it_status);
  92         LBUG();
  93
  94         return 0;
  95 }
  96 EXPORT_SYMBOL(it_open_error);
  97
  98 /* this must be called on a lockh that is known to have a referenced lock */
  99 int mdc_set_lock_data(struct obd_export *exp, const struct lustre_handle *lockh,
 100                       void *data, __u64 *bits)
 101 {
 102         struct ldlm_lock *lock;
 103         struct inode *new_inode = data;
 104
 105         ENTRY;
 106         if (bits)
 107                 *bits = 0;
 108
 109         if (!lustre_handle_is_used(lockh))
 110                 RETURN(0);
 111
 112         lock = ldlm_handle2lock(lockh);
 113
 114         LASSERT(lock != NULL);
 115         lock_res_and_lock(lock);
 116         if (lock->l_resource->lr_lvb_inode &&
 117             lock->l_resource->lr_lvb_inode != data) {
 118                 struct inode *old_inode = lock->l_resource->lr_lvb_inode;
 119
 120                 LASSERTF(old_inode->i_state & I_FREEING,
 121                          "Found existing inode %p/%lu/%u state %lu in lock: setting data to %p/%lu/%u\n",
 122                          old_inode, old_inode->i_ino, old_inode->i_generation,
 123                          old_inode->i_state,
 124                          new_inode, new_inode->i_ino, new_inode->i_generation);
 125         }
 126         lock->l_resource->lr_lvb_inode = new_inode;
 127         if (bits)
 128                 *bits = lock->l_policy_data.l_inodebits.bits;
 129
 130         unlock_res_and_lock(lock);
 131         LDLM_LOCK_PUT(lock);
 132
 133         RETURN(0);
 134 }
 135
 136 enum ldlm_mode mdc_lock_match(struct obd_export *exp, __u64 flags,
 137                               const struct lu_fid *fid, enum ldlm_type type,
 138                               union ldlm_policy_data *policy,
 139                               enum ldlm_mode mode, struct lustre_handle *lockh)
 140 {
 141         struct ldlm_res_id res_id;
 142         enum ldlm_mode rc;
 143
 144         ENTRY;
 145         fid_build_reg_res_name(fid, &res_id);
 146         /* LU-4405: Clear bits not supported by server */
 147         policy->l_inodebits.bits &= exp_connect_ibits(exp);
 148         rc = ldlm_lock_match(class_exp2obd(exp)->obd_namespace, flags,
 149                              &res_id, type, policy, mode, lockh, 0);
 150         RETURN(rc);
 151 }
 152
 153 int mdc_cancel_unused(struct obd_export *exp, const struct lu_fid *fid,
 154                       union ldlm_policy_data *policy, enum ldlm_mode mode,
 155                       enum ldlm_cancel_flags flags, void *opaque)
 156 {
 157         struct obd_device *obd = class_exp2obd(exp);
 158         struct ldlm_res_id res_id;
 159         int rc;
 160
 161         ENTRY;
 162         fid_build_reg_res_name(fid, &res_id);
 163         rc = ldlm_cli_cancel_unused_resource(obd->obd_namespace, &res_id,
 164                                              policy, mode, flags, opaque);
 165         RETURN(rc);
 166 }
 167
 168 int mdc_null_inode(struct obd_export *exp,
 169                    const struct lu_fid *fid)
 170 {
 171         struct ldlm_res_id res_id;
 172         struct ldlm_resource *res;
 173         struct ldlm_namespace *ns = class_exp2obd(exp)->obd_namespace;
 174
 175         ENTRY;
 176         LASSERTF(ns != NULL, "no namespace passed\n");
 177
 178         fid_build_reg_res_name(fid, &res_id);
 179
 180         res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
 181         if (IS_ERR(res))
 182                 RETURN(0);
 183
 184         lock_res(res);
 185         res->lr_lvb_inode = NULL;
 186         unlock_res(res);
 187
 188         ldlm_resource_putref(res);
 189         RETURN(0);
 190 }
 191
 192 static inline void mdc_clear_replay_flag(struct ptlrpc_request *req, int rc)
 193 {
 194         /* Don't hold error requests for replay. */
 195         if (req->rq_replay) {
 196                 spin_lock(&req->rq_lock);
 197                 req->rq_replay = 0;
 198                 spin_unlock(&req->rq_lock);
 199         }
 200         if (rc && req->rq_transno != 0) {
 201                 DEBUG_REQ(D_ERROR, req, "transno returned on error: rc = %d",
 202                           rc);
 203                 LBUG();
 204         }
 205 }
 206
 207 /* Save a large LOV EA into the request buffer so that it is available
 208  * for replay.  We don't do this in the initial request because the
 209  * original request doesn't need this buffer (at most it sends just the
 210  * lov_mds_md) and it is a waste of RAM/bandwidth to send the empty
 211  * buffer and may also be difficult to allocate and save a very large
 212  * request buffer for each open. (b=5707)
 213  *
 214  * OOM here may cause recovery failure if lmm is needed (only for the
 215  * original open if the MDS crashed just when this client also OOM'd)
 216  * but this is incredibly unlikely, and questionable whether the client
 217  * could do MDS recovery under OOM anyways...
 218  */
 219 int mdc_save_lovea(struct ptlrpc_request *req,
 220                    const struct req_msg_field *field, void *data, u32 size)
 221 {
 222         struct req_capsule *pill = &req->rq_pill;
 223         struct lov_user_md *lmm;
 224         int rc = 0;
 225
 226         if (req_capsule_get_size(pill, field, RCL_CLIENT) < size) {
 227                 rc = sptlrpc_cli_enlarge_reqbuf(req, field, size);
 228                 if (rc) {
 229                         CERROR("%s: Can't enlarge ea size to %d: rc = %d\n",
 230                                req->rq_export->exp_obd->obd_name,
 231                                size, rc);
 232                         return rc;
 233                 }
 234         } else {
 235                 req_capsule_shrink(pill, field, size, RCL_CLIENT);
 236         }
 237
 238         req_capsule_set_size(pill, field, RCL_CLIENT, size);
 239         lmm = req_capsule_client_get(pill, field);
 240         if (lmm) {
 241                 memcpy(lmm, data, size);
 242                 /* overwrite layout generation returned from the MDS */
 243                 lmm->lmm_stripe_offset =
 244                   (typeof(lmm->lmm_stripe_offset))LOV_OFFSET_DEFAULT;
 245         }
 246
 247         return rc;
 248 }
 249
 250 static struct ptlrpc_request *
 251 mdc_intent_open_pack(struct obd_export *exp, struct lookup_intent *it,
 252                      struct md_op_data *op_data, __u32 acl_bufsize)
 253 {
 254         struct ptlrpc_request *req;
 255         struct obd_device *obd = class_exp2obd(exp);
 256         struct ldlm_intent *lit;
 257         const void *lmm = op_data->op_data;
 258         __u32 lmmsize = op_data->op_data_size;
 259         __u32  mdt_md_capsule_size;
 260         LIST_HEAD(cancels);
 261         int count = 0;
 262         enum ldlm_mode mode;
 263         int repsize, repsize_estimate;
 264         int rc;
 265
 266         ENTRY;
 267
 268         mdt_md_capsule_size = obd->u.cli.cl_default_mds_easize;
 269
 270         it->it_create_mode = (it->it_create_mode & ~S_IFMT) | S_IFREG;
 271
 272         /* XXX: openlock is not cancelled for cross-refs. */
 273         /* If inode is known, cancel conflicting OPEN locks. */
 274         if (fid_is_sane(&op_data->op_fid2)) {
 275                 if (it->it_flags & MDS_OPEN_LEASE) { /* try to get lease */
