fs/ceph/mds_client.c

   1 #include <linux/ceph/ceph_debug.h>
   2
   3 #include <linux/fs.h>
   4 #include <linux/wait.h>
   5 #include <linux/slab.h>
   6 #include <linux/gfp.h>
   7 #include <linux/sched.h>
   8 #include <linux/debugfs.h>
   9 #include <linux/seq_file.h>
  10 #include <linux/utsname.h>
  11 #include <linux/ratelimit.h>
  12
  13 #include "super.h"
  14 #include "mds_client.h"
  15
  16 #include <linux/ceph/ceph_features.h>
  17 #include <linux/ceph/messenger.h>
  18 #include <linux/ceph/decode.h>
  19 #include <linux/ceph/pagelist.h>
  20 #include <linux/ceph/auth.h>
  21 #include <linux/ceph/debugfs.h>
  22
  23 /*
  24  * A cluster of MDS (metadata server) daemons is responsible for
  25  * managing the file system namespace (the directory hierarchy and
  26  * inodes) and for coordinating shared access to storage.  Metadata is
  27  * partitioning hierarchically across a number of servers, and that
  28  * partition varies over time as the cluster adjusts the distribution
  29  * in order to balance load.
  30  *
  31  * The MDS client is primarily responsible to managing synchronous
  32  * metadata requests for operations like open, unlink, and so forth.
  33  * If there is a MDS failure, we find out about it when we (possibly
  34  * request and) receive a new MDS map, and can resubmit affected
  35  * requests.
  36  *
  37  * For the most part, though, we take advantage of a lossless
  38  * communications channel to the MDS, and do not need to worry about
  39  * timing out or resubmitting requests.
  40  *
  41  * We maintain a stateful "session" with each MDS we interact with.
  42  * Within each session, we sent periodic heartbeat messages to ensure
  43  * any capabilities or leases we have been issues remain valid.  If
  44  * the session times out and goes stale, our leases and capabilities
  45  * are no longer valid.
  46  */
  47
  48 struct ceph_reconnect_state {
  49         int nr_caps;
  50         struct ceph_pagelist *pagelist;
  51         bool flock;
  52 };
  53
  54 static void __wake_requests(struct ceph_mds_client *mdsc,
  55                             struct list_head *head);
  56
  57 static const struct ceph_connection_operations mds_con_ops;
  58
  59
  60 /*
  61  * mds reply parsing
  62  */
  63
  64 /*
  65  * parse individual inode info
  66  */
  67 static int parse_reply_info_in(void **p, void *end,
  68                                struct ceph_mds_reply_info_in *info,
  69                                u64 features)
  70 {
  71         int err = -EIO;
  72
  73         info->in = *p;
  74         *p += sizeof(struct ceph_mds_reply_inode) +
  75                 sizeof(*info->in->fragtree.splits) *
  76                 le32_to_cpu(info->in->fragtree.nsplits);
  77
  78         ceph_decode_32_safe(p, end, info->symlink_len, bad);
  79         ceph_decode_need(p, end, info->symlink_len, bad);
  80         info->symlink = *p;
  81         *p += info->symlink_len;
  82
  83         if (features & CEPH_FEATURE_DIRLAYOUTHASH)
  84                 ceph_decode_copy_safe(p, end, &info->dir_layout,
  85                                       sizeof(info->dir_layout), bad);
  86         else
  87                 memset(&info->dir_layout, 0, sizeof(info->dir_layout));
  88
  89         ceph_decode_32_safe(p, end, info->xattr_len, bad);
  90         ceph_decode_need(p, end, info->xattr_len, bad);
  91         info->xattr_data = *p;
  92         *p += info->xattr_len;
  93
  94         if (features & CEPH_FEATURE_MDS_INLINE_DATA) {
  95                 ceph_decode_64_safe(p, end, info->inline_version, bad);
  96                 ceph_decode_32_safe(p, end, info->inline_len, bad);
  97                 ceph_decode_need(p, end, info->inline_len, bad);
  98                 info->inline_data = *p;
  99                 *p += info->inline_len;
 100         } else
 101                 info->inline_version = CEPH_INLINE_NONE;
 102
 103         if (features & CEPH_FEATURE_FS_FILE_LAYOUT_V2) {
 104                 ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
 105                 ceph_decode_need(p, end, info->pool_ns_len, bad);
 106                 *p += info->pool_ns_len;
 107         } else {
 108                 info->pool_ns_len = 0;
 109         }
 110
 111         return 0;
 112 bad:
 113         return err;
 114 }
 115
 116 /*
 117  * parse a normal reply, which may contain a (dir+)dentry and/or a
 118  * target inode.
 119  */
 120 static int parse_reply_info_trace(void **p, void *end,
 121                                   struct ceph_mds_reply_info_parsed *info,
 122                                   u64 features)
 123 {
 124         int err;
 125
 126         if (info->head->is_dentry) {
 127                 err = parse_reply_info_in(p, end, &info->diri, features);
 128                 if (err < 0)
 129                         goto out_bad;
 130
 131                 if (unlikely(*p + sizeof(*info->dirfrag) > end))
 132                         goto bad;
 133                 info->dirfrag = *p;
 134                 *p += sizeof(*info->dirfrag) +
 135                         sizeof(u32)*le32_to_cpu(info->dirfrag->ndist);
 136                 if (unlikely(*p > end))
 137                         goto bad;
 138
 139                 ceph_decode_32_safe(p, end, info->dname_len, bad);
 140                 ceph_decode_need(p, end, info->dname_len, bad);
 141                 info->dname = *p;
 142                 *p += info->dname_len;
 143                 info->dlease = *p;
 144                 *p += sizeof(*info->dlease);
 145         }
 146
 147         if (info->head->is_target) {
 148                 err = parse_reply_info_in(p, end, &info->targeti, features);
 149                 if (err < 0)
 150                         goto out_bad;
 151         }
 152
 153         if (unlikely(*p != end))
 154                 goto bad;
 155         return 0;
 156
 157 bad:
 158         err = -EIO;
 159 out_bad:
 160         pr_err("problem parsing mds trace %d\n", err);
 161         return err;
 162 }
 163
 164 /*
 165  * parse readdir results
 166  */
 167 static int parse_reply_info_dir(void **p, void *end,
 168                                 struct ceph_mds_reply_info_parsed *info,
 169                                 u64 features)
 170 {
 171         u32 num, i = 0;
 172         int err;
 173
 174         info->dir_dir = *p;
 175         if (*p + sizeof(*info->dir_dir) > end)
 176                 goto bad;
 177         *p += sizeof(*info->dir_dir) +
 178                 sizeof(u32)*le32_to_cpu(info->dir_dir->ndist);
 179         if (*p > end)
 180                 goto bad;
 181
 182         ceph_decode_need(p, end, sizeof(num) + 2, bad);
 183         num = ceph_decode_32(p);
 184         info->dir_end = ceph_decode_8(p);
 185         info->dir_complete = ceph_decode_8(p);
 186         if (num == 0)
 187                 goto done;
 188
 189         BUG_ON(!info->dir_in);
 190         info->dir_dname = (void *)(info->dir_in + num);
 191         info->dir_dname_len = (void *)(info->dir_dname + num);
 192         info->dir_dlease = (void *)(info->dir_dname_len + num);
 193         if ((unsigned long)(info->dir_dlease + num) >
 194             (unsigned long)info->dir_in + info->dir_buf_size) {
 195                 pr_err("dir contents are larger than expected\n");
 196                 WARN_ON(1);
 197                 goto bad;
 198         }
 199
 200         info->dir_nr = num;
 201         while (num) {
 202                 /* dentry */
 203                 ceph_decode_need(p, end, sizeof(u32)*2, bad);
 204                 info->dir_dname_len[i] = ceph_decode_32(p);
 205                 ceph_decode_need(p, end, info->dir_dname_len[i], bad);
 206                 info->dir_dname[i] = *p;
 207                 *p += info->dir_dname_len[i];
 208                 dout("parsed dir dname '%.*s'\n", info->dir_dname_len[i],
 209                      info->dir_dname[i]);
 210                 info->dir_dlease[i] = *p;
 211                 *p += sizeof(struct ceph_mds_reply_lease);
 212
 213                 /* inode */
 214                 err = parse_reply_info_in(p, end, &info->dir_in[i], features);
 215                 if (err < 0)
 216                         goto out_bad;
 217                 i++;
 218                 num--;
 219         }
 220
 221 done:
 222         if (*p != end)
 223                 goto bad;
 224         return 0;
 225
 226 bad:
 227         err = -EIO;
 228 out_bad:
 229         pr_err("problem parsing dir contents %d\n", err);
 230         return err;
 231 }
 232
 233 /*
 234  * parse fcntl F_GETLK results
 235  */
 236 static int parse_reply_info_filelock(void **p, void *end,
 237                                      struct ceph_mds_reply_info_parsed *info,
 238                                      u64 features)
 239 {
 240         if (*p + sizeof(*info->filelock_reply) > end)
 241                 goto bad;
 242
 243         info->filelock_reply = *p;
 244         *p += sizeof(*info->filelock_reply);
 245
 246         if (unlikely(*p != end))
 247                 goto bad;
 248         return 0;
 249
 250 bad:
 251         return -EIO;
 252 }
 253
 254 /*
 255  * parse create results
 256  */
 257 static int parse_reply_info_create(void **p, void *end,
 258                                   struct ceph_mds_reply_info_parsed *info,
 259                                   u64 features)
 260 {
 261         if (features & CEPH_FEATURE_REPLY_CREATE_INODE) {
 262                 if (*p == end) {
 263                         info->has_create_ino = false;
 264                 } else {
 265                         info->has_create_ino = true;
 266                         info->ino = ceph_decode_64(p);
 267                 }
 268         }
 269
 270         if (unlikely(*p != end))
 271                 goto bad;
 272         return 0;
 273
 274 bad:
 275         return -EIO;
 276 }
 277
 278 /*
 279  * parse extra results
 280  */
 281 static int parse_reply_info_extra(void **p, void *end,
 282                                   struct ceph_mds_reply_info_parsed *info,
 283                                   u64 features)
 284 {
 285         if (info->head->op == CEPH_MDS_OP_GETFILELOCK)
 286                 return parse_reply_info_filelock(p, end, info, features);
 287         else if (info->head->op == CEPH_MDS_OP_READDIR ||
 288                  info->head->op == CEPH_MDS_OP_LSSNAP)
 289                 return parse_reply_info_dir(p, end, info, features);
 290         else if (info->head->op == CEPH_MDS_OP_CREATE)
 291                 return parse_reply_info_create(p, end, info, features);
