drivers/staging/lustre/lustre/fld/fld_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.sun.com/software/products/lustre/docs/GPLv2.pdf
  19  *
  20  * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
  21  * CA 95054 USA or visit www.sun.com if you need additional information or
  22  * have any questions.
  23  *
  24  * GPL HEADER END
  25  */
  26 /*
  27  * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
  28  * Use is subject to license terms.
  29  *
  30  * Copyright (c) 2011, 2013, Intel Corporation.
  31  */
  32 /*
  33  * This file is part of Lustre, http://www.lustre.org/
  34  * Lustre is a trademark of Sun Microsystems, Inc.
  35  *
  36  * lustre/fld/fld_request.c
  37  *
  38  * FLD (Fids Location Database)
  39  *
  40  * Author: Yury Umanets <umka@clusterfs.com>
  41  */
  42
  43 #define DEBUG_SUBSYSTEM S_FLD
  44
  45 # include <linux/libcfs/libcfs.h>
  46 # include <linux/module.h>
  47 # include <asm/div64.h>
  48
  49 #include <obd.h>
  50 #include <obd_class.h>
  51 #include <lustre_ver.h>
  52 #include <obd_support.h>
  53 #include <lprocfs_status.h>
  54
  55 #include <dt_object.h>
  56 #include <md_object.h>
  57 #include <lustre_req_layout.h>
  58 #include <lustre_fld.h>
  59 #include <lustre_mdc.h>
  60 #include "fld_internal.h"
  61
  62 /* TODO: these 3 functions are copies of flow-control code from mdc_lib.c
  63  * It should be common thing. The same about mdc RPC lock */
  64 static int fld_req_avail(struct client_obd *cli, struct mdc_cache_waiter *mcw)
  65 {
  66         int rc;
  67
  68         client_obd_list_lock(&cli->cl_loi_list_lock);
  69         rc = list_empty(&mcw->mcw_entry);
  70         client_obd_list_unlock(&cli->cl_loi_list_lock);
  71         return rc;
  72 };
  73
  74 static void fld_enter_request(struct client_obd *cli)
  75 {
  76         struct mdc_cache_waiter mcw;
  77         struct l_wait_info lwi = { 0 };
  78
  79         client_obd_list_lock(&cli->cl_loi_list_lock);
  80         if (cli->cl_r_in_flight >= cli->cl_max_rpcs_in_flight) {
  81                 list_add_tail(&mcw.mcw_entry, &cli->cl_cache_waiters);
  82                 init_waitqueue_head(&mcw.mcw_waitq);
  83                 client_obd_list_unlock(&cli->cl_loi_list_lock);
  84                 l_wait_event(mcw.mcw_waitq, fld_req_avail(cli, &mcw), &lwi);
  85         } else {
  86                 cli->cl_r_in_flight++;
  87                 client_obd_list_unlock(&cli->cl_loi_list_lock);
  88         }
  89 }
  90
  91 static void fld_exit_request(struct client_obd *cli)
  92 {
  93         struct list_head *l, *tmp;
  94         struct mdc_cache_waiter *mcw;
  95
  96         client_obd_list_lock(&cli->cl_loi_list_lock);
  97         cli->cl_r_in_flight--;
  98         list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
  99
 100                 if (cli->cl_r_in_flight >= cli->cl_max_rpcs_in_flight) {
 101                         /* No free request slots anymore */
 102                         break;
 103                 }
 104
 105                 mcw = list_entry(l, struct mdc_cache_waiter, mcw_entry);
 106                 list_del_init(&mcw->mcw_entry);
 107                 cli->cl_r_in_flight++;
 108                 wake_up(&mcw->mcw_waitq);
 109         }
 110         client_obd_list_unlock(&cli->cl_loi_list_lock);
 111 }
 112
 113 static int fld_rrb_hash(struct lu_client_fld *fld,
 114                         seqno_t seq)
 115 {
 116         LASSERT(fld->lcf_count > 0);
 117         return do_div(seq, fld->lcf_count);
 118 }
 119
 120 static struct lu_fld_target *
 121 fld_rrb_scan(struct lu_client_fld *fld, seqno_t seq)
 122 {
 123         struct lu_fld_target *target;
 124         int hash;
 125
 126         /* Because almost all of special sequence located in MDT0,
 127          * it should go to index 0 directly, instead of calculating
 128          * hash again, and also if other MDTs is not being connected,
 129          * the fld lookup requests(for seq on MDT0) should not be
 130          * blocked because of other MDTs */
 131         if (fid_seq_is_norm(seq))
 132                 hash = fld_rrb_hash(fld, seq);
 133         else
 134                 hash = 0;
 135
 136         list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
 137                 if (target->ft_idx == hash)
 138                         return target;
 139         }
 140
 141         CERROR("%s: Can't find target by hash %d (seq "LPX64"). "
 142                "Targets (%d):\n", fld->lcf_name, hash, seq,
 143                fld->lcf_count);
 144
 145         list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
 146                 const char *srv_name = target->ft_srv != NULL  ?