 276                         if (it->it_flags & MDS_FMODE_WRITE)
 277                                 mode = LCK_EX;
 278                         else
 279                                 mode = LCK_PR;
 280                 } else {
 281                         if (it->it_flags & (MDS_FMODE_WRITE | MDS_OPEN_TRUNC))
 282                                 mode = LCK_CW;
 283 #ifdef FMODE_EXEC
 284                         else if (it->it_flags & FMODE_EXEC)
 285                                 mode = LCK_PR;
 286 #endif
 287                         else
 288                                 mode = LCK_CR;
 289                 }
 290                 count = mdc_resource_get_unused(exp, &op_data->op_fid2,
 291                                                 &cancels, mode,
 292                                                 MDS_INODELOCK_OPEN);
 293         }
 294
 295         /* If CREATE, cancel parent's UPDATE lock. */
 296         if (it->it_op & IT_CREAT)
 297                 mode = LCK_EX;
 298         else
 299                 mode = LCK_CR;
 300         count += mdc_resource_get_unused(exp, &op_data->op_fid1,
 301                                          &cancels, mode,
 302                                          MDS_INODELOCK_UPDATE);
 303
 304         req = ptlrpc_request_alloc(class_exp2cliimp(exp),
 305                                    &RQF_LDLM_INTENT_OPEN);
 306         if (req == NULL) {
 307                 ldlm_lock_list_put(&cancels, l_bl_ast, count);
 308                 RETURN(ERR_PTR(-ENOMEM));
 309         }
 310
 311         req_capsule_set_size(&req->rq_pill, &RMF_NAME, RCL_CLIENT,
 312                              op_data->op_namelen + 1);
 313         if (cl_is_lov_delay_create(it->it_flags)) {
 314                 /* open(O_LOV_DELAY_CREATE) won't pack lmm */
 315                 LASSERT(lmmsize == 0);
 316                 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT, 0);
 317         } else {
 318                 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT,
 319                              max(lmmsize, obd->u.cli.cl_default_mds_easize));
 320         }
 321
 322         req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
 323                              RCL_CLIENT, op_data->op_file_secctx_name != NULL ?
 324                              op_data->op_file_secctx_name_size : 0);
 325
 326         req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX, RCL_CLIENT,
 327                              op_data->op_file_secctx_size);
 328
 329         /* get SELinux policy info if any */
 330         rc = sptlrpc_get_sepol(req);
 331         if (rc < 0) {
 332                 ptlrpc_request_free(req);
 333                 RETURN(ERR_PTR(rc));
 334         }
 335         req_capsule_set_size(&req->rq_pill, &RMF_SELINUX_POL, RCL_CLIENT,
 336                              strlen(req->rq_sepol) ?
 337                              strlen(req->rq_sepol) + 1 : 0);
 338
 339         rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
 340         if (rc < 0) {
 341                 ptlrpc_request_free(req);
 342                 RETURN(ERR_PTR(rc));
 343         }
 344
 345         spin_lock(&req->rq_lock);
 346         req->rq_replay = req->rq_import->imp_replayable;
 347         spin_unlock(&req->rq_lock);
 348
 349         /* pack the intent */
 350         lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
 351         lit->opc = (__u64)it->it_op;
 352
 353         /* pack the intended request */
 354         mdc_open_pack(req, op_data, it->it_create_mode, 0, it->it_flags, lmm,
 355                       lmmsize);
 356
 357         req_capsule_set_size(&req->rq_pill, &RMF_MDT_MD, RCL_SERVER,
 358                              mdt_md_capsule_size);
 359         req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, acl_bufsize);
 360
 361         if (!(it->it_op & IT_CREAT) && it->it_op & IT_OPEN &&
 362             req_capsule_has_field(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
 363                                   RCL_CLIENT) &&
 364             op_data->op_file_secctx_name_size > 0 &&
 365             op_data->op_file_secctx_name != NULL) {
 366                 char *secctx_name;
 367
 368                 secctx_name = req_capsule_client_get(&req->rq_pill,
 369                                                      &RMF_FILE_SECCTX_NAME);
 370                 memcpy(secctx_name, op_data->op_file_secctx_name,
 371                        op_data->op_file_secctx_name_size);
 372                 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
 373                                      RCL_SERVER,
 374                                      obd->u.cli.cl_max_mds_easize);
 375
 376                 CDEBUG(D_SEC, "packed '%.*s' as security xattr name\n",
 377                        op_data->op_file_secctx_name_size,
 378                        op_data->op_file_secctx_name);
 379
 380         } else {
 381                 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
 382                                      RCL_SERVER, 0);
 383         }
 384
 385         /**
 386          * Inline buffer for possible data from Data-on-MDT files.
 387          */
 388         req_capsule_set_size(&req->rq_pill, &RMF_NIOBUF_INLINE, RCL_SERVER,
 389                              sizeof(struct niobuf_remote));
 390         ptlrpc_request_set_replen(req);
 391
 392         /* Get real repbuf allocated size as rounded up power of 2 */
 393         repsize = size_roundup_power2(req->rq_replen +
 394                                       lustre_msg_early_size());
 395         /* Estimate free space for DoM files in repbuf */
 396         repsize_estimate = repsize - (req->rq_replen -
 397                            mdt_md_capsule_size +
 398                            sizeof(struct lov_comp_md_v1) +
 399                            sizeof(struct lov_comp_md_entry_v1) +
 400                            lov_mds_md_size(0, LOV_MAGIC_V3));
 401
 402         if (repsize_estimate < obd->u.cli.cl_dom_min_inline_repsize) {
 403                 repsize = obd->u.cli.cl_dom_min_inline_repsize -
 404                           repsize_estimate + sizeof(struct niobuf_remote);
 405                 req_capsule_set_size(&req->rq_pill, &RMF_NIOBUF_INLINE,
 406                                      RCL_SERVER,
 407                                      sizeof(struct niobuf_remote) + repsize);
 408                 ptlrpc_request_set_replen(req);
 409                 CDEBUG(D_INFO, "Increase repbuf by %d bytes, total: %d\n",
 410                        repsize, req->rq_replen);
 411                 repsize = size_roundup_power2(req->rq_replen +
 412                                               lustre_msg_early_size());
 413         }
 414         /* The only way to report real allocated repbuf size to the server
 415          * is the lm_repsize but it must be set prior buffer allocation itself
 416          * due to security reasons - it is part of buffer used in signature
 417          * calculation (see LU-11414). Therefore the saved size is predicted
 418          * value as rq_replen rounded to the next higher power of 2.
 419          * Such estimation is safe. Though the final allocated buffer might
 420          * be even larger, it is not possible to know that at this point.