 292         else
 293                 return -EIO;
 294 }
 295
 296 /*
 297  * parse entire mds reply
 298  */
 299 static int parse_reply_info(struct ceph_msg *msg,
 300                             struct ceph_mds_reply_info_parsed *info,
 301                             u64 features)
 302 {
 303         void *p, *end;
 304         u32 len;
 305         int err;
 306
 307         info->head = msg->front.iov_base;
 308         p = msg->front.iov_base + sizeof(struct ceph_mds_reply_head);
 309         end = p + msg->front.iov_len - sizeof(struct ceph_mds_reply_head);
 310
 311         /* trace */
 312         ceph_decode_32_safe(&p, end, len, bad);
 313         if (len > 0) {
 314                 ceph_decode_need(&p, end, len, bad);
 315                 err = parse_reply_info_trace(&p, p+len, info, features);
 316                 if (err < 0)
 317                         goto out_bad;
 318         }
 319
 320         /* extra */
 321         ceph_decode_32_safe(&p, end, len, bad);
 322         if (len > 0) {
 323                 ceph_decode_need(&p, end, len, bad);
 324                 err = parse_reply_info_extra(&p, p+len, info, features);
 325                 if (err < 0)
 326                         goto out_bad;
 327         }
 328
 329         /* snap blob */
 330         ceph_decode_32_safe(&p, end, len, bad);
 331         info->snapblob_len = len;
 332         info->snapblob = p;
 333         p += len;
 334
 335         if (p != end)
 336                 goto bad;
 337         return 0;
 338
 339 bad:
 340         err = -EIO;
 341 out_bad:
 342         pr_err("mds parse_reply err %d\n", err);
 343         return err;
 344 }
 345
 346 static void destroy_reply_info(struct ceph_mds_reply_info_parsed *info)
 347 {
 348         if (!info->dir_in)
 349                 return;
 350         free_pages((unsigned long)info->dir_in, get_order(info->dir_buf_size));
 351 }
 352
 353
 354 /*
 355  * sessions
 356  */
 357 const char *ceph_session_state_name(int s)
 358 {
 359         switch (s) {
 360         case CEPH_MDS_SESSION_NEW: return "new";
 361         case CEPH_MDS_SESSION_OPENING: return "opening";
 362         case CEPH_MDS_SESSION_OPEN: return "open";
 363         case CEPH_MDS_SESSION_HUNG: return "hung";
 364         case CEPH_MDS_SESSION_CLOSING: return "closing";
 365         case CEPH_MDS_SESSION_RESTARTING: return "restarting";
 366         case CEPH_MDS_SESSION_RECONNECTING: return "reconnecting";
 367         default: return "???";
 368         }
 369 }
 370
 371 static struct ceph_mds_session *get_session(struct ceph_mds_session *s)
 372 {
 373         if (atomic_inc_not_zero(&s->s_ref)) {
 374                 dout("mdsc get_session %p %d -> %d\n", s,
 375                      atomic_read(&s->s_ref)-1, atomic_read(&s->s_ref));
 376                 return s;
 377         } else {
 378                 dout("mdsc get_session %p 0 -- FAIL", s);
 379                 return NULL;
 380         }
 381 }
 382
 383 void ceph_put_mds_session(struct ceph_mds_session *s)
 384 {
 385         dout("mdsc put_session %p %d -> %d\n", s,
 386              atomic_read(&s->s_ref), atomic_read(&s->s_ref)-1);
 387         if (atomic_dec_and_test(&s->s_ref)) {
 388                 if (s->s_auth.authorizer)
 389                         ceph_auth_destroy_authorizer(s->s_auth.authorizer);
 390                 kfree(s);
 391         }
 392 }
 393
 394 /*
 395  * called under mdsc->mutex
 396  */
 397 struct ceph_mds_session *__ceph_lookup_mds_session(struct ceph_mds_client *mdsc,
 398                                                    int mds)
 399 {
 400         struct ceph_mds_session *session;
 401
 402         if (mds >= mdsc->max_sessions || mdsc->sessions[mds] == NULL)
 403                 return NULL;
 404         session = mdsc->sessions[mds];
 405         dout("lookup_mds_session %p %d\n", session,
 406              atomic_read(&session->s_ref));
 407         get_session(session);
 408         return session;
 409 }
 410
 411 static bool __have_session(struct ceph_mds_client *mdsc, int mds)
 412 {
 413         if (mds >= mdsc->max_sessions)
 414                 return false;
 415         return mdsc->sessions[mds];
 416 }
 417
 418 static int __verify_registered_session(struct ceph_mds_client *mdsc,
 419                                        struct ceph_mds_session *s)
 420 {
 421         if (s->s_mds >= mdsc->max_sessions ||
 422             mdsc->sessions[s->s_mds] != s)
 423                 return -ENOENT;
 424         return 0;
 425 }
 426
 427 /*
 428  * create+register a new session for given mds.
 429  * called under mdsc->mutex.
 430  */
 431 static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
 432                                                  int mds)
 433 {
 434         struct ceph_mds_session *s;
 435
 436         if (mds >= mdsc->mdsmap->m_max_mds)
 437                 return ERR_PTR(-EINVAL);
 438
 439         s = kzalloc(sizeof(*s), GFP_NOFS);
 440         if (!s)
 441                 return ERR_PTR(-ENOMEM);
 442         s->s_mdsc = mdsc;
 443         s->s_mds = mds;
 444         s->s_state = CEPH_MDS_SESSION_NEW;
 445         s->s_ttl = 0;
 446         s->s_seq = 0;
 447         mutex_init(&s->s_mutex);
 448
 449         ceph_con_init(&s->s_con, s, &mds_con_ops, &mdsc->fsc->client->msgr);
 450
 451         spin_lock_init(&s->s_gen_ttl_lock);
 452         s->s_cap_gen = 0;
 453         s->s_cap_ttl = jiffies - 1;
 454
 455         spin_lock_init(&s->s_cap_lock);
 456         s->s_renew_requested = 0;
 457         s->s_renew_seq = 0;
 458         INIT_LIST_HEAD(&s->s_caps);
 459         s->s_nr_caps = 0;
 460         s->s_trim_caps = 0;
 461         atomic_set(&s->s_ref, 1);
 462         INIT_LIST_HEAD(&s->s_waiting);
 463         INIT_LIST_HEAD(&s->s_unsafe);
 464         s->s_num_cap_releases = 0;
 465         s->s_cap_reconnect = 0;
 466         s->s_cap_iterator = NULL;
 467         INIT_LIST_HEAD(&s->s_cap_releases);
 468         INIT_LIST_HEAD(&s->s_cap_flushing);
 469         INIT_LIST_HEAD(&s->s_cap_snaps_flushing);
 470
 471         dout("register_session mds%d\n", mds);
 472         if (mds >= mdsc->max_sessions) {
 473                 int newmax = 1 << get_count_order(mds+1);
 474                 struct ceph_mds_session **sa;
 475
 476                 dout("register_session realloc to %d\n", newmax);
 477                 sa = kcalloc(newmax, sizeof(void *), GFP_NOFS);
 478                 if (sa == NULL)
 479                         goto fail_realloc;
 480                 if (mdsc->sessions) {
 481                         memcpy(sa, mdsc->sessions,
 482                                mdsc->max_sessions * sizeof(void *));
 483                         kfree(mdsc->sessions);
 484                 }
 485                 mdsc->sessions = sa;
 486                 mdsc->max_sessions = newmax;
 487         }
 488         mdsc->sessions[mds] = s;
 489         atomic_inc(&mdsc->num_sessions);
 490         atomic_inc(&s->s_ref);  /* one ref to sessions[], one to caller */
 491
 492         ceph_con_open(&s->s_con, CEPH_ENTITY_TYPE_MDS, mds,
 493                       ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
 494
 495         return s;
 496
 497 fail_realloc:
 498         kfree(s);
 499         return ERR_PTR(-ENOMEM);
 500 }
 501
 502 /*
 503  * called under mdsc->mutex
 504  */
 505 static void __unregister_session(struct ceph_mds_client *mdsc,
 506                                struct ceph_mds_session *s)
 507 {
 508         dout("__unregister_session mds%d %p\n", s->s_mds, s);
 509         BUG_ON(mdsc->sessions[s->s_mds] != s);
 510         mdsc->sessions[s->s_mds] = NULL;
 511         ceph_con_close(&s->s_con);
 512         ceph_put_mds_session(s);
 513         atomic_dec(&mdsc->num_sessions);
 514 }
 515
 516 /*
 517  * drop session refs in request.
 518  *
 519  * should be last request ref, or hold mdsc->mutex
 520  */
 521 static void put_request_session(struct ceph_mds_request *req)
 522 {
 523         if (req->r_session) {
 524                 ceph_put_mds_session(req->r_session);
 525                 req->r_session = NULL;
 526         }
 527 }
 528
 529 void ceph_mdsc_release_request(struct kref *kref)
 530 {
 531         struct ceph_mds_request *req = container_of(kref,
 532                                                     struct ceph_mds_request,
 533                                                     r_kref);
 534         destroy_reply_info(&req->r_reply_info);
 535         if (req->r_request)
 536                 ceph_msg_put(req->r_request);
 537         if (req->r_reply)
 538                 ceph_msg_put(req->r_reply);
 539         if (req->r_inode) {
 540                 ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
 541                 iput(req->r_inode);
 542         }
 543         if (req->r_locked_dir)
 544                 ceph_put_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN);
 545         iput(req->r_target_inode);
 546         if (req->r_dentry)
 547                 dput(req->r_dentry);
 548         if (req->r_old_dentry)
 549                 dput(req->r_old_dentry);
 550         if (req->r_old_dentry_dir) {
 551                 /*
 552                  * track (and drop pins for) r_old_dentry_dir
 553                  * separately, since r_old_dentry's d_parent may have
 554                  * changed between the dir mutex being dropped and
 555                  * this request being freed.
 556                  */
 557                 ceph_put_cap_refs(ceph_inode(req->r_old_dentry_dir),
 558                                   CEPH_CAP_PIN);
 559                 iput(req->r_old_dentry_dir);
 560         }
 561         kfree(req->r_path1);
 562         kfree(req->r_path2);
 563         if (req->r_pagelist)
 564                 ceph_pagelist_release(req->r_pagelist);
 565         put_request_session(req);
 566         ceph_unreserve_caps(req->r_mdsc, &req->r_caps_reservation);
 567         kfree(req);
 568 }
 569
 570 DEFINE_RB_FUNCS(request, struct ceph_mds_request, r_tid, r_node)
 571
 572 /*
 573  * lookup session, bump ref if found.
 574  *
 575  * called under mdsc->mutex.
 576  */
 577 static struct ceph_mds_request *
 578 lookup_get_request(struct ceph_mds_client *mdsc, u64 tid)
 579 {
 580         struct ceph_mds_request *req;
 581
 582         req = lookup_request(&mdsc->request_tree, tid);
 583         if (req)
 584                 ceph_mdsc_get_request(req);
 585
 586         return req;
 587 }
 588
 589 /*
 590  * Register an in-flight request, and assign a tid.  Link to directory
 591  * are modifying (if any).
 592  *
 593  * Called under mdsc->mutex.