 147                         target->ft_srv->lsf_name : "<null>";
 148                 const char *exp_name = target->ft_exp != NULL ?
 149                         (char *)target->ft_exp->exp_obd->obd_uuid.uuid :
 150                         "<null>";
 151
 152                 CERROR("  exp: 0x%p (%s), srv: 0x%p (%s), idx: "LPU64"\n",
 153                        target->ft_exp, exp_name, target->ft_srv,
 154                        srv_name, target->ft_idx);
 155         }
 156
 157         /*
 158          * If target is not found, there is logical error anyway, so here is
 159          * LBUG() to catch this situation.
 160          */
 161         LBUG();
 162         return NULL;
 163 }
 164
 165 struct lu_fld_hash fld_hash[] = {
 166         {
 167                 .fh_name = "RRB",
 168                 .fh_hash_func = fld_rrb_hash,
 169                 .fh_scan_func = fld_rrb_scan
 170         },
 171         {
 172                 0,
 173         }
 174 };
 175
 176 static struct lu_fld_target *
 177 fld_client_get_target(struct lu_client_fld *fld, seqno_t seq)
 178 {
 179         struct lu_fld_target *target;
 180
 181         LASSERT(fld->lcf_hash != NULL);
 182
 183         spin_lock(&fld->lcf_lock);
 184         target = fld->lcf_hash->fh_scan_func(fld, seq);
 185         spin_unlock(&fld->lcf_lock);
 186
 187         if (target != NULL) {
 188                 CDEBUG(D_INFO, "%s: Found target (idx "LPU64
 189                        ") by seq "LPX64"\n", fld->lcf_name,
 190                        target->ft_idx, seq);
 191         }
 192
 193         return target;
 194 }
 195
 196 /*
 197  * Add export to FLD. This is usually done by CMM and LMV as they are main users
 198  * of FLD module.