 421          */
 422         req->rq_reqmsg->lm_repsize = repsize;
 423         RETURN(req);
 424 }
 425
 426 #define GA_DEFAULT_EA_NAME_LEN   20
 427 #define GA_DEFAULT_EA_VAL_LEN   250
 428 #define GA_DEFAULT_EA_NUM        10
 429
 430 static struct ptlrpc_request *
 431 mdc_intent_getxattr_pack(struct obd_export *exp, struct lookup_intent *it,
 432                          struct md_op_data *op_data)
 433 {
 434         struct ptlrpc_request *req;
 435         struct ldlm_intent *lit;
 436         int rc, count = 0;
 437         LIST_HEAD(cancels);
 438         u32 ea_vals_buf_size = GA_DEFAULT_EA_VAL_LEN * GA_DEFAULT_EA_NUM;
 439
 440         ENTRY;
 441         req = ptlrpc_request_alloc(class_exp2cliimp(exp),
 442                                         &RQF_LDLM_INTENT_GETXATTR);
 443         if (req == NULL)
 444                 RETURN(ERR_PTR(-ENOMEM));
 445
 446         /* get SELinux policy info if any */
 447         rc = sptlrpc_get_sepol(req);
 448         if (rc < 0) {
 449                 ptlrpc_request_free(req);
 450                 RETURN(ERR_PTR(rc));
 451         }
 452         req_capsule_set_size(&req->rq_pill, &RMF_SELINUX_POL, RCL_CLIENT,
 453                              strlen(req->rq_sepol) ?
 454                              strlen(req->rq_sepol) + 1 : 0);
 455
 456         rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
 457         if (rc) {
 458                 ptlrpc_request_free(req);
 459                 RETURN(ERR_PTR(rc));
 460         }
 461
 462         /* pack the intent */
 463         lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
 464         lit->opc = IT_GETXATTR;
 465         CDEBUG(D_INFO, "%s: get xattrs for "DFID"\n",
 466                exp->exp_obd->obd_name, PFID(&op_data->op_fid1));
 467
 468 #if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(3, 0, 53, 0)
 469         /* If the supplied buffer is too small then the server will return
 470          * -ERANGE and llite will fallback to using non cached xattr
 471          * operations. On servers before 2.10.1 a (non-cached) listxattr RPC
 472          * for an orphan or dead file causes an oops. So let's try to avoid
 473          * sending too small a buffer to too old a server. This is effectively
 474          * undoing the memory conservation of LU-9417 when it would be *more*
 475          * likely to crash the server. See LU-9856.
 476          */
 477         if (exp->exp_connect_data.ocd_version < OBD_OCD_VERSION(2, 10, 1, 0))
 478                 ea_vals_buf_size = max_t(u32, ea_vals_buf_size,
 479                                          exp->exp_connect_data.ocd_max_easize);
 480 #endif
 481
 482         /* pack the intended request */
 483         mdc_pack_body(req, &op_data->op_fid1, op_data->op_valid,
 484                       ea_vals_buf_size, -1, 0);
 485
 486         /* get SELinux policy info if any */
 487         mdc_file_sepol_pack(req);
 488
 489         req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_SERVER,
 490                              GA_DEFAULT_EA_NAME_LEN * GA_DEFAULT_EA_NUM);
 491
 492         req_capsule_set_size(&req->rq_pill, &RMF_EAVALS, RCL_SERVER,
 493                              ea_vals_buf_size);
 494
 495         req_capsule_set_size(&req->rq_pill, &RMF_EAVALS_LENS, RCL_SERVER,
 496                              sizeof(u32) * GA_DEFAULT_EA_NUM);
 497
 498         req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, 0);
 499
 500         ptlrpc_request_set_replen(req);
 501
 502         RETURN(req);
 503 }
 504
 505 static struct ptlrpc_request *
 506 mdc_intent_getattr_pack(struct obd_export *exp, struct lookup_intent *it,
 507                         struct md_op_data *op_data, __u32 acl_bufsize)
 508 {
 509         struct ptlrpc_request *req;
 510         struct obd_device *obd = class_exp2obd(exp);
 511         u64 valid = OBD_MD_FLGETATTR | OBD_MD_FLEASIZE | OBD_MD_FLMODEASIZE |
 512                     OBD_MD_FLDIREA | OBD_MD_MEA | OBD_MD_FLACL |
 513                     OBD_MD_DEFAULT_MEA;
 514         struct ldlm_intent *lit;
 515         __u32 easize;
 516         bool have_secctx = false;
 517         int rc;
 518
 519         ENTRY;
 520         req = ptlrpc_request_alloc(class_exp2cliimp(exp),
 521                                    &RQF_LDLM_INTENT_GETATTR);
 522         if (req == NULL)
 523                 RETURN(ERR_PTR(-ENOMEM));
 524
 525         /* send name of security xattr to get upon intent */
 526         if (it->it_op & (IT_LOOKUP | IT_GETATTR) &&
 527             req_capsule_has_field(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
 528                                   RCL_CLIENT) &&
 529             op_data->op_file_secctx_name_size > 0 &&
 530             op_data->op_file_secctx_name != NULL) {
 531                 have_secctx = true;
 532                 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
 533                                      RCL_CLIENT,
 534                                      op_data->op_file_secctx_name_size);
 535         }
 536
 537         req_capsule_set_size(&req->rq_pill, &RMF_NAME, RCL_CLIENT,
 538                              op_data->op_namelen + 1);
 539
 540         rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
 541         if (rc) {
 542                 ptlrpc_request_free(req);
 543                 RETURN(ERR_PTR(rc));
 544         }
 545
 546         /* pack the intent */
 547         lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
 548         lit->opc = (__u64)it->it_op;
 549
 550         easize = obd->u.cli.cl_default_mds_easize;
 551
 552         /* pack the intended request */
 553         mdc_getattr_pack(req, valid, it->it_flags, op_data, easize);
 554
 555         req_capsule_set_size(&req->rq_pill, &RMF_MDT_MD, RCL_SERVER, easize);
 556         req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, acl_bufsize);
 557         req_capsule_set_size(&req->rq_pill, &RMF_DEFAULT_MDT_MD, RCL_SERVER,
 558                              sizeof(struct lmv_user_md));
 559
 560         if (have_secctx) {
 561                 char *secctx_name;
 562
 563                 secctx_name = req_capsule_client_get(&req->rq_pill,
 564                                                      &RMF_FILE_SECCTX_NAME);
 565                 memcpy(secctx_name, op_data->op_file_secctx_name,
 566                        op_data->op_file_secctx_name_size);
 567
 568                 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
 569                                      RCL_SERVER, easize);
 570
 571                 CDEBUG(D_SEC, "packed '%.*s' as security xattr name\n",
 572                        op_data->op_file_secctx_name_size,
 573                        op_data->op_file_secctx_name);
 574         } else {
 575                 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
 576                                      RCL_SERVER, 0);
 577         }
 578
 579         ptlrpc_request_set_replen(req);
 580         RETURN(req);
 581 }
 582
 583 static struct ptlrpc_request *mdc_intent_layout_pack(struct obd_export *exp,
 584                                                      struct lookup_intent *it,
 585                                                      struct md_op_data *op_data)
 586 {
 587         struct obd_device *obd = class_exp2obd(exp);
 588         struct ptlrpc_request *req;
 589         struct ldlm_intent *lit;
 590         struct layout_intent *layout;
 591         LIST_HEAD(cancels);
 592         int count = 0, rc;
 593
 594         ENTRY;
 595         req = ptlrpc_request_alloc(class_exp2cliimp(exp),
 596                                 &RQF_LDLM_INTENT_LAYOUT);
 597         if (req == NULL)
 598                 RETURN(ERR_PTR(-ENOMEM));
 599
 600         if (fid_is_sane(&op_data->op_fid2) && (it->it_op & IT_LAYOUT) &&
 601             (it->it_flags & FMODE_WRITE)) {
 602                 count = mdc_resource_get_unused(exp, &op_data->op_fid2,
 603                                                 &cancels, LCK_EX,
 604                                                 MDS_INODELOCK_LAYOUT);
 605         }
 606
 607         req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT, 0);
 608         rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
 609         if (rc) {
 610                 ptlrpc_request_free(req);
 611                 RETURN(ERR_PTR(rc));
 612         }
 613
 614         /* pack the intent */
 615         lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
 616         lit->opc = (__u64)it->it_op;
 617
 618         /* pack the layout intent request */
 619         layout = req_capsule_client_get(&req->rq_pill, &RMF_LAYOUT_INTENT);
 620         LASSERT(op_data->op_data != NULL);
 621         LASSERT(op_data->op_data_size == sizeof(*layout));
 622         memcpy(layout, op_data->op_data, sizeof(*layout));
 623
 624         req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
 625                              obd->u.cli.cl_default_mds_easize);
 626         ptlrpc_request_set_replen(req);
 627         RETURN(req);
 628 }
 629
 630 static struct ptlrpc_request *mdc_enqueue_pack(struct obd_export *exp,
 631                                                int lvb_len)
 632 {
 633         struct ptlrpc_request *req;
 634         int rc;
 635
 636         ENTRY;
 637         req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_LDLM_ENQUEUE);
 638         if (req == NULL)
 639                 RETURN(ERR_PTR(-ENOMEM));
 640
 641         rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
 642         if (rc) {
 643                 ptlrpc_request_free(req);
 644                 RETURN(ERR_PTR(rc));
 645         }
 646
 647         req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER, lvb_len);
 648         ptlrpc_request_set_replen(req);
 649         RETURN(req);
 650 }
 651
 652 static int mdc_finish_enqueue(struct obd_export *exp,
 653                               struct ptlrpc_request *req,
 654                               struct ldlm_enqueue_info *einfo,
 655                               struct lookup_intent *it,
 656                               struct lustre_handle *lockh, int rc)
 657 {
 658         struct req_capsule *pill = &req->rq_pill;
 659         struct ldlm_request *lockreq;
 660         struct ldlm_reply *lockrep;
 661         struct ldlm_lock *lock;
 662         struct mdt_body *body = NULL;
 663         void *lvb_data = NULL;
 664         __u32 lvb_len = 0;
 665
 666         ENTRY;
 667         LASSERT(rc >= 0);
 668         /* Similarly, if we're going to replay this request, we don't want to
 669          * actually get a lock, just perform the intent.