 594  */
 595 static void __register_request(struct ceph_mds_client *mdsc,
 596                                struct ceph_mds_request *req,
 597                                struct inode *dir)
 598 {
 599         req->r_tid = ++mdsc->last_tid;
 600         if (req->r_num_caps)
 601                 ceph_reserve_caps(mdsc, &req->r_caps_reservation,
 602                                   req->r_num_caps);
 603         dout("__register_request %p tid %lld\n", req, req->r_tid);
 604         ceph_mdsc_get_request(req);
 605         insert_request(&mdsc->request_tree, req);
 606
 607         req->r_uid = current_fsuid();
 608         req->r_gid = current_fsgid();
 609
 610         if (mdsc->oldest_tid == 0 && req->r_op != CEPH_MDS_OP_SETFILELOCK)
 611                 mdsc->oldest_tid = req->r_tid;
 612
 613         if (dir) {
 614                 ihold(dir);
 615                 req->r_unsafe_dir = dir;
 616         }
 617 }
 618
 619 static void __unregister_request(struct ceph_mds_client *mdsc,
 620                                  struct ceph_mds_request *req)
 621 {
 622         dout("__unregister_request %p tid %lld\n", req, req->r_tid);
 623
 624         if (req->r_tid == mdsc->oldest_tid) {
 625                 struct rb_node *p = rb_next(&req->r_node);
 626                 mdsc->oldest_tid = 0;
 627                 while (p) {
 628                         struct ceph_mds_request *next_req =
 629                                 rb_entry(p, struct ceph_mds_request, r_node);
 630                         if (next_req->r_op != CEPH_MDS_OP_SETFILELOCK) {
 631                                 mdsc->oldest_tid = next_req->r_tid;
 632                                 break;
 633                         }
 634                         p = rb_next(p);
 635                 }
 636         }
 637
 638         erase_request(&mdsc->request_tree, req);
 639
 640         if (req->r_unsafe_dir && req->r_got_unsafe) {
 641                 struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir);
 642                 spin_lock(&ci->i_unsafe_lock);
 643                 list_del_init(&req->r_unsafe_dir_item);
 644                 spin_unlock(&ci->i_unsafe_lock);
 645         }
 646         if (req->r_target_inode && req->r_got_unsafe) {
 647                 struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
 648                 spin_lock(&ci->i_unsafe_lock);
 649                 list_del_init(&req->r_unsafe_target_item);
 650                 spin_unlock(&ci->i_unsafe_lock);
 651         }
 652
 653         if (req->r_unsafe_dir) {
 654                 iput(req->r_unsafe_dir);
 655                 req->r_unsafe_dir = NULL;
 656         }
 657
 658         complete_all(&req->r_safe_completion);
 659
 660         ceph_mdsc_put_request(req);
 661 }
 662
 663 /*
 664  * Choose mds to send request to next.  If there is a hint set in the
 665  * request (e.g., due to a prior forward hint from the mds), use that.
 666  * Otherwise, consult frag tree and/or caps to identify the
 667  * appropriate mds.  If all else fails, choose randomly.
 668  *
 669  * Called under mdsc->mutex.
 670  */
 671 static struct dentry *get_nonsnap_parent(struct dentry *dentry)
 672 {
 673         /*
 674          * we don't need to worry about protecting the d_parent access
 675          * here because we never renaming inside the snapped namespace
 676          * except to resplice to another snapdir, and either the old or new
 677          * result is a valid result.
 678          */
 679         while (!IS_ROOT(dentry) && ceph_snap(d_inode(dentry)) != CEPH_NOSNAP)
 680                 dentry = dentry->d_parent;
 681         return dentry;
 682 }
 683
 684 static int __choose_mds(struct ceph_mds_client *mdsc,
 685                         struct ceph_mds_request *req)
 686 {
 687         struct inode *inode;
 688         struct ceph_inode_info *ci;
 689         struct ceph_cap *cap;
 690         int mode = req->r_direct_mode;
 691         int mds = -1;
 692         u32 hash = req->r_direct_hash;
 693         bool is_hash = req->r_direct_is_hash;
 694
 695         /*
 696          * is there a specific mds we should try?  ignore hint if we have
 697          * no session and the mds is not up (active or recovering).
 698          */
 699         if (req->r_resend_mds >= 0 &&
 700             (__have_session(mdsc, req->r_resend_mds) ||
 701              ceph_mdsmap_get_state(mdsc->mdsmap, req->r_resend_mds) > 0)) {
 702                 dout("choose_mds using resend_mds mds%d\n",
 703                      req->r_resend_mds);
 704                 return req->r_resend_mds;
 705         }
 706
 707         if (mode == USE_RANDOM_MDS)
 708                 goto random;
 709
 710         inode = NULL;
 711         if (req->r_inode) {
 712                 inode = req->r_inode;
 713         } else if (req->r_dentry) {
 714                 /* ignore race with rename; old or new d_parent is okay */
 715                 struct dentry *parent = req->r_dentry->d_parent;
 716                 struct inode *dir = d_inode(parent);
 717
 718                 if (dir->i_sb != mdsc->fsc->sb) {
 719                         /* not this fs! */
 720                         inode = d_inode(req->r_dentry);
 721                 } else if (ceph_snap(dir) != CEPH_NOSNAP) {
 722                         /* direct snapped/virtual snapdir requests
 723                          * based on parent dir inode */
 724                         struct dentry *dn = get_nonsnap_parent(parent);
 725                         inode = d_inode(dn);
 726                         dout("__choose_mds using nonsnap parent %p\n", inode);
 727                 } else {
 728                         /* dentry target */
 729                         inode = d_inode(req->r_dentry);
 730                         if (!inode || mode == USE_AUTH_MDS) {
 731                                 /* dir + name */
 732                                 inode = dir;
 733                                 hash = ceph_dentry_hash(dir, req->r_dentry);
 734                                 is_hash = true;
 735                         }
 736                 }
 737         }
 738
 739         dout("__choose_mds %p is_hash=%d (%d) mode %d\n", inode, (int)is_hash,
 740              (int)hash, mode);
 741         if (!inode)
 742                 goto random;
 743         ci = ceph_inode(inode);
 744
 745         if (is_hash && S_ISDIR(inode->i_mode)) {
 746                 struct ceph_inode_frag frag;
 747                 int found;
 748
 749                 ceph_choose_frag(ci, hash, &frag, &found);
 750                 if (found) {
 751                         if (mode == USE_ANY_MDS && frag.ndist > 0) {
 752                                 u8 r;
 753
 754                                 /* choose a random replica */
 755                                 get_random_bytes(&r, 1);
 756                                 r %= frag.ndist;
 757                                 mds = frag.dist[r];
 758                                 dout("choose_mds %p %llx.%llx "
 759                                      "frag %u mds%d (%d/%d)\n",
 760                                      inode, ceph_vinop(inode),
 761                                      frag.frag, mds,
 762                                      (int)r, frag.ndist);
 763                                 if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
 764                                     CEPH_MDS_STATE_ACTIVE)
 765                                         return mds;
 766                         }
 767
 768                         /* since this file/dir wasn't known to be
 769                          * replicated, then we want to look for the
 770                          * authoritative mds. */
 771                         mode = USE_AUTH_MDS;
 772                         if (frag.mds >= 0) {
 773                                 /* choose auth mds */
 774                                 mds = frag.mds;
 775                                 dout("choose_mds %p %llx.%llx "
 776                                      "frag %u mds%d (auth)\n",
 777                                      inode, ceph_vinop(inode), frag.frag, mds);
 778                                 if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
 779                                     CEPH_MDS_STATE_ACTIVE)
 780                                         return mds;
 781                         }
 782                 }
 783         }
 784
 785         spin_lock(&ci->i_ceph_lock);
 786         cap = NULL;
 787         if (mode == USE_AUTH_MDS)
 788                 cap = ci->i_auth_cap;
 789         if (!cap && !RB_EMPTY_ROOT(&ci->i_caps))
 790                 cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
 791         if (!cap) {
 792                 spin_unlock(&ci->i_ceph_lock);
 793                 goto random;
 794         }
 795         mds = cap->session->s_mds;
 796         dout("choose_mds %p %llx.%llx mds%d (%scap %p)\n",
 797              inode, ceph_vinop(inode), mds,
 798              cap == ci->i_auth_cap ? "auth " : "", cap);
 799         spin_unlock(&ci->i_ceph_lock);
 800         return mds;
 801
 802 random:
 803         mds = ceph_mdsmap_get_random_mds(mdsc->mdsmap);
 804         dout("choose_mds chose random mds%d\n", mds);
 805         return mds;
 806 }
 807
 808
 809 /*
 810  * session messages
 811  */
 812 static struct ceph_msg *create_session_msg(u32 op, u64 seq)
 813 {
 814         struct ceph_msg *msg;
 815         struct ceph_mds_session_head *h;
 816
 817         msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS,
 818                            false);
 819         if (!msg) {
 820                 pr_err("create_session_msg ENOMEM creating msg\n");
 821                 return NULL;
 822         }
 823         h = msg->front.iov_base;
 824         h->op = cpu_to_le32(op);
 825         h->seq = cpu_to_le64(seq);
 826
 827         return msg;
 828 }
 829
 830 /*
 831  * session message, specialization for CEPH_SESSION_REQUEST_OPEN
 832  * to include additional client metadata fields.
 833  */
 834 static struct ceph_msg *create_session_open_msg(struct ceph_mds_client *mdsc, u64 seq)
 835 {
 836         struct ceph_msg *msg;
 837         struct ceph_mds_session_head *h;
 838         int i = -1;
 839         int metadata_bytes = 0;
 840         int metadata_key_count = 0;
 841         struct ceph_options *opt = mdsc->fsc->client->options;
 842         void *p;
 843
 844         const char* metadata[][2] = {
 845                 {"hostname", utsname()->nodename},
 846                 {"kernel_version", utsname()->release},
 847                 {"entity_id", opt->name ? opt->name : ""},
 848                 {NULL, NULL}
 849         };
 850
 851         /* Calculate serialized length of metadata */
 852         metadata_bytes = 4;  /* map length */
 853         for (i = 0; metadata[i][0] != NULL; ++i) {
 854                 metadata_bytes += 8 + strlen(metadata[i][0]) +
 855                         strlen(metadata[i][1]);
 856                 metadata_key_count++;
 857         }
 858
 859         /* Allocate the message */
 860         msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h) + metadata_bytes,
 861                            GFP_NOFS, false);
 862         if (!msg) {
 863                 pr_err("create_session_msg ENOMEM creating msg\n");
 864                 return NULL;
 865         }
 866         h = msg->front.iov_base;
 867         h->op = cpu_to_le32(CEPH_SESSION_REQUEST_OPEN);
 868         h->seq = cpu_to_le64(seq);
 869
 870         /*
 871          * Serialize client metadata into waiting buffer space, using
 872          * the format that userspace expects for map<string, string>
 873          *
 874          * ClientSession messages with metadata are v2
 875          */
 876         msg->hdr.version = cpu_to_le16(2);
 877         msg->hdr.compat_version = cpu_to_le16(1);
 878
 879         /* The write pointer, following the session_head structure */
 880         p = msg->front.iov_base + sizeof(*h);
 881
 882         /* Number of entries in the map */
 883         ceph_encode_32(&p, metadata_key_count);
 884
 885         /* Two length-prefixed strings for each entry in the map */
 886         for (i = 0; metadata[i][0] != NULL; ++i) {
 887                 size_t const key_len = strlen(metadata[i][0]);
 888                 size_t const val_len = strlen(metadata[i][1]);
 889
 890                 ceph_encode_32(&p, key_len);
 891                 memcpy(p, metadata[i][0], key_len);
 892                 p += key_len;
 893                 ceph_encode_32(&p, val_len);
 894                 memcpy(p, metadata[i][1], val_len);
 895                 p += val_len;
 896         }
 897
 898         return msg;
 899 }
 900
 901 /*
 902  * send session open request.