 199  */
 200 int fld_client_add_target(struct lu_client_fld *fld,
 201                           struct lu_fld_target *tar)
 202 {
 203         const char *name;
 204         struct lu_fld_target *target, *tmp;
 205
 206         LASSERT(tar != NULL);
 207         name = fld_target_name(tar);
 208         LASSERT(name != NULL);
 209         LASSERT(tar->ft_srv != NULL || tar->ft_exp != NULL);
 210
 211         if (fld->lcf_flags != LUSTRE_FLD_INIT) {
 212                 CERROR("%s: Attempt to add target %s (idx "LPU64") "
 213                        "on fly - skip it\n", fld->lcf_name, name,
 214                        tar->ft_idx);
 215                 return 0;
 216         } else {
 217                 CDEBUG(D_INFO, "%s: Adding target %s (idx "
 218                        LPU64")\n", fld->lcf_name, name, tar->ft_idx);
 219         }
 220
 221         OBD_ALLOC_PTR(target);
 222         if (target == NULL)
 223                 return -ENOMEM;
 224
 225         spin_lock(&fld->lcf_lock);
 226         list_for_each_entry(tmp, &fld->lcf_targets, ft_chain) {
 227                 if (tmp->ft_idx == tar->ft_idx) {
 228                         spin_unlock(&fld->lcf_lock);
 229                         OBD_FREE_PTR(target);
 230                         CERROR("Target %s exists in FLD and known as %s:#"LPU64"\n",
 231                                name, fld_target_name(tmp), tmp->ft_idx);
 232                         return -EEXIST;
 233                 }
 234         }
 235
 236         target->ft_exp = tar->ft_exp;
 237         if (target->ft_exp != NULL)
 238                 class_export_get(target->ft_exp);
 239         target->ft_srv = tar->ft_srv;
 240         target->ft_idx = tar->ft_idx;
 241
 242         list_add_tail(&target->ft_chain,
 243                           &fld->lcf_targets);
 244
 245         fld->lcf_count++;
 246         spin_unlock(&fld->lcf_lock);
 247
 248         return 0;
 249 }
 250 EXPORT_SYMBOL(fld_client_add_target);
 251
 252 /* Remove export from FLD */
 253 int fld_client_del_target(struct lu_client_fld *fld, __u64 idx)
 254 {
 255         struct lu_fld_target *target, *tmp;
 256
 257         spin_lock(&fld->lcf_lock);
 258         list_for_each_entry_safe(target, tmp,
 259                                      &fld->lcf_targets, ft_chain) {
 260                 if (target->ft_idx == idx) {
 261                         fld->lcf_count--;
 262                         list_del(&target->ft_chain);
 263                         spin_unlock(&fld->lcf_lock);
 264
 265                         if (target->ft_exp != NULL)
 266                                 class_export_put(target->ft_exp);
 267
 268                         OBD_FREE_PTR(target);
 269                         return 0;
 270                 }
 271         }
 272         spin_unlock(&fld->lcf_lock);
 273         return -ENOENT;
 274 }
 275 EXPORT_SYMBOL(fld_client_del_target);
 276
 277 #ifdef LPROCFS
 278 struct proc_dir_entry *fld_type_proc_dir = NULL;
 279
 280 static int fld_client_proc_init(struct lu_client_fld *fld)
 281 {
 282         int rc;
 283
 284         fld->lcf_proc_dir = lprocfs_register(fld->lcf_name,
 285                                              fld_type_proc_dir,
 286                                              NULL, NULL);
 287
 288         if (IS_ERR(fld->lcf_proc_dir)) {
 289                 CERROR("%s: LProcFS failed in fld-init\n",
 290                        fld->lcf_name);
 291                 rc = PTR_ERR(fld->lcf_proc_dir);
 292                 return rc;
 293         }
 294
 295         rc = lprocfs_add_vars(fld->lcf_proc_dir,
 296                               fld_client_proc_list, fld);
 297         if (rc) {
 298                 CERROR("%s: Can't init FLD proc, rc %d\n",