 670          */
 671         if (req->rq_transno || req->rq_replay) {
 672                 lockreq = req_capsule_client_get(pill, &RMF_DLM_REQ);
 673                 lockreq->lock_flags |= ldlm_flags_to_wire(LDLM_FL_INTENT_ONLY);
 674         }
 675
 676         if (rc == ELDLM_LOCK_ABORTED) {
 677                 einfo->ei_mode = 0;
 678                 memset(lockh, 0, sizeof(*lockh));
 679                 rc = 0;
 680         } else { /* rc = 0 */
 681                 lock = ldlm_handle2lock(lockh);
 682                 LASSERT(lock != NULL);
 683
 684                 /* If server returned a different lock mode, fix up variables */
 685                 if (lock->l_req_mode != einfo->ei_mode) {
 686                         ldlm_lock_addref(lockh, lock->l_req_mode);
 687                         ldlm_lock_decref(lockh, einfo->ei_mode);
 688                         einfo->ei_mode = lock->l_req_mode;
 689                 }
 690                 LDLM_LOCK_PUT(lock);
 691         }
 692
 693         lockrep = req_capsule_server_get(pill, &RMF_DLM_REP);
 694         LASSERT(lockrep != NULL); /* checked by ldlm_cli_enqueue() */
 695
 696         it->it_disposition = (int)lockrep->lock_policy_res1;
 697         it->it_status = (int)lockrep->lock_policy_res2;
 698         it->it_lock_mode = einfo->ei_mode;
 699         it->it_lock_handle = lockh->cookie;
 700         it->it_request = req;
 701
 702         /* Technically speaking rq_transno must already be zero if
 703          * it_status is in error, so the check is a bit redundant.
 704          */
 705         if ((!req->rq_transno || it->it_status < 0) && req->rq_replay)
 706                 mdc_clear_replay_flag(req, it->it_status);
 707
 708         /* If we're doing an IT_OPEN which did not result in an actual
 709          * successful open, then we need to remove the bit which saves
 710          * this request for unconditional replay.
 711          *
 712          * It's important that we do this first!  Otherwise we might exit the
 713          * function without doing so, and try to replay a failed create.
 714          * (b=3440)
 715          */
 716         if (it->it_op & IT_OPEN && req->rq_replay &&
 717             (!it_disposition(it, DISP_OPEN_OPEN) || it->it_status != 0))
 718                 mdc_clear_replay_flag(req, it->it_status);
 719
 720         DEBUG_REQ(D_RPCTRACE, req, "op=%x disposition=%x, status=%d",
 721                   it->it_op, it->it_disposition, it->it_status);
 722
 723         /* We know what to expect, so we do any byte flipping required here */
 724         if (it_has_reply_body(it)) {
 725                 body = req_capsule_server_get(pill, &RMF_MDT_BODY);
 726                 if (body == NULL) {
 727                         rc = -EPROTO;
 728                         CERROR("%s: cannot swab mdt_body: rc = %d\n",
 729                                exp->exp_obd->obd_name, rc);
 730                         RETURN(rc);
 731                 }
 732
 733                 if (it_disposition(it, DISP_OPEN_OPEN) &&
 734                     !it_open_error(DISP_OPEN_OPEN, it)) {
 735                         /*
 736                          * If this is a successful OPEN request, we need to set
 737                          * replay handler and data early, so that if replay
 738                          * happens immediately after swabbing below, new reply
 739                          * is swabbed by that handler correctly.
 740                          */
 741                         mdc_set_open_replay_data(NULL, NULL, it);
 742                 }
 743
 744                 if (it_disposition(it, DISP_OPEN_CREATE) &&
 745                     !it_open_error(DISP_OPEN_CREATE, it)) {
 746                         lprocfs_counter_incr(exp->exp_obd->obd_md_stats,
 747                                              LPROC_MD_CREATE);
 748                 }
 749
 750                 if (body->mbo_valid & (OBD_MD_FLDIREA | OBD_MD_FLEASIZE)) {
 751                         void *eadata;
 752
 753                         mdc_update_max_ea_from_body(exp, body);
 754
 755                         /*
 756                          * The eadata is opaque; just check that it is there.
 757                          * Eventually, obd_unpackmd() will check the contents.
 758                          */
 759                         eadata = req_capsule_server_sized_get(pill, &RMF_MDT_MD,
 760                                                         body->mbo_eadatasize);
 761                         if (eadata == NULL)
 762                                 RETURN(-EPROTO);
 763
 764                         /* save LVB data and length if for layout lock */
 765                         lvb_data = eadata;
 766                         lvb_len = body->mbo_eadatasize;
 767
 768                         /*
 769                          * We save the reply LOV EA in case we have to replay a
 770                          * create for recovery.  If we didn't allocate a large
 771                          * enough request buffer above we need to reallocate it
 772                          * here to hold the actual LOV EA.
 773                          *
 774                          * To not save LOV EA if request is not going to replay
 775                          * (for example error one).