 903  *
 904  * called under mdsc->mutex
 905  */
 906 static int __open_session(struct ceph_mds_client *mdsc,
 907                           struct ceph_mds_session *session)
 908 {
 909         struct ceph_msg *msg;
 910         int mstate;
 911         int mds = session->s_mds;
 912
 913         /* wait for mds to go active? */
 914         mstate = ceph_mdsmap_get_state(mdsc->mdsmap, mds);
 915         dout("open_session to mds%d (%s)\n", mds,
 916              ceph_mds_state_name(mstate));
 917         session->s_state = CEPH_MDS_SESSION_OPENING;
 918         session->s_renew_requested = jiffies;
 919
 920         /* send connect message */
 921         msg = create_session_open_msg(mdsc, session->s_seq);
 922         if (!msg)
 923                 return -ENOMEM;
 924         ceph_con_send(&session->s_con, msg);
 925         return 0;
 926 }
 927
 928 /*
 929  * open sessions for any export targets for the given mds
 930  *
 931  * called under mdsc->mutex
 932  */
 933 static struct ceph_mds_session *
 934 __open_export_target_session(struct ceph_mds_client *mdsc, int target)
 935 {
 936         struct ceph_mds_session *session;
 937
 938         session = __ceph_lookup_mds_session(mdsc, target);
 939         if (!session) {
 940                 session = register_session(mdsc, target);
 941                 if (IS_ERR(session))
 942                         return session;
 943         }
 944         if (session->s_state == CEPH_MDS_SESSION_NEW ||
 945             session->s_state == CEPH_MDS_SESSION_CLOSING)
 946                 __open_session(mdsc, session);
 947
 948         return session;
 949 }
 950
 951 struct ceph_mds_session *
 952 ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target)
 953 {
 954         struct ceph_mds_session *session;
 955
 956         dout("open_export_target_session to mds%d\n", target);
 957
 958         mutex_lock(&mdsc->mutex);
 959         session = __open_export_target_session(mdsc, target);
 960         mutex_unlock(&mdsc->mutex);
 961
 962         return session;
 963 }
 964
 965 static void __open_export_target_sessions(struct ceph_mds_client *mdsc,
 966                                           struct ceph_mds_session *session)
 967 {
 968         struct ceph_mds_info *mi;
 969         struct ceph_mds_session *ts;
 970         int i, mds = session->s_mds;
 971
 972         if (mds >= mdsc->mdsmap->m_max_mds)
 973                 return;
 974
 975         mi = &mdsc->mdsmap->m_info[mds];
 976         dout("open_export_target_sessions for mds%d (%d targets)\n",
 977              session->s_mds, mi->num_export_targets);
 978
 979         for (i = 0; i < mi->num_export_targets; i++) {
 980                 ts = __open_export_target_session(mdsc, mi->export_targets[i]);
 981                 if (!IS_ERR(ts))
 982                         ceph_put_mds_session(ts);
 983         }
 984 }
 985
 986 void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
 987                                            struct ceph_mds_session *session)
 988 {
 989         mutex_lock(&mdsc->mutex);
 990         __open_export_target_sessions(mdsc, session);
 991         mutex_unlock(&mdsc->mutex);
 992 }
 993
 994 /*
 995  * session caps
 996  */
 997
 998 /* caller holds s_cap_lock, we drop it */
 999 static void cleanup_cap_releases(struct ceph_mds_client *mdsc,
1000                                  struct ceph_mds_session *session)
1001         __releases(session->s_cap_lock)
1002 {
1003         LIST_HEAD(tmp_list);
1004         list_splice_init(&session->s_cap_releases, &tmp_list);
1005         session->s_num_cap_releases = 0;
1006         spin_unlock(&session->s_cap_lock);
1007
1008         dout("cleanup_cap_releases mds%d\n", session->s_mds);
1009         while (!list_empty(&tmp_list)) {
1010                 struct ceph_cap *cap;
1011                 /* zero out the in-progress message */
1012                 cap = list_first_entry(&tmp_list,
1013                                         struct ceph_cap, session_caps);
1014                 list_del(&cap->session_caps);
1015                 ceph_put_cap(mdsc, cap);
1016         }
1017 }
1018
1019 static void cleanup_session_requests(struct ceph_mds_client *mdsc,
1020                                      struct ceph_mds_session *session)
1021 {
1022         struct ceph_mds_request *req;
1023         struct rb_node *p;
1024
1025         dout("cleanup_session_requests mds%d\n", session->s_mds);
1026         mutex_lock(&mdsc->mutex);
1027         while (!list_empty(&session->s_unsafe)) {
1028                 req = list_first_entry(&session->s_unsafe,
1029                                        struct ceph_mds_request, r_unsafe_item);
1030                 list_del_init(&req->r_unsafe_item);
1031                 pr_warn_ratelimited(" dropping unsafe request %llu\n",
1032                                     req->r_tid);
1033                 __unregister_request(mdsc, req);
1034         }
1035         /* zero r_attempts, so kick_requests() will re-send requests */
1036         p = rb_first(&mdsc->request_tree);
1037         while (p) {
1038                 req = rb_entry(p, struct ceph_mds_request, r_node);
1039                 p = rb_next(p);
1040                 if (req->r_session &&
1041                     req->r_session->s_mds == session->s_mds)
1042                         req->r_attempts = 0;
1043         }
1044         mutex_unlock(&mdsc->mutex);
1045 }
1046
1047 /*
1048  * Helper to safely iterate over all caps associated with a session, with
1049  * special care taken to handle a racing __ceph_remove_cap().
1050  *
1051  * Caller must hold session s_mutex.
1052  */
1053 static int iterate_session_caps(struct ceph_mds_session *session,
1054                                  int (*cb)(struct inode *, struct ceph_cap *,
1055                                             void *), void *arg)
1056 {
1057         struct list_head *p;
1058         struct ceph_cap *cap;
1059         struct inode *inode, *last_inode = NULL;
1060         struct ceph_cap *old_cap = NULL;
1061         int ret;
1062
1063         dout("iterate_session_caps %p mds%d\n", session, session->s_mds);
1064         spin_lock(&session->s_cap_lock);
1065         p = session->s_caps.next;
1066         while (p != &session->s_caps) {
1067                 cap = list_entry(p, struct ceph_cap, session_caps);
1068                 inode = igrab(&cap->ci->vfs_inode);
1069                 if (!inode) {
1070                         p = p->next;
1071                         continue;
1072                 }
1073                 session->s_cap_iterator = cap;
1074                 spin_unlock(&session->s_cap_lock);
1075
1076                 if (last_inode) {
1077                         iput(last_inode);
1078                         last_inode = NULL;
1079                 }
1080                 if (old_cap) {
1081                         ceph_put_cap(session->s_mdsc, old_cap);
1082                         old_cap = NULL;
1083                 }
1084
1085                 ret = cb(inode, cap, arg);
1086                 last_inode = inode;
1087
1088                 spin_lock(&session->s_cap_lock);
1089                 p = p->next;
1090                 if (cap->ci == NULL) {
1091                         dout("iterate_session_caps  finishing cap %p removal\n",
1092                              cap);
1093                         BUG_ON(cap->session != session);
1094                         cap->session = NULL;
1095                         list_del_init(&cap->session_caps);
1096                         session->s_nr_caps--;
1097                         if (cap->queue_release) {
1098                                 list_add_tail(&cap->session_caps,
1099                                               &session->s_cap_releases);
1100                                 session->s_num_cap_releases++;
1101                         } else {
1102                                 old_cap = cap;  /* put_cap it w/o locks held */
1103                         }
1104                 }
1105                 if (ret < 0)
1106                         goto out;
1107         }
1108         ret = 0;
1109 out:
1110         session->s_cap_iterator = NULL;
1111         spin_unlock(&session->s_cap_lock);
1112
1113         iput(last_inode);
1114         if (old_cap)
1115                 ceph_put_cap(session->s_mdsc, old_cap);
1116
1117         return ret;
1118 }
1119
1120 static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
1121                                   void *arg)
1122 {
1123         struct ceph_inode_info *ci = ceph_inode(inode);
1124         LIST_HEAD(to_remove);
1125         int drop = 0;
1126
1127         dout("removing cap %p, ci is %p, inode is %p\n",
1128              cap, ci, &ci->vfs_inode);
1129         spin_lock(&ci->i_ceph_lock);
1130         __ceph_remove_cap(cap, false);
1131         if (!ci->i_auth_cap) {
1132                 struct ceph_cap_flush *cf;
1133                 struct ceph_mds_client *mdsc =
1134                         ceph_sb_to_client(inode->i_sb)->mdsc;
1135
1136                 while (true) {
1137                         struct rb_node *n = rb_first(&ci->i_cap_flush_tree);
1138                         if (!n)
1139                                 break;
1140                         cf = rb_entry(n, struct ceph_cap_flush, i_node);
1141                         rb_erase(&cf->i_node, &ci->i_cap_flush_tree);
1142                         list_add(&cf->list, &to_remove);
1143                 }
1144
1145                 spin_lock(&mdsc->cap_dirty_lock);
1146
1147                 list_for_each_entry(cf, &to_remove, list)
1148                         rb_erase(&cf->g_node, &mdsc->cap_flush_tree);
1149
1150                 if (!list_empty(&ci->i_dirty_item)) {
1151                         pr_warn_ratelimited(
1152                                 " dropping dirty %s state for %p %lld\n",
1153                                 ceph_cap_string(ci->i_dirty_caps),
1154                                 inode, ceph_ino(inode));
1155                         ci->i_dirty_caps = 0;
1156                         list_del_init(&ci->i_dirty_item);
1157                         drop = 1;
1158                 }
1159                 if (!list_empty(&ci->i_flushing_item)) {
1160                         pr_warn_ratelimited(
1161                                 " dropping dirty+flushing %s state for %p %lld\n",
1162                                 ceph_cap_string(ci->i_flushing_caps),
1163                                 inode, ceph_ino(inode));
1164                         ci->i_flushing_caps = 0;
1165                         list_del_init(&ci->i_flushing_item);
1166                         mdsc->num_cap_flushing--;
1167                         drop = 1;
1168                 }
1169                 spin_unlock(&mdsc->cap_dirty_lock);
1170
1171                 if (!ci->i_dirty_caps && ci->i_prealloc_cap_flush) {
1172                         list_add(&ci->i_prealloc_cap_flush->list, &to_remove);
1173                         ci->i_prealloc_cap_flush = NULL;
1174                 }
1175         }
1176         spin_unlock(&ci->i_ceph_lock);
1177         while (!list_empty(&to_remove)) {
1178                 struct ceph_cap_flush *cf;
1179                 cf = list_first_entry(&to_remove,
1180                                       struct ceph_cap_flush, list);
1181                 list_del(&cf->list);
1182                 ceph_free_cap_flush(cf);
1183         }
1184         while (drop--)
1185                 iput(inode);
1186         return 0;
1187 }
1188
1189 /*
1190  * caller must hold session s_mutex
1191  */
1192 static void remove_session_caps(struct ceph_mds_session *session)
1193 {
1194         dout("remove_session_caps on %p\n", session);
1195         iterate_session_caps(session, remove_session_caps_cb, NULL);
1196
1197         spin_lock(&session->s_cap_lock);
1198         if (session->s_nr_caps > 0) {
1199                 struct super_block *sb = session->s_mdsc->fsc->sb;
1200                 struct inode *inode;
1201                 struct ceph_cap *cap, *prev = NULL;
1202                 struct ceph_vino vino;
1203                 /*
1204                  * iterate_session_caps() skips inodes that are being
1205                  * deleted, we need to wait until deletions are complete.