 299                        fld->lcf_name, rc);
 300                 GOTO(out_cleanup, rc);
 301         }
 302
 303         return 0;
 304
 305 out_cleanup:
 306         fld_client_proc_fini(fld);
 307         return rc;
 308 }
 309
 310 void fld_client_proc_fini(struct lu_client_fld *fld)
 311 {
 312         if (fld->lcf_proc_dir) {
 313                 if (!IS_ERR(fld->lcf_proc_dir))
 314                         lprocfs_remove(&fld->lcf_proc_dir);
 315                 fld->lcf_proc_dir = NULL;
 316         }
 317 }
 318 #else
 319 static int fld_client_proc_init(struct lu_client_fld *fld)
 320 {
 321         return 0;
 322 }
 323
 324 void fld_client_proc_fini(struct lu_client_fld *fld)
 325 {
 326         return;
 327 }
 328 #endif
 329
 330 EXPORT_SYMBOL(fld_client_proc_fini);
 331
 332 static inline int hash_is_sane(int hash)
 333 {
 334         return (hash >= 0 && hash < ARRAY_SIZE(fld_hash));
 335 }
 336
 337 int fld_client_init(struct lu_client_fld *fld,
 338                     const char *prefix, int hash)
 339 {
 340         int cache_size, cache_threshold;
 341         int rc;
 342
 343         LASSERT(fld != NULL);
 344
 345         snprintf(fld->lcf_name, sizeof(fld->lcf_name),
 346                  "cli-%s", prefix);
 347
 348         if (!hash_is_sane(hash)) {
 349                 CERROR("%s: Wrong hash function %#x\n",
 350                        fld->lcf_name, hash);
 351                 return -EINVAL;
 352         }
 353
 354         fld->lcf_count = 0;
 355         spin_lock_init(&fld->lcf_lock);
 356         fld->lcf_hash = &fld_hash[hash];
 357         fld->lcf_flags = LUSTRE_FLD_INIT;
 358         INIT_LIST_HEAD(&fld->lcf_targets);
 359
 360         cache_size = FLD_CLIENT_CACHE_SIZE /
 361                 sizeof(struct fld_cache_entry);
 362
 363         cache_threshold = cache_size *
 364                 FLD_CLIENT_CACHE_THRESHOLD / 100;
 365
 366         fld->lcf_cache = fld_cache_init(fld->lcf_name,
 367                                         cache_size, cache_threshold);
 368         if (IS_ERR(fld->lcf_cache)) {
 369                 rc = PTR_ERR(fld->lcf_cache);
 370                 fld->lcf_cache = NULL;
 371                 GOTO(out, rc);
 372         }
 373
 374         rc = fld_client_proc_init(fld);
 375         if (rc)
 376                 GOTO(out, rc);
 377 out:
 378         if (rc)
 379                 fld_client_fini(fld);
 380         else
 381                 CDEBUG(D_INFO, "%s: Using \"%s\" hash\n",
 382                        fld->lcf_name, fld->lcf_hash->fh_name);
 383         return rc;
 384 }
 385 EXPORT_SYMBOL(fld_client_init);
 386
 387 void fld_client_fini(struct lu_client_fld *fld)
 388 {
 389         struct lu_fld_target *target, *tmp;
 390
 391         spin_lock(&fld->lcf_lock);
 392         list_for_each_entry_safe(target, tmp,
 393                                      &fld->lcf_targets, ft_chain) {
 394                 fld->lcf_count--;
 395                 list_del(&target->ft_chain);
 396                 if (target->ft_exp != NULL)
 397                         class_export_put(target->ft_exp);
 398                 OBD_FREE_PTR(target);
 399         }
 400         spin_unlock(&fld->lcf_lock);
 401
 402         if (fld->lcf_cache != NULL) {
 403                 if (!IS_ERR(fld->lcf_cache))
 404                         fld_cache_fini(fld->lcf_cache);
 405                 fld->lcf_cache = NULL;
 406         }
 407 }
 408 EXPORT_SYMBOL(fld_client_fini);
 409
 410 int fld_client_rpc(struct obd_export *exp,
 411                    struct lu_seq_range *range, __u32 fld_op)
 412 {
 413         struct ptlrpc_request *req;
 414         struct lu_seq_range   *prange;
 415         __u32            *op;
 416         int                 rc;
 417         struct obd_import     *imp;
 418