 776                          */
 777                         if ((it->it_op & IT_OPEN) && req->rq_replay) {
 778                                 rc = mdc_save_lovea(req, &RMF_EADATA, eadata,
 779                                                     body->mbo_eadatasize);
 780                                 if (rc) {
 781                                         body->mbo_valid &= ~OBD_MD_FLEASIZE;
 782                                         body->mbo_eadatasize = 0;
 783                                         rc = 0;
 784                                 }
 785                         }
 786                 }
 787         } else if (it->it_op & IT_LAYOUT) {
 788                 /* maybe the lock was granted right away and layout
 789                  * is packed into RMF_DLM_LVB of req
 790                  */
 791                 lvb_len = req_capsule_get_size(pill, &RMF_DLM_LVB, RCL_SERVER);
 792                 CDEBUG(D_INFO, "%s: layout return lvb %d transno %lld\n",
 793                        class_exp2obd(exp)->obd_name, lvb_len, req->rq_transno);
 794                 if (lvb_len > 0) {
 795                         lvb_data = req_capsule_server_sized_get(pill,
 796                                                         &RMF_DLM_LVB, lvb_len);
 797                         if (lvb_data == NULL)
 798                                 RETURN(-EPROTO);
 799
 800                         /**
 801                          * save replied layout data to the request buffer for
 802                          * recovery consideration (lest MDS reinitialize
 803                          * another set of OST objects).
 804                          */
 805                         if (req->rq_transno)
 806                                 (void)mdc_save_lovea(req, &RMF_EADATA, lvb_data,
 807                                                      lvb_len);
 808                 }
 809         }
 810
 811         /* fill in stripe data for layout lock.
 812          * LU-6581: trust layout data only if layout lock is granted. The MDT
 813          * has stopped sending layout unless the layout lock is granted. The
 814          * client still does this checking in case it's talking with an old
 815          * server. - Jinshan
 816          */
 817         lock = ldlm_handle2lock(lockh);
 818         if (lock == NULL)
 819                 RETURN(rc);
 820
 821         if (ldlm_has_layout(lock) && lvb_data != NULL &&
 822             !(lockrep->lock_flags & LDLM_FL_BLOCKED_MASK)) {
 823                 void *lmm;
 824
 825                 LDLM_DEBUG(lock, "layout lock returned by: %s, lvb_len: %d",
 826                         ldlm_it2str(it->it_op), lvb_len);
 827
 828                 OBD_ALLOC_LARGE(lmm, lvb_len);
 829                 if (lmm == NULL)
 830                         GOTO(out_lock, rc = -ENOMEM);
 831
 832                 memcpy(lmm, lvb_data, lvb_len);
 833
 834                 /* install lvb_data */
 835                 lock_res_and_lock(lock);
 836                 if (lock->l_lvb_data == NULL) {
 837                         lock->l_lvb_type = LVB_T_LAYOUT;
 838                         lock->l_lvb_data = lmm;
 839                         lock->l_lvb_len = lvb_len;
 840                         lmm = NULL;
 841                 }
 842                 unlock_res_and_lock(lock);
 843                 if (lmm != NULL)
 844                         OBD_FREE_LARGE(lmm, lvb_len);
 845         }
 846
 847         if (ldlm_has_dom(lock)) {
 848                 LASSERT(lock->l_glimpse_ast == mdc_ldlm_glimpse_ast);
 849
 850                 body = req_capsule_server_get(pill, &RMF_MDT_BODY);
 851                 if (!(body->mbo_valid & OBD_MD_DOM_SIZE)) {
 852                         LDLM_ERROR(lock, "%s: DoM lock without size.",
 853                                    exp->exp_obd->obd_name);
 854                         GOTO(out_lock, rc = -EPROTO);
 855                 }
 856
 857                 LDLM_DEBUG(lock, "DoM lock is returned by: %s, size: %llu",
 858                            ldlm_it2str(it->it_op), body->mbo_dom_size);
 859
 860                 rc = mdc_fill_lvb(req, &lock->l_ost_lvb);
 861         }
 862 out_lock:
 863         LDLM_LOCK_PUT(lock);
 864
 865         RETURN(rc);
 866 }
 867
 868 static inline bool mdc_skip_mod_rpc_slot(const struct lookup_intent *it)
 869 {
 870         if (it != NULL &&
 871             (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
 872              it->it_op == IT_READDIR ||
 873              (it->it_op == IT_LAYOUT && !(it->it_flags & MDS_FMODE_WRITE))))
 874                 return true;
 875         return false;
 876 }
 877
 878 /* We always reserve enough space in the reply packet for a stripe MD, because
 879  * we don't know in advance the file type.
 880  */
 881 static int mdc_enqueue_base(struct obd_export *exp,
 882                             struct ldlm_enqueue_info *einfo,
 883                             const union ldlm_policy_data *policy,
 884                             struct lookup_intent *it,
 885                             struct md_op_data *op_data,
 886                             struct lustre_handle *lockh,
 887                             __u64 extra_lock_flags)
 888 {
 889         struct obd_device *obd = class_exp2obd(exp);
 890         struct ptlrpc_request *req;
 891         __u64 flags, saved_flags = extra_lock_flags;
 892         struct ldlm_res_id res_id;
 893         static const union ldlm_policy_data lookup_policy = {
 894                                   .l_inodebits = { MDS_INODELOCK_LOOKUP } };
 895         static const union ldlm_policy_data update_policy = {
 896                                   .l_inodebits = { MDS_INODELOCK_UPDATE } };
 897         static const union ldlm_policy_data layout_policy = {
 898                                   .l_inodebits = { MDS_INODELOCK_LAYOUT } };
 899         static const union ldlm_policy_data getxattr_policy = {
 900                                   .l_inodebits = { MDS_INODELOCK_XATTR } };
 901         int generation, resends = 0;
 902         struct ldlm_reply *lockrep;
 903         struct obd_import *imp = class_exp2cliimp(exp);
 904         __u32 acl_bufsize;
 905         enum lvb_type lvb_type = 0;
 906         int rc;
 907
 908         ENTRY;
 909         LASSERTF(!it || einfo->ei_type == LDLM_IBITS, "lock type %d\n",
 910                  einfo->ei_type);
 911         fid_build_reg_res_name(&op_data->op_fid1, &res_id);
 912
 913         if (it != NULL) {
 914                 LASSERT(policy == NULL);
 915
 916                 saved_flags |= LDLM_FL_HAS_INTENT;
 917                 if (it->it_op & (IT_GETATTR | IT_READDIR))
 918                         policy = &update_policy;
 919                 else if (it->it_op & IT_LAYOUT)
 920                         policy = &layout_policy;
 921                 else if (it->it_op & IT_GETXATTR)
 922                         policy = &getxattr_policy;
 923                 else
 924                         policy = &lookup_policy;
 925         }
 926
 927         generation = obd->u.cli.cl_import->imp_generation;
 928         if (!it || (it->it_op & (IT_OPEN | IT_CREAT)))
 929                 acl_bufsize = min_t(__u32, imp->imp_connect_data.ocd_max_easize,
 930                                     XATTR_SIZE_MAX);
 931         else
 932                 acl_bufsize = LUSTRE_POSIX_ACL_MAX_SIZE_OLD;
 933
 934 resend:
 935         flags = saved_flags;
 936         if (it == NULL) {
 937                 /* The only way right now is FLOCK. */
 938                 LASSERTF(einfo->ei_type == LDLM_FLOCK, "lock type %d\n",
 939                          einfo->ei_type);
 940                 res_id.name[3] = LDLM_FLOCK;
 941                 req = ldlm_enqueue_pack(exp, 0);
 942         } else if (it->it_op & IT_OPEN) {
 943                 req = mdc_intent_open_pack(exp, it, op_data, acl_bufsize);
 944         } else if (it->it_op & (IT_GETATTR | IT_LOOKUP)) {
 945                 req = mdc_intent_getattr_pack(exp, it, op_data, acl_bufsize);
 946         } else if (it->it_op & IT_READDIR) {
 947                 req = mdc_enqueue_pack(exp, 0);
 948         } else if (it->it_op & IT_LAYOUT) {
 949                 if (!imp_connect_lvb_type(imp))
 950                         RETURN(-EOPNOTSUPP);
 951                 req = mdc_intent_layout_pack(exp, it, op_data);
 952                 lvb_type = LVB_T_LAYOUT;
 953         } else if (it->it_op & IT_GETXATTR) {
 954                 req = mdc_intent_getxattr_pack(exp, it, op_data);
 955         } else {
 956                 LBUG();
 957                 RETURN(-EINVAL);
 958         }
 959
 960         if (IS_ERR(req))
 961                 RETURN(PTR_ERR(req));
 962
 963         if (resends) {
 964                 req->rq_generation_set = 1;
 965                 req->rq_import_generation = generation;
 966                 req->rq_sent = ktime_get_real_seconds() + resends;
 967         }
 968
 969         einfo->ei_enq_slot = !mdc_skip_mod_rpc_slot(it);
 970
 971         /* With Data-on-MDT the glimpse callback is needed too.