1206                  * __wait_on_freeing_inode() is designed for the job,
1207                  * but it is not exported, so use lookup inode function
1208                  * to access it.
1209                  */
1210                 while (!list_empty(&session->s_caps)) {
1211                         cap = list_entry(session->s_caps.next,
1212                                          struct ceph_cap, session_caps);
1213                         if (cap == prev)
1214                                 break;
1215                         prev = cap;
1216                         vino = cap->ci->i_vino;
1217                         spin_unlock(&session->s_cap_lock);
1218
1219                         inode = ceph_find_inode(sb, vino);
1220                         iput(inode);
1221
1222                         spin_lock(&session->s_cap_lock);
1223                 }
1224         }
1225
1226         // drop cap expires and unlock s_cap_lock
1227         cleanup_cap_releases(session->s_mdsc, session);
1228
1229         BUG_ON(session->s_nr_caps > 0);
1230         BUG_ON(!list_empty(&session->s_cap_flushing));
1231 }
1232
1233 /*
1234  * wake up any threads waiting on this session's caps.  if the cap is
1235  * old (didn't get renewed on the client reconnect), remove it now.
1236  *
1237  * caller must hold s_mutex.
1238  */
1239 static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap,
1240                               void *arg)
1241 {
1242         struct ceph_inode_info *ci = ceph_inode(inode);
1243
1244         wake_up_all(&ci->i_cap_wq);
1245         if (arg) {
1246                 spin_lock(&ci->i_ceph_lock);
1247                 ci->i_wanted_max_size = 0;
1248                 ci->i_requested_max_size = 0;
1249                 spin_unlock(&ci->i_ceph_lock);
1250         }
1251         return 0;
1252 }
1253
1254 static void wake_up_session_caps(struct ceph_mds_session *session,
1255                                  int reconnect)
1256 {
1257         dout("wake_up_session_caps %p mds%d\n", session, session->s_mds);
1258         iterate_session_caps(session, wake_up_session_cb,
1259                              (void *)(unsigned long)reconnect);
1260 }
1261
1262 /*
1263  * Send periodic message to MDS renewing all currently held caps.  The
1264  * ack will reset the expiration for all caps from this session.
1265  *
1266  * caller holds s_mutex
1267  */
1268 static int send_renew_caps(struct ceph_mds_client *mdsc,
1269                            struct ceph_mds_session *session)
1270 {
1271         struct ceph_msg *msg;
1272         int state;
1273
1274         if (time_after_eq(jiffies, session->s_cap_ttl) &&
1275             time_after_eq(session->s_cap_ttl, session->s_renew_requested))
1276                 pr_info("mds%d caps stale\n", session->s_mds);
1277         session->s_renew_requested = jiffies;
1278
1279         /* do not try to renew caps until a recovering mds has reconnected
1280          * with its clients. */
1281         state = ceph_mdsmap_get_state(mdsc->mdsmap, session->s_mds);
1282         if (state < CEPH_MDS_STATE_RECONNECT) {
1283                 dout("send_renew_caps ignoring mds%d (%s)\n",
1284                      session->s_mds, ceph_mds_state_name(state));
1285                 return 0;
1286         }
1287
1288         dout("send_renew_caps to mds%d (%s)\n", session->s_mds,
1289                 ceph_mds_state_name(state));
1290         msg = create_session_msg(CEPH_SESSION_REQUEST_RENEWCAPS,
1291                                  ++session->s_renew_seq);
1292         if (!msg)
1293                 return -ENOMEM;
1294         ceph_con_send(&session->s_con, msg);
1295         return 0;
1296 }
1297
1298 static int send_flushmsg_ack(struct ceph_mds_client *mdsc,
1299                              struct ceph_mds_session *session, u64 seq)
1300 {
1301         struct ceph_msg *msg;
1302
1303         dout("send_flushmsg_ack to mds%d (%s)s seq %lld\n",
1304              session->s_mds, ceph_session_state_name(session->s_state), seq);
1305         msg = create_session_msg(CEPH_SESSION_FLUSHMSG_ACK, seq);
1306         if (!msg)
1307                 return -ENOMEM;
1308         ceph_con_send(&session->s_con, msg);
1309         return 0;
1310 }
1311
1312
1313 /*
1314  * Note new cap ttl, and any transition from stale -> not stale (fresh?).
1315  *
1316  * Called under session->s_mutex
1317  */
1318 static void renewed_caps(struct ceph_mds_client *mdsc,
1319                          struct ceph_mds_session *session, int is_renew)
1320 {
1321         int was_stale;
1322         int wake = 0;
1323
1324         spin_lock(&session->s_cap_lock);
1325         was_stale = is_renew && time_after_eq(jiffies, session->s_cap_ttl);
1326
1327         session->s_cap_ttl = session->s_renew_requested +
1328                 mdsc->mdsmap->m_session_timeout*HZ;
1329
1330         if (was_stale) {
1331                 if (time_before(jiffies, session->s_cap_ttl)) {
1332                         pr_info("mds%d caps renewed\n", session->s_mds);
1333                         wake = 1;
1334                 } else {
1335                         pr_info("mds%d caps still stale\n", session->s_mds);
1336                 }
1337         }
1338         dout("renewed_caps mds%d ttl now %lu, was %s, now %s\n",
1339              session->s_mds, session->s_cap_ttl, was_stale ? "stale" : "fresh",
1340              time_before(jiffies, session->s_cap_ttl) ? "stale" : "fresh");
1341         spin_unlock(&session->s_cap_lock);
1342
1343         if (wake)
1344                 wake_up_session_caps(session, 0);
1345 }
1346
1347 /*
1348  * send a session close request
1349  */
1350 static int request_close_session(struct ceph_mds_client *mdsc,
1351                                  struct ceph_mds_session *session)
1352 {
1353         struct ceph_msg *msg;
1354
1355         dout("request_close_session mds%d state %s seq %lld\n",
1356              session->s_mds, ceph_session_state_name(session->s_state),
1357              session->s_seq);
1358         msg = create_session_msg(CEPH_SESSION_REQUEST_CLOSE, session->s_seq);
1359         if (!msg)
1360                 return -ENOMEM;
1361         ceph_con_send(&session->s_con, msg);
1362         return 0;
1363 }
1364
1365 /*
1366  * Called with s_mutex held.
1367  */
1368 static int __close_session(struct ceph_mds_client *mdsc,
1369                          struct ceph_mds_session *session)
1370 {
1371         if (session->s_state >= CEPH_MDS_SESSION_CLOSING)
1372                 return 0;
1373         session->s_state = CEPH_MDS_SESSION_CLOSING;
1374         return request_close_session(mdsc, session);
1375 }
1376
1377 /*
1378  * Trim old(er) caps.
1379  *
1380  * Because we can't cache an inode without one or more caps, we do
1381  * this indirectly: if a cap is unused, we prune its aliases, at which
1382  * point the inode will hopefully get dropped to.
1383  *
1384  * Yes, this is a bit sloppy.  Our only real goal here is to respond to
1385  * memory pressure from the MDS, though, so it needn't be perfect.
1386  */
1387 static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
1388 {
1389         struct ceph_mds_session *session = arg;
1390         struct ceph_inode_info *ci = ceph_inode(inode);
1391         int used, wanted, oissued, mine;
1392
1393         if (session->s_trim_caps <= 0)
1394                 return -1;
1395
1396         spin_lock(&ci->i_ceph_lock);
1397         mine = cap->issued | cap->implemented;
1398         used = __ceph_caps_used(ci);
1399         wanted = __ceph_caps_file_wanted(ci);
1400         oissued = __ceph_caps_issued_other(ci, cap);
1401
1402         dout("trim_caps_cb %p cap %p mine %s oissued %s used %s wanted %s\n",
1403              inode, cap, ceph_cap_string(mine), ceph_cap_string(oissued),
1404              ceph_cap_string(used), ceph_cap_string(wanted));
1405         if (cap == ci->i_auth_cap) {
1406                 if (ci->i_dirty_caps || ci->i_flushing_caps ||
1407                     !list_empty(&ci->i_cap_snaps))
1408                         goto out;
1409                 if ((used | wanted) & CEPH_CAP_ANY_WR)
1410                         goto out;
1411         }
1412         /* The inode has cached pages, but it's no longer used.
1413          * we can safely drop it */
1414         if (wanted == 0 && used == CEPH_CAP_FILE_CACHE &&
1415             !(oissued & CEPH_CAP_FILE_CACHE)) {
1416           used = 0;
1417           oissued = 0;
1418         }
1419         if ((used | wanted) & ~oissued & mine)
1420                 goto out;   /* we need these caps */
1421
1422         session->s_trim_caps--;
1423         if (oissued) {
1424                 /* we aren't the only cap.. just remove us */
1425                 __ceph_remove_cap(cap, true);
1426         } else {
1427                 /* try dropping referring dentries */
1428                 spin_unlock(&ci->i_ceph_lock);
1429                 d_prune_aliases(inode);
1430                 dout("trim_caps_cb %p cap %p  pruned, count now %d\n",
1431                      inode, cap, atomic_read(&inode->i_count));
1432                 return 0;
1433         }
1434
1435 out:
1436         spin_unlock(&ci->i_ceph_lock);
1437         return 0;
1438 }
1439
1440 /*
1441  * Trim session cap count down to some max number.
1442  */
1443 static int trim_caps(struct ceph_mds_client *mdsc,
1444                      struct ceph_mds_session *session,
1445                      int max_caps)
1446 {
1447         int trim_caps = session->s_nr_caps - max_caps;
1448
1449         dout("trim_caps mds%d start: %d / %d, trim %d\n",
1450              session->s_mds, session->s_nr_caps, max_caps, trim_caps);
1451         if (trim_caps > 0) {
1452                 session->s_trim_caps = trim_caps;
1453                 iterate_session_caps(session, trim_caps_cb, session);
1454                 dout("trim_caps mds%d done: %d / %d, trimmed %d\n",
1455                      session->s_mds, session->s_nr_caps, max_caps,
1456                         trim_caps - session->s_trim_caps);
1457                 session->s_trim_caps = 0;
1458         }
1459
1460         ceph_send_cap_releases(mdsc, session);
1461         return 0;
1462 }
1463
1464 static int check_capsnap_flush(struct ceph_inode_info *ci,
1465                                u64 want_snap_seq)
1466 {
1467         int ret = 1;
1468         spin_lock(&ci->i_ceph_lock);
1469         if (want_snap_seq > 0 && !list_empty(&ci->i_cap_snaps)) {
1470                 struct ceph_cap_snap *capsnap =
1471                         list_first_entry(&ci->i_cap_snaps,
1472                                          struct ceph_cap_snap, ci_item);
1473                 ret = capsnap->follows >= want_snap_seq;
1474         }
1475         spin_unlock(&ci->i_ceph_lock);
1476         return ret;
1477 }
1478
1479 static int check_caps_flush(struct ceph_mds_client *mdsc,
1480                             u64 want_flush_tid)
1481 {
1482         struct rb_node *n;
1483         struct ceph_cap_flush *cf;
1484         int ret = 1;
1485
1486         spin_lock(&mdsc->cap_dirty_lock);
1487         n = rb_first(&mdsc->cap_flush_tree);
1488         cf = n ? rb_entry(n, struct ceph_cap_flush, g_node) : NULL;
1489         if (cf && cf->tid <= want_flush_tid) {
1490                 dout("check_caps_flush still flushing tid %llu <= %llu\n",
1491                      cf->tid, want_flush_tid);
1492                 ret = 0;
1493         }
1494         spin_unlock(&mdsc->cap_dirty_lock);
1495         return ret;
1496 }
1497
1498 /*
1499  * flush all dirty inode data to disk.