 419         LASSERT(exp != NULL);
 420
 421         imp = class_exp2cliimp(exp);
 422         req = ptlrpc_request_alloc_pack(imp, &RQF_FLD_QUERY, LUSTRE_MDS_VERSION,
 423                                         FLD_QUERY);
 424         if (req == NULL)
 425                 return -ENOMEM;
 426
 427         op = req_capsule_client_get(&req->rq_pill, &RMF_FLD_OPC);
 428         *op = fld_op;
 429
 430         prange = req_capsule_client_get(&req->rq_pill, &RMF_FLD_MDFLD);
 431         *prange = *range;
 432
 433         ptlrpc_request_set_replen(req);
 434         req->rq_request_portal = FLD_REQUEST_PORTAL;
 435         ptlrpc_at_set_req_timeout(req);
 436
 437         if (fld_op == FLD_LOOKUP &&
 438             imp->imp_connect_flags_orig & OBD_CONNECT_MDS_MDS)
 439                 req->rq_allow_replay = 1;
 440
 441         if (fld_op != FLD_LOOKUP)
 442                 mdc_get_rpc_lock(exp->exp_obd->u.cli.cl_rpc_lock, NULL);
 443         fld_enter_request(&exp->exp_obd->u.cli);
 444         rc = ptlrpc_queue_wait(req);
 445         fld_exit_request(&exp->exp_obd->u.cli);
 446         if (fld_op != FLD_LOOKUP)
 447                 mdc_put_rpc_lock(exp->exp_obd->u.cli.cl_rpc_lock, NULL);
 448         if (rc)
 449                 GOTO(out_req, rc);
 450
 451         prange = req_capsule_server_get(&req->rq_pill, &RMF_FLD_MDFLD);
 452         if (prange == NULL)
 453                 GOTO(out_req, rc = -EFAULT);
 454         *range = *prange;
 455 out_req:
 456         ptlrpc_req_finished(req);
 457         return rc;
 458 }
 459
 460 int fld_client_lookup(struct lu_client_fld *fld, seqno_t seq, mdsno_t *mds,
 461                       __u32 flags, const struct lu_env *env)
 462 {
 463         struct lu_seq_range res = { 0 };
 464         struct lu_fld_target *target;
 465         int rc;
 466
 467         fld->lcf_flags |= LUSTRE_FLD_RUN;
 468
 469         rc = fld_cache_lookup(fld->lcf_cache, seq, &res);
 470         if (rc == 0) {
 471                 *mds = res.lsr_index;
 472                 return 0;
 473         }
 474
 475         /* Can not find it in the cache */
 476         target = fld_client_get_target(fld, seq);
 477         LASSERT(target != NULL);
 478
 479         CDEBUG(D_INFO, "%s: Lookup fld entry (seq: "LPX64") on "
 480                "target %s (idx "LPU64")\n", fld->lcf_name, seq,
 481                fld_target_name(target), target->ft_idx);
 482
 483         res.lsr_start = seq;
 484         fld_range_set_type(&res, flags);
 485         rc = fld_client_rpc(target->ft_exp, &res, FLD_LOOKUP);
 486
 487         if (rc == 0) {
 488                 *mds = res.lsr_index;
 489
 490                 fld_cache_insert(fld->lcf_cache, &res);
 491         }
 492         return rc;
 493 }
 494 EXPORT_SYMBOL(fld_client_lookup);
 495
 496 void fld_client_flush(struct lu_client_fld *fld)
 497 {
 498         fld_cache_flush(fld->lcf_cache);
 499 }
 500 EXPORT_SYMBOL(fld_client_flush);
 501
 502 static int __init fld_mod_init(void)
 503 {
 504         fld_type_proc_dir = lprocfs_register(LUSTRE_FLD_NAME,
 505                                              proc_lustre_root,
 506                                              NULL, NULL);
 507         if (IS_ERR(fld_type_proc_dir))
 508                 return PTR_ERR(fld_type_proc_dir);
 509
 510         return 0;
 511 }
 512
 513 static void __exit fld_mod_exit(void)
 514 {
 515         if (fld_type_proc_dir != NULL && !IS_ERR(fld_type_proc_dir)) {
 516                 lprocfs_remove(&fld_type_proc_dir);
 517                 fld_type_proc_dir = NULL;
 518         }
 519 }
 520
 521 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
 522 MODULE_DESCRIPTION("Lustre FLD");
 523 MODULE_LICENSE("GPL");
 524
 525 module_init(fld_mod_init)
 526 module_exit(fld_mod_exit)