 972          * It is set here in advance but not in mdc_finish_enqueue()
 973          * to avoid possible races. It is safe to have glimpse handler
 974          * for non-DOM locks and costs nothing.
 975          */
 976         if (einfo->ei_cb_gl == NULL)
 977                 einfo->ei_cb_gl = mdc_ldlm_glimpse_ast;
 978
 979         rc = ldlm_cli_enqueue(exp, &req, einfo, &res_id, policy, &flags, NULL,
 980                               0, lvb_type, lockh, 0);
 981
 982         if (!it) {
 983                 /* For flock requests we immediatelly return without further
 984                  * delay and let caller deal with the rest, since rest of
 985                  * this function metadata processing makes no sense for flock
 986                  * requests anyway. But in case of problem during comms with
 987                  * server (-ETIMEDOUT) or any signal/kill attempt (-EINTR),
 988                  * we cannot rely on caller and this mainly for F_UNLCKs
 989                  * (explicits or automatically generated by kernel to clean
 990                  * current flocks upon exit) that can't be trashed.
 991                  */
 992                 ptlrpc_req_finished(req);
 993                 if (((rc == -EINTR) || (rc == -ETIMEDOUT)) &&
 994                     (einfo->ei_type == LDLM_FLOCK) &&
 995                     (einfo->ei_mode == LCK_NL))
 996                         goto resend;
 997                 RETURN(rc);
 998         }
 999
1000         if (rc < 0) {
1001                 CDEBUG(D_INFO,
1002                       "%s: ldlm_cli_enqueue "DFID":"DFID"=%s failed: rc = %d\n",
1003                       obd->obd_name, PFID(&op_data->op_fid1),
1004                       PFID(&op_data->op_fid2), op_data->op_name ?: "", rc);
1005
1006                 mdc_clear_replay_flag(req, rc);
1007                 ptlrpc_req_finished(req);
1008                 RETURN(rc);
1009         }
1010
1011         lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1012         LASSERT(lockrep != NULL);
1013
1014         lockrep->lock_policy_res2 =
1015                 ptlrpc_status_ntoh(lockrep->lock_policy_res2);
1016
1017         /* Retry infinitely when the server returns -EINPROGRESS for the
1018          * intent operation, when server returns -EINPROGRESS for acquiring
1019          * intent lock, we'll retry in after_reply().
1020          */
1021         if (it && (int)lockrep->lock_policy_res2 == -EINPROGRESS) {
1022                 mdc_clear_replay_flag(req, rc);
1023                 ptlrpc_req_finished(req);
1024                 if (generation == obd->u.cli.cl_import->imp_generation) {
1025                         if (signal_pending(current))
1026                                 RETURN(-EINTR);
1027
1028                         resends++;
1029                         CDEBUG(D_HA, "%s: resend:%d op:%d "DFID"/"DFID"\n",
1030                                obd->obd_name, resends, it->it_op,
1031                                PFID(&op_data->op_fid1),
1032                                PFID(&op_data->op_fid2));
1033                         goto resend;
1034                 } else {
1035                         CDEBUG(D_HA, "resend cross eviction\n");
1036                         RETURN(-EIO);
1037                 }
1038         }
1039
1040         if ((int)lockrep->lock_policy_res2 == -ERANGE &&
1041             it->it_op & (IT_OPEN | IT_GETATTR | IT_LOOKUP) &&
1042             acl_bufsize == LUSTRE_POSIX_ACL_MAX_SIZE_OLD) {
1043                 mdc_clear_replay_flag(req, -ERANGE);
1044                 ptlrpc_req_finished(req);
1045                 acl_bufsize = min_t(__u32, imp->imp_connect_data.ocd_max_easize,
1046                                     XATTR_SIZE_MAX);
1047                 goto resend;
1048         }
1049
1050         rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
1051         if (rc < 0) {
1052                 if (lustre_handle_is_used(lockh)) {
1053                         ldlm_lock_decref(lockh, einfo->ei_mode);
1054                         memset(lockh, 0, sizeof(*lockh));
1055                 }
1056                 ptlrpc_req_finished(req);
1057
1058                 it->it_lock_handle = 0;
1059                 it->it_lock_mode = 0;
1060                 it->it_request = NULL;
1061         }
1062
1063         RETURN(rc);
1064 }
1065
1066 int mdc_enqueue(struct obd_export *exp, struct ldlm_enqueue_info *einfo,
1067                 const union ldlm_policy_data *policy,
1068                 struct md_op_data *op_data,
1069                 struct lustre_handle *lockh, __u64 extra_lock_flags)
1070 {
1071         return mdc_enqueue_base(exp, einfo, policy, NULL,
1072                                 op_data, lockh, extra_lock_flags);
1073 }
1074
1075 static int mdc_finish_intent_lock(struct obd_export *exp,
1076                                   struct ptlrpc_request *request,
1077                                   struct md_op_data *op_data,
1078                                   struct lookup_intent *it,
1079                                   struct lustre_handle *lockh)
1080 {
1081         struct lustre_handle old_lock;
1082         struct ldlm_lock *lock;
1083         int rc = 0;
1084
1085         ENTRY;
1086         LASSERT(request != NULL);
1087         LASSERT(request != LP_POISON);
1088         LASSERT(request->rq_repmsg != LP_POISON);
1089
1090         if (it->it_op & IT_READDIR)
1091                 RETURN(0);
1092
1093         if (it->it_op & (IT_GETXATTR | IT_LAYOUT)) {
1094                 if (it->it_status != 0)
1095                         GOTO(out, rc = it->it_status);
1096         } else {
1097                 if (!it_disposition(it, DISP_IT_EXECD)) {
1098                         /* The server failed before it even started executing
1099                          * the intent, i.e. because it couldn't unpack the
1100                          * request.