1500  *
1501  * returns true if we've flushed through want_flush_tid
1502  */
1503 static void wait_caps_flush(struct ceph_mds_client *mdsc,
1504                             u64 want_flush_tid, u64 want_snap_seq)
1505 {
1506         int mds;
1507
1508         dout("check_caps_flush want %llu snap want %llu\n",
1509              want_flush_tid, want_snap_seq);
1510         mutex_lock(&mdsc->mutex);
1511         for (mds = 0; mds < mdsc->max_sessions; ) {
1512                 struct ceph_mds_session *session = mdsc->sessions[mds];
1513                 struct inode *inode = NULL;
1514
1515                 if (!session) {
1516                         mds++;
1517                         continue;
1518                 }
1519                 get_session(session);
1520                 mutex_unlock(&mdsc->mutex);
1521
1522                 mutex_lock(&session->s_mutex);
1523                 if (!list_empty(&session->s_cap_snaps_flushing)) {
1524                         struct ceph_cap_snap *capsnap =
1525                                 list_first_entry(&session->s_cap_snaps_flushing,
1526                                                  struct ceph_cap_snap,
1527                                                  flushing_item);
1528                         struct ceph_inode_info *ci = capsnap->ci;
1529                         if (!check_capsnap_flush(ci, want_snap_seq)) {
1530                                 dout("check_cap_flush still flushing snap %p "
1531                                      "follows %lld <= %lld to mds%d\n",
1532                                      &ci->vfs_inode, capsnap->follows,
1533                                      want_snap_seq, mds);
1534                                 inode = igrab(&ci->vfs_inode);
1535                         }
1536                 }
1537                 mutex_unlock(&session->s_mutex);
1538                 ceph_put_mds_session(session);
1539
1540                 if (inode) {
1541                         wait_event(mdsc->cap_flushing_wq,
1542                                    check_capsnap_flush(ceph_inode(inode),
1543                                                        want_snap_seq));
1544                         iput(inode);
1545                 } else {
1546                         mds++;
1547                 }
1548
1549                 mutex_lock(&mdsc->mutex);
1550         }
1551         mutex_unlock(&mdsc->mutex);
1552
1553         wait_event(mdsc->cap_flushing_wq,
1554                    check_caps_flush(mdsc, want_flush_tid));
1555
1556         dout("check_caps_flush ok, flushed thru %llu\n", want_flush_tid);
1557 }
1558
1559 /*
1560  * called under s_mutex
1561  */
1562 void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
1563                             struct ceph_mds_session *session)
1564 {
1565         struct ceph_msg *msg = NULL;
1566         struct ceph_mds_cap_release *head;
1567         struct ceph_mds_cap_item *item;
1568         struct ceph_cap *cap;
1569         LIST_HEAD(tmp_list);
1570         int num_cap_releases;
1571
1572         spin_lock(&session->s_cap_lock);
1573 again:
1574         list_splice_init(&session->s_cap_releases, &tmp_list);
1575         num_cap_releases = session->s_num_cap_releases;
1576         session->s_num_cap_releases = 0;
1577         spin_unlock(&session->s_cap_lock);
1578
1579         while (!list_empty(&tmp_list)) {
1580                 if (!msg) {
1581                         msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE,
1582                                         PAGE_SIZE, GFP_NOFS, false);
1583                         if (!msg)
1584                                 goto out_err;
1585                         head = msg->front.iov_base;
1586                         head->num = cpu_to_le32(0);
1587                         msg->front.iov_len = sizeof(*head);
1588                 }
1589                 cap = list_first_entry(&tmp_list, struct ceph_cap,
1590                                         session_caps);
1591                 list_del(&cap->session_caps);
1592                 num_cap_releases--;
1593
1594                 head = msg->front.iov_base;
1595                 le32_add_cpu(&head->num, 1);
1596                 item = msg->front.iov_base + msg->front.iov_len;
1597                 item->ino = cpu_to_le64(cap->cap_ino);
1598                 item->cap_id = cpu_to_le64(cap->cap_id);
1599                 item->migrate_seq = cpu_to_le32(cap->mseq);
1600                 item->seq = cpu_to_le32(cap->issue_seq);
1601                 msg->front.iov_len += sizeof(*item);
1602
1603                 ceph_put_cap(mdsc, cap);
1604
1605                 if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
1606                         msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1607                         dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
1608                         ceph_con_send(&session->s_con, msg);
1609                         msg = NULL;
1610                 }
1611         }
1612
1613         BUG_ON(num_cap_releases != 0);
1614
1615         spin_lock(&session->s_cap_lock);
1616         if (!list_empty(&session->s_cap_releases))
1617                 goto again;
1618         spin_unlock(&session->s_cap_lock);
1619
1620         if (msg) {
1621                 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1622                 dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
1623                 ceph_con_send(&session->s_con, msg);
1624         }
1625         return;
1626 out_err:
1627         pr_err("send_cap_releases mds%d, failed to allocate message\n",
1628                 session->s_mds);
1629         spin_lock(&session->s_cap_lock);
1630         list_splice(&tmp_list, &session->s_cap_releases);
1631         session->s_num_cap_releases += num_cap_releases;
1632         spin_unlock(&session->s_cap_lock);
1633 }
1634
1635 /*
1636  * requests
1637  */
1638
1639 int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
1640                                     struct inode *dir)
1641 {
1642         struct ceph_inode_info *ci = ceph_inode(dir);
1643         struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
1644         struct ceph_mount_options *opt = req->r_mdsc->fsc->mount_options;
1645         size_t size = sizeof(*rinfo->dir_in) + sizeof(*rinfo->dir_dname_len) +
1646                       sizeof(*rinfo->dir_dname) + sizeof(*rinfo->dir_dlease);
1647         int order, num_entries;
1648
1649         spin_lock(&ci->i_ceph_lock);
1650         num_entries = ci->i_files + ci->i_subdirs;
1651         spin_unlock(&ci->i_ceph_lock);
1652         num_entries = max(num_entries, 1);
1653         num_entries = min(num_entries, opt->max_readdir);
1654
1655         order = get_order(size * num_entries);
1656         while (order >= 0) {
1657                 rinfo->dir_in = (void*)__get_free_pages(GFP_KERNEL |
1658                                                         __GFP_NOWARN,
1659                                                         order);
1660                 if (rinfo->dir_in)
1661                         break;
1662                 order--;
1663         }
1664         if (!rinfo->dir_in)
1665                 return -ENOMEM;
1666
1667         num_entries = (PAGE_SIZE << order) / size;
1668         num_entries = min(num_entries, opt->max_readdir);
1669
1670         rinfo->dir_buf_size = PAGE_SIZE << order;
1671         req->r_num_caps = num_entries + 1;
1672         req->r_args.readdir.max_entries = cpu_to_le32(num_entries);
1673         req->r_args.readdir.max_bytes = cpu_to_le32(opt->max_readdir_bytes);
1674         return 0;
1675 }
1676
1677 /*
1678  * Create an mds request.
1679  */
1680 struct ceph_mds_request *
1681 ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
1682 {
1683         struct ceph_mds_request *req = kzalloc(sizeof(*req), GFP_NOFS);
1684
1685         if (!req)
1686                 return ERR_PTR(-ENOMEM);
1687
1688         mutex_init(&req->r_fill_mutex);
1689         req->r_mdsc = mdsc;
1690         req->r_started = jiffies;
1691         req->r_resend_mds = -1;
1692         INIT_LIST_HEAD(&req->r_unsafe_dir_item);
1693         INIT_LIST_HEAD(&req->r_unsafe_target_item);
1694         req->r_fmode = -1;
1695         kref_init(&req->r_kref);
1696         RB_CLEAR_NODE(&req->r_node);
1697         INIT_LIST_HEAD(&req->r_wait);
1698         init_completion(&req->r_completion);
1699         init_completion(&req->r_safe_completion);
1700         INIT_LIST_HEAD(&req->r_unsafe_item);
1701
1702         req->r_stamp = current_fs_time(mdsc->fsc->sb);
1703
1704         req->r_op = op;
1705         req->r_direct_mode = mode;
1706         return req;
1707 }
1708
1709 /*
1710  * return oldest (lowest) request, tid in request tree, 0 if none.
1711  *
1712  * called under mdsc->mutex.
1713  */
1714 static struct ceph_mds_request *__get_oldest_req(struct ceph_mds_client *mdsc)
1715 {
1716         if (RB_EMPTY_ROOT(&mdsc->request_tree))
1717                 return NULL;
1718         return rb_entry(rb_first(&mdsc->request_tree),
1719                         struct ceph_mds_request, r_node);
1720 }
1721
1722 static inline  u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
1723 {
1724         return mdsc->oldest_tid;
1725 }
1726
1727 /*
1728  * Build a dentry's path.  Allocate on heap; caller must kfree.  Based
1729  * on build_path_from_dentry in fs/cifs/dir.c.
1730  *
1731  * If @stop_on_nosnap, generate path relative to the first non-snapped
1732  * inode.
1733  *
1734  * Encode hidden .snap dirs as a double /, i.e.