1101                          */
1102                         LASSERT(it->it_status != 0);
1103                         GOTO(out, rc = it->it_status);
1104                 }
1105                 rc = it_open_error(DISP_IT_EXECD, it);
1106                 if (rc)
1107                         GOTO(out, rc);
1108
1109                 rc = it_open_error(DISP_LOOKUP_EXECD, it);
1110                 if (rc)
1111                         GOTO(out, rc);
1112
1113                 /* keep requests around for the multiple phases of the call
1114                  * this shows the DISP_XX must guarantee we make it into the
1115                  * call
1116                  */
1117                 if (!it_disposition(it, DISP_ENQ_CREATE_REF) &&
1118                     it_disposition(it, DISP_OPEN_CREATE) &&
1119                     !it_open_error(DISP_OPEN_CREATE, it)) {
1120                         it_set_disposition(it, DISP_ENQ_CREATE_REF);
1121                         /* balanced in ll_create_node */
1122                         ptlrpc_request_addref(request);
1123                 }
1124                 if (!it_disposition(it, DISP_ENQ_OPEN_REF) &&
1125                     it_disposition(it, DISP_OPEN_OPEN) &&
1126                     !it_open_error(DISP_OPEN_OPEN, it)) {
1127                         it_set_disposition(it, DISP_ENQ_OPEN_REF);
1128                         /* balanced in ll_file_open */
1129                         ptlrpc_request_addref(request);
1130                         /* eviction in middle of open RPC processing b=11546 */
1131                         OBD_FAIL_TIMEOUT(OBD_FAIL_MDC_ENQUEUE_PAUSE,
1132                                          obd_timeout);
1133                 }
1134
1135                 if (it->it_op & IT_CREAT) {
1136                         /* XXX this belongs in ll_create_it */
1137                 } else if (it->it_op == IT_OPEN) {
1138                         LASSERT(!it_disposition(it, DISP_OPEN_CREATE));
1139                 } else {
1140                         LASSERT(it->it_op & (IT_GETATTR | IT_LOOKUP));
1141                 }
1142         }
1143
1144         /* If we already have a matching lock, then cancel the new
1145          * one.  We have to set the data here instead of in
1146          * mdc_enqueue, because we need to use the child's inode as
1147          * the l_ast_data to match, and that's not available until
1148          * intent_finish has performed the iget().
1149          */
1150         lock = ldlm_handle2lock(lockh);
1151         if (lock) {
1152                 union ldlm_policy_data policy = lock->l_policy_data;
1153
1154                 LDLM_DEBUG(lock, "matching against this");
1155
1156                 if (it_has_reply_body(it)) {
1157                         struct mdt_body *body;
1158
1159                         body = req_capsule_server_get(&request->rq_pill,
1160                                                       &RMF_MDT_BODY);
1161                         /* mdc_enqueue checked */
1162                         LASSERT(body != NULL);
1163                         LASSERTF(fid_res_name_eq(&body->mbo_fid1,
1164                                                  &lock->l_resource->lr_name),
1165                                  "Lock res_id: "DLDLMRES", fid: "DFID"\n",
1166                                  PLDLMRES(lock->l_resource),
1167                                  PFID(&body->mbo_fid1));
1168                 }
1169                 LDLM_LOCK_PUT(lock);
1170
1171                 memcpy(&old_lock, lockh, sizeof(*lockh));
1172                 if (ldlm_lock_match(NULL, LDLM_FL_BLOCK_GRANTED, NULL,
1173                                    LDLM_IBITS, &policy, LCK_NL, &old_lock, 0)) {
1174                         ldlm_lock_decref_and_cancel(lockh, it->it_lock_mode);
1175                         memcpy(lockh, &old_lock, sizeof(old_lock));
1176                         it->it_lock_handle = lockh->cookie;
1177                 }
1178         }
1179
1180         EXIT;
1181 out:
1182         CDEBUG(D_DENTRY,
1183                "D_IT dentry=%.*s intent=%s status=%d disp=%x: rc = %d\n",
1184                 (int)op_data->op_namelen, op_data->op_name,
1185                 ldlm_it2str(it->it_op), it->it_status, it->it_disposition, rc);
1186
1187         return rc;
1188 }
1189
1190 int mdc_revalidate_lock(struct obd_export *exp, struct lookup_intent *it,
1191                         struct lu_fid *fid, __u64 *bits)
1192 {
1193         /* We could just return 1 immediately, but as we should only be called
1194          * in revalidate_it if we already have a lock, let's verify that.
1195          */
1196         struct ldlm_res_id res_id;
1197         struct lustre_handle lockh;
1198         union ldlm_policy_data policy;
1199         enum ldlm_mode mode;
1200
1201         ENTRY;
1202         if (it->it_lock_handle) {
1203                 lockh.cookie = it->it_lock_handle;
1204                 mode = ldlm_revalidate_lock_handle(&lockh, bits);
1205         } else {
1206                 fid_build_reg_res_name(fid, &res_id);
1207                 switch (it->it_op) {
1208                 case IT_GETATTR:
1209                         /* File attributes are held under multiple bits:
1210                          * nlink is under lookup lock, size and times are
1211                          * under UPDATE lock and recently we've also got
1212                          * a separate permissions lock for owner/group/acl that
1213                          * were protected by lookup lock before.
1214                          * Getattr must provide all of that information,
1215                          * so we need to ensure we have all of those locks.
1216                          * Unfortunately, if the bits are split across multiple
1217                          * locks, there's no easy way to match all of them here,
1218                          * so an extra RPC would be performed to fetch all
1219                          * of those bits at once for now.
1220                          */
1221                         /* For new MDTs(> 2.4), UPDATE|PERM should be enough,
1222                          * but for old MDTs (< 2.4), permission is covered
1223                          * by LOOKUP lock, so it needs to match all bits here.
1224                          */
1225                         policy.l_inodebits.bits = MDS_INODELOCK_UPDATE |
1226                                                   MDS_INODELOCK_LOOKUP |
1227                                                   MDS_INODELOCK_PERM;
1228                         break;
1229                 case IT_READDIR:
1230                         policy.l_inodebits.bits = MDS_INODELOCK_UPDATE;
1231                         break;
1232                 case IT_LAYOUT:
1233                         policy.l_inodebits.bits = MDS_INODELOCK_LAYOUT;
1234                         break;
1235                 default:
1236                         policy.l_inodebits.bits = MDS_INODELOCK_LOOKUP;
1237                         break;
1238                 }
1239
1240                 mode = mdc_lock_match(exp, LDLM_FL_BLOCK_GRANTED, fid,
1241                                       LDLM_IBITS, &policy,
1242                                       LCK_CR | LCK_CW | LCK_PR | LCK_PW,
1243                                       &lockh);
1244         }
1245
1246         if (mode) {
1247                 it->it_lock_handle = lockh.cookie;
1248                 it->it_lock_mode = mode;
1249         } else {
1250                 it->it_lock_handle = 0;
1251                 it->it_lock_mode = 0;
1252         }
1253
1254         RETURN(!!mode);
1255 }
1256
1257 /*
1258  * This long block is all about fixing up the lock and request state
1259  * so that it is correct as of the moment _before_ the operation was
1260  * applied; that way, the VFS will think that everything is normal and
1261  * call Lustre's regular VFS methods.
1262  *
1263  * If we're performing a creation, that means that unless the creation
1264  * failed with EEXIST, we should fake up a negative dentry.
1265  *
1266  * For everything else, we want to lookup to succeed.
1267  *
1268  * One additional note: if CREATE or OPEN succeeded, we add an extra
1269  * reference to the request because we need to keep it around until
1270  * ll_create/ll_open gets called.
1271  *
1272  * The server will return to us, in it_disposition, an indication of
1273  * exactly what it_status refers to.
1274  *
1275  * If DISP_OPEN_OPEN is set, then it_status refers to the open() call,
1276  * otherwise if DISP_OPEN_CREATE is set, then it status is the
1277  * creation failure mode.  In either case, one of DISP_LOOKUP_NEG or
1278  * DISP_LOOKUP_POS will be set, indicating whether the child lookup
1279  * was successful.