1735  *   foo/.snap/bar -> foo//bar
1736  */
1737 char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *base,
1738                            int stop_on_nosnap)
1739 {
1740         struct dentry *temp;
1741         char *path;
1742         int len, pos;
1743         unsigned seq;
1744
1745         if (dentry == NULL)
1746                 return ERR_PTR(-EINVAL);
1747
1748 retry:
1749         len = 0;
1750         seq = read_seqbegin(&rename_lock);
1751         rcu_read_lock();
1752         for (temp = dentry; !IS_ROOT(temp);) {
1753                 struct inode *inode = d_inode(temp);
1754                 if (inode && ceph_snap(inode) == CEPH_SNAPDIR)
1755                         len++;  /* slash only */
1756                 else if (stop_on_nosnap && inode &&
1757                          ceph_snap(inode) == CEPH_NOSNAP)
1758                         break;
1759                 else
1760                         len += 1 + temp->d_name.len;
1761                 temp = temp->d_parent;
1762         }
1763         rcu_read_unlock();
1764         if (len)
1765                 len--;  /* no leading '/' */
1766
1767         path = kmalloc(len+1, GFP_NOFS);
1768         if (path == NULL)
1769                 return ERR_PTR(-ENOMEM);
1770         pos = len;
1771         path[pos] = 0;  /* trailing null */
1772         rcu_read_lock();
1773         for (temp = dentry; !IS_ROOT(temp) && pos != 0; ) {
1774                 struct inode *inode;
1775
1776                 spin_lock(&temp->d_lock);
1777                 inode = d_inode(temp);
1778                 if (inode && ceph_snap(inode) == CEPH_SNAPDIR) {
1779                         dout("build_path path+%d: %p SNAPDIR\n",
1780                              pos, temp);
1781                 } else if (stop_on_nosnap && inode &&
1782                            ceph_snap(inode) == CEPH_NOSNAP) {
1783                         spin_unlock(&temp->d_lock);
1784                         break;
1785                 } else {
1786                         pos -= temp->d_name.len;
1787                         if (pos < 0) {
1788                                 spin_unlock(&temp->d_lock);
1789                                 break;
1790                         }
1791                         strncpy(path + pos, temp->d_name.name,
1792                                 temp->d_name.len);
1793                 }
1794                 spin_unlock(&temp->d_lock);
1795                 if (pos)
1796                         path[--pos] = '/';
1797                 temp = temp->d_parent;
1798         }
1799         rcu_read_unlock();
1800         if (pos != 0 || read_seqretry(&rename_lock, seq)) {
1801                 pr_err("build_path did not end path lookup where "
1802                        "expected, namelen is %d, pos is %d\n", len, pos);
1803                 /* presumably this is only possible if racing with a
1804                    rename of one of the parent directories (we can not
1805                    lock the dentries above us to prevent this, but
1806                    retrying should be harmless) */
1807                 kfree(path);
1808                 goto retry;
1809         }
1810
1811         *base = ceph_ino(d_inode(temp));
1812         *plen = len;
1813         dout("build_path on %p %d built %llx '%.*s'\n",
1814              dentry, d_count(dentry), *base, len, path);
1815         return path;
1816 }
1817
1818 static int build_dentry_path(struct dentry *dentry,
1819                              const char **ppath, int *ppathlen, u64 *pino,
1820                              int *pfreepath)
1821 {
1822         char *path;
1823
1824         if (ceph_snap(d_inode(dentry->d_parent)) == CEPH_NOSNAP) {
1825                 *pino = ceph_ino(d_inode(dentry->d_parent));
1826                 *ppath = dentry->d_name.name;
1827                 *ppathlen = dentry->d_name.len;
1828                 return 0;
1829         }
1830         path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
1831         if (IS_ERR(path))
1832                 return PTR_ERR(path);
1833         *ppath = path;
1834         *pfreepath = 1;
1835         return 0;
1836 }
1837
1838 static int build_inode_path(struct inode *inode,
1839                             const char **ppath, int *ppathlen, u64 *pino,
1840                             int *pfreepath)
1841 {
1842         struct dentry *dentry;
1843         char *path;
1844
1845         if (ceph_snap(inode) == CEPH_NOSNAP) {
1846                 *pino = ceph_ino(inode);
1847                 *ppathlen = 0;
1848                 return 0;
1849         }
1850         dentry = d_find_alias(inode);
1851         path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
1852         dput(dentry);
1853         if (IS_ERR(path))
1854                 return PTR_ERR(path);
1855         *ppath = path;
1856         *pfreepath = 1;
1857         return 0;
1858 }
1859
1860 /*
1861  * request arguments may be specified via an inode *, a dentry *, or
1862  * an explicit ino+path.
1863  */
1864 static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry,
1865                                   const char *rpath, u64 rino,
1866                                   const char **ppath, int *pathlen,
1867                                   u64 *ino, int *freepath)
1868 {
1869         int r = 0;
1870
1871         if (rinode) {
1872                 r = build_inode_path(rinode, ppath, pathlen, ino, freepath);
1873                 dout(" inode %p %llx.%llx\n", rinode, ceph_ino(rinode),
1874                      ceph_snap(rinode));
1875         } else if (rdentry) {
1876                 r = build_dentry_path(rdentry, ppath, pathlen, ino, freepath);
1877                 dout(" dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen,
1878                      *ppath);
1879         } else if (rpath || rino) {
1880                 *ino = rino;
1881                 *ppath = rpath;
1882                 *pathlen = rpath ? strlen(rpath) : 0;
1883                 dout(" path %.*s\n", *pathlen, rpath);
1884         }
1885
1886         return r;
1887 }
1888
1889 /*
1890  * called under mdsc->mutex
1891  */
1892 static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
1893                                                struct ceph_mds_request *req,
1894                                                int mds, bool drop_cap_releases)
1895 {
1896         struct ceph_msg *msg;
1897         struct ceph_mds_request_head *head;
1898         const char *path1 = NULL;
1899         const char *path2 = NULL;
1900         u64 ino1 = 0, ino2 = 0;
1901         int pathlen1 = 0, pathlen2 = 0;
1902         int freepath1 = 0, freepath2 = 0;
1903         int len;
1904         u16 releases;
1905         void *p, *end;
1906         int ret;
1907
1908         ret = set_request_path_attr(req->r_inode, req->r_dentry,
1909                               req->r_path1, req->r_ino1.ino,
1910                               &path1, &pathlen1, &ino1, &freepath1);
1911         if (ret < 0) {
1912                 msg = ERR_PTR(ret);
1913                 goto out;
1914         }
1915
1916         ret = set_request_path_attr(NULL, req->r_old_dentry,
1917                               req->r_path2, req->r_ino2.ino,
1918                               &path2, &pathlen2, &ino2, &freepath2);
1919         if (ret < 0) {
1920                 msg = ERR_PTR(ret);
1921                 goto out_free1;
1922         }
1923
1924         len = sizeof(*head) +
1925                 pathlen1 + pathlen2 + 2*(1 + sizeof(u32) + sizeof(u64)) +
1926                 sizeof(struct ceph_timespec);
1927
1928         /* calculate (max) length for cap releases */
1929         len += sizeof(struct ceph_mds_request_release) *
1930                 (!!req->r_inode_drop + !!req->r_dentry_drop +
1931                  !!req->r_old_inode_drop + !!req->r_old_dentry_drop);
1932         if (req->r_dentry_drop)
1933                 len += req->r_dentry->d_name.len;
1934         if (req->r_old_dentry_drop)
1935                 len += req->r_old_dentry->d_name.len;
1936
1937         msg = ceph_msg_new(CEPH_MSG_CLIENT_REQUEST, len, GFP_NOFS, false);
1938         if (!msg) {
1939                 msg = ERR_PTR(-ENOMEM);
1940                 goto out_free2;
1941         }
1942
1943         msg->hdr.version = cpu_to_le16(2);
1944         msg->hdr.tid = cpu_to_le64(req->r_tid);
1945
1946         head = msg->front.iov_base;
1947         p = msg->front.iov_base + sizeof(*head);
1948         end = msg->front.iov_base + msg->front.iov_len;
1949
1950         head->mdsmap_epoch = cpu_to_le32(mdsc->mdsmap->m_epoch);
1951         head->op = cpu_to_le32(req->r_op);
1952         head->caller_uid = cpu_to_le32(from_kuid(&init_user_ns, req->r_uid));
1953         head->caller_gid = cpu_to_le32(from_kgid(&init_user_ns, req->r_gid));
1954         head->args = req->r_args;
1955
1956         ceph_encode_filepath(&p, end, ino1, path1);
1957         ceph_encode_filepath(&p, end, ino2, path2);
1958
1959         /* make note of release offset, in case we need to replay */
1960         req->r_request_release_offset = p - msg->front.iov_base;
1961
1962         /* cap releases */
1963         releases = 0;
1964         if (req->r_inode_drop)
1965                 releases += ceph_encode_inode_release(&p,
1966                       req->r_inode ? req->r_inode : d_inode(req->r_dentry),
1967                       mds, req->r_inode_drop, req->r_inode_unless, 0);
1968         if (req->r_dentry_drop)
1969                 releases += ceph_encode_dentry_release(&p, req->r_dentry,
1970                        mds, req->r_dentry_drop, req->r_dentry_unless);
1971         if (req->r_old_dentry_drop)
1972                 releases += ceph_encode_dentry_release(&p, req->r_old_dentry,
1973                        mds, req->r_old_dentry_drop, req->r_old_dentry_unless);
1974         if (req->r_old_inode_drop)
1975                 releases += ceph_encode_inode_release(&p,
1976                       d_inode(req->r_old_dentry),
1977                       mds, req->r_old_inode_drop, req->r_old_inode_unless, 0);
1978
1979         if (drop_cap_releases) {
1980                 releases = 0;
1981                 p = msg->front.iov_base + req->r_request_release_offset;
1982         }
1983
1984         head->num_releases = cpu_to_le16(releases);
1985
1986         /* time stamp */
1987         {
1988                 struct ceph_timespec ts;
1989                 ceph_encode_timespec(&ts, &req->r_stamp);
1990                 ceph_encode_copy(&p, &ts, sizeof(ts));
1991         }
1992
1993         BUG_ON(p > end);
1994         msg->front.iov_len = p - msg->front.iov_base;
1995         msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1996
1997         if (req->r_pagelist) {
1998                 struct ceph_pagelist *pagelist = req->r_pagelist;
1999                 atomic_inc(&pagelist->refcnt);
2000                 ceph_msg_data_add_pagelist(msg, pagelist);
2001                 msg->hdr.data_len = cpu_to_le32(pagelist->length);
2002         } else {
2003                 msg->hdr.data_len = 0;
2004         }
2005
2006         msg->hdr.data_off = cpu_to_le16(0);
2007
2008 out_free2:
2009         if (freepath2)
2010                 kfree((char *)path2);
2011 out_free1:
2012         if (freepath1)
2013                 kfree((char *)path1);
2014 out:
2015         return msg;
2016 }
2017
2018 /*
2019  * called under mdsc->mutex if error, under no mutex if
2020  * success.
2021  */
2022 static void complete_request(struct ceph_mds_client *mdsc,
2023                              struct ceph_mds_request *req)
2024 {
2025         if (req->r_callback)
2026                 req->r_callback(mdsc, req);
2027         else
2028                 complete_all(&req->r_completion);
2029 }
2030
2031 /*
2032  * called under mdsc->mutex
2033  */
2034 static int __prepare_send_request(struct ceph_mds_client *mdsc,
2035                                   struct ceph_mds_request *req,
2036                                   int mds, bool drop_cap_releases)
2037 {
2038         struct ceph_mds_request_head *rhead;
2039         struct ceph_msg *msg;
2040         int flags = 0;
2041
2042         req->r_attempts++;
2043         if (req->r_inode) {
2044                 struct ceph_cap *cap =
2045                         ceph_get_cap_for_mds(ceph_inode(req->r_inode), mds);
2046
2047                 if (cap)
2048                         req->r_sent_on_mseq = cap->mseq;
2049                 else
2050                         req->r_sent_on_mseq = -1;
2051         }
2052         dout("prepare_send_request %p tid %lld %s (attempt %d)\n", req,
2053              req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts);
2054
2055         if (req->r_got_unsafe) {
2056                 void *p;
2057                 /*
2058                  * Replay.  Do not regenerate message (and rebuild
2059                  * paths, etc.); just use the original message.