1280  *
1281  * Else, if DISP_LOOKUP_EXECD then it_status is the rc of the
1282  * child lookup.
1283  */
1284 int mdc_intent_lock(struct obd_export *exp, struct md_op_data *op_data,
1285                     struct lookup_intent *it, struct ptlrpc_request **reqp,
1286                     ldlm_blocking_callback cb_blocking, __u64 extra_lock_flags)
1287 {
1288         struct ldlm_enqueue_info einfo = {
1289                 .ei_type        = LDLM_IBITS,
1290                 .ei_mode        = it_to_lock_mode(it),
1291                 .ei_cb_bl       = cb_blocking,
1292                 .ei_cb_cp       = ldlm_completion_ast,
1293                 .ei_cb_gl       = mdc_ldlm_glimpse_ast,
1294         };
1295         struct lustre_handle lockh;
1296         int rc = 0;
1297
1298         ENTRY;
1299         LASSERT(it);
1300         CDEBUG(D_DLMTRACE, "(name: %.*s,"DFID") in obj "DFID
1301                 ", intent: %s flags %#llo\n", (int)op_data->op_namelen,
1302                 op_data->op_name, PFID(&op_data->op_fid2),
1303                 PFID(&op_data->op_fid1), ldlm_it2str(it->it_op),
1304                 it->it_flags);
1305
1306         lockh.cookie = 0;
1307         if (fid_is_sane(&op_data->op_fid2) &&
1308             (it->it_op & (IT_LOOKUP | IT_GETATTR | IT_READDIR))) {
1309                 /* We could just return 1 immediately, but since we should only
1310                  * be called in revalidate_it if we already have a lock, let's
1311                  * verify that.
1312                  */
1313                 it->it_lock_handle = 0;
1314                 rc = mdc_revalidate_lock(exp, it, &op_data->op_fid2, NULL);
1315                 /* Only return failure if it was not GETATTR by cfid
1316                  * (from inode_revalidate()).
1317                  */
1318                 if (rc || op_data->op_namelen != 0)
1319                         RETURN(rc);
1320         }
1321
1322         /* For case if upper layer did not alloc fid, do it now. */
1323         if (!fid_is_sane(&op_data->op_fid2) && it->it_op & IT_CREAT) {
1324                 rc = mdc_fid_alloc(NULL, exp, &op_data->op_fid2, op_data);
1325                 if (rc < 0) {
1326                         CERROR("%s: cannot allocate new FID: rc=%d\n",
1327                                exp->exp_obd->obd_name, rc);
1328                         RETURN(rc);
1329                 }
1330         }
1331
1332         rc = mdc_enqueue_base(exp, &einfo, NULL, it, op_data, &lockh,
1333                               extra_lock_flags);
1334         if (rc < 0)
1335                 RETURN(rc);
1336
1337         *reqp = it->it_request;
1338         rc = mdc_finish_intent_lock(exp, *reqp, op_data, it, &lockh);
1339         RETURN(rc);
1340 }
1341
1342 static int mdc_intent_getattr_async_interpret(const struct lu_env *env,
1343                                               struct ptlrpc_request *req,
1344                                               void *args, int rc)
1345 {
1346         struct mdc_getattr_args *ga = args;
1347         struct obd_export *exp = ga->ga_exp;
1348         struct md_enqueue_info *minfo = ga->ga_minfo;
1349         struct ldlm_enqueue_info *einfo = &minfo->mi_einfo;
1350         struct lookup_intent *it = &minfo->mi_it;
1351         struct lustre_handle *lockh = &minfo->mi_lockh;
1352         struct ldlm_reply *lockrep;
1353         __u64 flags = LDLM_FL_HAS_INTENT;
1354
1355         ENTRY;
1356         if (OBD_FAIL_CHECK(OBD_FAIL_MDC_GETATTR_ENQUEUE))
1357                 rc = -ETIMEDOUT;
1358
1359         rc = ldlm_cli_enqueue_fini(exp, req, einfo->ei_type, 1, einfo->ei_mode,
1360                                    &flags, NULL, 0, lockh, rc);
1361         if (rc < 0) {
1362                 CERROR("%s: ldlm_cli_enqueue_fini() failed: rc = %d\n",
1363                        exp->exp_obd->obd_name, rc);
1364                 mdc_clear_replay_flag(req, rc);
1365                 GOTO(out, rc);
1366         }
1367
1368         lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1369         LASSERT(lockrep != NULL);
1370
1371         lockrep->lock_policy_res2 =
1372                 ptlrpc_status_ntoh(lockrep->lock_policy_res2);
1373
1374         rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
1375         if (rc)
1376                 GOTO(out, rc);
1377
1378         rc = mdc_finish_intent_lock(exp, req, &minfo->mi_data, it, lockh);
1379         EXIT;
1380
1381 out:
1382         minfo->mi_cb(req, minfo, rc);
1383         return 0;
1384 }
1385
1386 int mdc_intent_getattr_async(struct obd_export *exp,
1387                              struct md_enqueue_info *minfo)
1388 {
1389         struct md_op_data *op_data = &minfo->mi_data;
1390         struct lookup_intent *it = &minfo->mi_it;
1391         struct ptlrpc_request *req;
1392         struct mdc_getattr_args *ga;
1393         struct ldlm_res_id res_id;
1394         union ldlm_policy_data policy = {
1395                 .l_inodebits = { MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE }
1396         };
1397         __u64 flags = LDLM_FL_HAS_INTENT;
1398         int rc = 0;
1399
1400         ENTRY;
1401         CDEBUG(D_DLMTRACE,
1402                "name: %.*s in inode "DFID", intent: %s flags %#llo\n",
1403                (int)op_data->op_namelen, op_data->op_name,
1404                PFID(&op_data->op_fid1), ldlm_it2str(it->it_op), it->it_flags);
1405
1406         fid_build_reg_res_name(&op_data->op_fid1, &res_id);
1407         /* If the MDT return -ERANGE because of large ACL, then the sponsor
1408          * of the async getattr RPC will handle that by itself.
1409          */
1410         req = mdc_intent_getattr_pack(exp, it, op_data,
1411                                       LUSTRE_POSIX_ACL_MAX_SIZE_OLD);
1412         if (IS_ERR(req))
1413                 RETURN(PTR_ERR(req));
1414
1415         /* With Data-on-MDT the glimpse callback is needed too.
1416          * It is set here in advance but not in mdc_finish_enqueue()
1417          * to avoid possible races. It is safe to have glimpse handler
1418          * for non-DOM locks and costs nothing.
1419          */
1420         if (minfo->mi_einfo.ei_cb_gl == NULL)
1421                 minfo->mi_einfo.ei_cb_gl = mdc_ldlm_glimpse_ast;
1422
1423         rc = ldlm_cli_enqueue(exp, &req, &minfo->mi_einfo, &res_id, &policy,
1424                               &flags, NULL, 0, LVB_T_NONE, &minfo->mi_lockh, 1);
1425         if (rc < 0) {
1426                 ptlrpc_req_finished(req);
1427                 RETURN(rc);
1428         }
1429
1430         ga = ptlrpc_req_async_args(ga, req);
1431         ga->ga_exp = exp;
1432         ga->ga_minfo = minfo;
1433
1434         req->rq_interpret_reply = mdc_intent_getattr_async_interpret;
1435         ptlrpcd_add_req(req);
1436
1437         RETURN(0);
1438 }