2060                  * Rebuilding paths will break for renames because
2061                  * d_move mangles the src name.
2062                  */
2063                 msg = req->r_request;
2064                 rhead = msg->front.iov_base;
2065
2066                 flags = le32_to_cpu(rhead->flags);
2067                 flags |= CEPH_MDS_FLAG_REPLAY;
2068                 rhead->flags = cpu_to_le32(flags);
2069
2070                 if (req->r_target_inode)
2071                         rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
2072
2073                 rhead->num_retry = req->r_attempts - 1;
2074
2075                 /* remove cap/dentry releases from message */
2076                 rhead->num_releases = 0;
2077
2078                 /* time stamp */
2079                 p = msg->front.iov_base + req->r_request_release_offset;
2080                 {
2081                         struct ceph_timespec ts;
2082                         ceph_encode_timespec(&ts, &req->r_stamp);
2083                         ceph_encode_copy(&p, &ts, sizeof(ts));
2084                 }
2085
2086                 msg->front.iov_len = p - msg->front.iov_base;
2087                 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
2088                 return 0;
2089         }
2090
2091         if (req->r_request) {
2092                 ceph_msg_put(req->r_request);
2093                 req->r_request = NULL;
2094         }
2095         msg = create_request_message(mdsc, req, mds, drop_cap_releases);
2096         if (IS_ERR(msg)) {
2097                 req->r_err = PTR_ERR(msg);
2098                 return PTR_ERR(msg);
2099         }
2100         req->r_request = msg;
2101
2102         rhead = msg->front.iov_base;
2103         rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc));
2104         if (req->r_got_unsafe)
2105                 flags |= CEPH_MDS_FLAG_REPLAY;
2106         if (req->r_locked_dir)
2107                 flags |= CEPH_MDS_FLAG_WANT_DENTRY;
2108         rhead->flags = cpu_to_le32(flags);
2109         rhead->num_fwd = req->r_num_fwd;
2110         rhead->num_retry = req->r_attempts - 1;
2111         rhead->ino = 0;
2112
2113         dout(" r_locked_dir = %p\n", req->r_locked_dir);
2114         return 0;
2115 }
2116
2117 /*
2118  * send request, or put it on the appropriate wait list.
2119  */
2120 static int __do_request(struct ceph_mds_client *mdsc,
2121                         struct ceph_mds_request *req)
2122 {
2123         struct ceph_mds_session *session = NULL;
2124         int mds = -1;
2125         int err = 0;
2126
2127         if (req->r_err || req->r_got_result) {
2128                 if (req->r_aborted)
2129                         __unregister_request(mdsc, req);
2130                 goto out;
2131         }
2132
2133         if (req->r_timeout &&
2134             time_after_eq(jiffies, req->r_started + req->r_timeout)) {
2135                 dout("do_request timed out\n");
2136                 err = -EIO;
2137                 goto finish;
2138         }
2139         if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
2140                 dout("do_request forced umount\n");
2141                 err = -EIO;
2142                 goto finish;
2143         }
2144
2145         put_request_session(req);
2146
2147         mds = __choose_mds(mdsc, req);
2148         if (mds < 0 ||
2149             ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
2150                 dout("do_request no mds or not active, waiting for map\n");
2151                 list_add(&req->r_wait, &mdsc->waiting_for_map);
2152                 goto out;
2153         }
2154
2155         /* get, open session */
2156         session = __ceph_lookup_mds_session(mdsc, mds);
2157         if (!session) {
2158                 session = register_session(mdsc, mds);
2159                 if (IS_ERR(session)) {
2160                         err = PTR_ERR(session);
2161                         goto finish;
2162                 }
2163         }
2164         req->r_session = get_session(session);
2165
2166         dout("do_request mds%d session %p state %s\n", mds, session,
2167              ceph_session_state_name(session->s_state));
2168         if (session->s_state != CEPH_MDS_SESSION_OPEN &&
2169             session->s_state != CEPH_MDS_SESSION_HUNG) {
2170                 if (session->s_state == CEPH_MDS_SESSION_NEW ||
2171                     session->s_state == CEPH_MDS_SESSION_CLOSING)
2172                         __open_session(mdsc, session);
2173                 list_add(&req->r_wait, &session->s_waiting);
2174                 goto out_session;
2175         }
2176
2177         /* send request */
2178         req->r_resend_mds = -1;   /* forget any previous mds hint */
2179
2180         if (req->r_request_started == 0)   /* note request start time */
2181                 req->r_request_started = jiffies;
2182
2183         err = __prepare_send_request(mdsc, req, mds, false);
2184         if (!err) {
2185                 ceph_msg_get(req->r_request);
2186                 ceph_con_send(&session->s_con, req->r_request);
2187         }
2188
2189 out_session:
2190         ceph_put_mds_session(session);
2191 finish:
2192         if (err) {
2193                 dout("__do_request early error %d\n", err);
2194                 req->r_err = err;
2195                 complete_request(mdsc, req);
2196                 __unregister_request(mdsc, req);
2197         }
2198 out:
2199         return err;
2200 }
2201
2202 /*
2203  * called under mdsc->mutex
2204  */
2205 static void __wake_requests(struct ceph_mds_client *mdsc,
2206                             struct list_head *head)
2207 {
2208         struct ceph_mds_request *req;
2209         LIST_HEAD(tmp_list);
2210
2211         list_splice_init(head, &tmp_list);
2212
2213         while (!list_empty(&tmp_list)) {
2214                 req = list_entry(tmp_list.next,
2215                                  struct ceph_mds_request, r_wait);
2216                 list_del_init(&req->r_wait);
2217                 dout(" wake request %p tid %llu\n", req, req->r_tid);
2218                 __do_request(mdsc, req);
2219         }
2220 }
2221
2222 /*
2223  * Wake up threads with requests pending for @mds, so that they can
2224  * resubmit their requests to a possibly different mds.
2225  */
2226 static void kick_requests(struct ceph_mds_client *mdsc, int mds)
2227 {
2228         struct ceph_mds_request *req;
2229         struct rb_node *p = rb_first(&mdsc->request_tree);
2230
2231         dout("kick_requests mds%d\n", mds);
2232         while (p) {
2233                 req = rb_entry(p, struct ceph_mds_request, r_node);
2234                 p = rb_next(p);
2235                 if (req->r_got_unsafe)
2236                         continue;
2237                 if (req->r_attempts > 0)
2238                         continue; /* only new requests */
2239                 if (req->r_session &&
2240                     req->r_session->s_mds == mds) {
2241                         dout(" kicking tid %llu\n", req->r_tid);
2242                         list_del_init(&req->r_wait);
2243                         __do_request(mdsc, req);
2244                 }
2245         }
2246 }
2247
2248 void ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
2249                               struct ceph_mds_request *req)
2250 {
2251         dout("submit_request on %p\n", req);
2252         mutex_lock(&mdsc->mutex);
2253         __register_request(mdsc, req, NULL);
2254         __do_request(mdsc, req);
2255         mutex_unlock(&mdsc->mutex);
2256 }
2257
2258 /*
2259  * Synchrously perform an mds request.  Take care of all of the
2260  * session setup, forwarding, retry details.
2261  */
2262 int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
2263                          struct inode *dir,
2264                          struct ceph_mds_request *req)
2265 {
2266         int err;
2267
2268         dout("do_request on %p\n", req);
2269
2270         /* take CAP_PIN refs for r_inode, r_locked_dir, r_old_dentry */
2271         if (req->r_inode)
2272                 ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
2273         if (req->r_locked_dir)
2274                 ceph_get_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN);
2275         if (req->r_old_dentry_dir)
2276                 ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir),
2277                                   CEPH_CAP_PIN);
2278
2279         /* deny access to directories with pool_ns layouts */
2280         if (req->r_inode && S_ISDIR(req->r_inode->i_mode) &&
2281             ceph_inode(req->r_inode)->i_pool_ns_len)
2282                 return -EIO;
2283         if (req->r_locked_dir &&
2284             ceph_inode(req->r_locked_dir)->i_pool_ns_len)
2285                 return -EIO;
2286
2287         /* issue */
2288         mutex_lock(&mdsc->mutex);
2289         __register_request(mdsc, req, dir);
2290         __do_request(mdsc, req);
2291
2292         if (req->r_err) {
2293                 err = req->r_err;
2294                 goto out;
2295         }
2296
2297         /* wait */
2298         mutex_unlock(&mdsc->mutex);
2299         dout("do_request waiting\n");
2300         if (!req->r_timeout && req->r_wait_for_completion) {
2301                 err = req->r_wait_for_completion(mdsc, req);
2302         } else {
2303                 long timeleft = wait_for_completion_killable_timeout(
2304                                         &req->r_completion,
2305                                         ceph_timeout_jiffies(req->r_timeout));
2306                 if (timeleft > 0)
2307                         err = 0;
2308                 else if (!timeleft)
2309                         err = -EIO;  /* timed out */
2310                 else
2311                         err = timeleft;  /* killed */
2312         }
2313         dout("do_request waited, got %d\n", err);
2314         mutex_lock(&mdsc->mutex);
2315
2316         /* only abort if we didn't race with a real reply */
2317         if (req->r_got_result) {
2318                 err = le32_to_cpu(req->r_reply_info.head->result);
2319         } else if (err < 0) {
2320                 dout("aborted request %lld with %d\n", req->r_tid, err);
2321
2322                 /*
2323                  * ensure we aren't running concurrently with
2324                  * ceph_fill_trace or ceph_readdir_prepopulate, which
2325                  * rely on locks (dir mutex) held by our caller.
2326                  */
2327                 mutex_lock(&req->r_fill_mutex);
2328                 req->r_err = err;
2329                 req->r_aborted = true;
2330                 mutex_unlock(&req->r_fill_mutex);
2331
2332                 if (req->r_locked_dir &&
2333                     (req->r_op & CEPH_MDS_OP_WRITE))
2334                         ceph_invalidate_dir_request(req);
2335         } else {
2336                 err = req->r_err;
2337         }
2338
2339 out:
2340         mutex_unlock(&mdsc->mutex);
2341         dout("do_request %p done, result %d\n", req, err);
2342         return err;
2343 }
2344
2345 /*
2346  * Invalidate dir's completeness, dentry lease state on an aborted MDS
2347  * namespace request.
2348  */
2349 void ceph_invalidate_dir_request(struct ceph_mds_request *req)
2350 {
2351         struct inode *inode = req->r_locked_dir;
2352
2353         dout("invalidate_dir_request %p (complete, lease(s))\n", inode);
2354
2355         ceph_dir_clear_complete(inode);
2356         if (req->r_dentry)
2357                 ceph_invalidate_dentry_lease(req->r_dentry);
2358         if (req->r_old_dentry)
2359                 ceph_invalidate_dentry_lease(req->r_old_dentry);
2360 }
2361
2362 /*
2363  * Handle mds reply.
2364  *
2365  * We take the session mutex and parse and process the reply immediately.
2366  * This preserves the logical ordering of replies, capabilities, etc., sent
2367  * by the MDS as they are applied to our local cache.
2368  */
2369 static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
2370 {
2371         struct ceph_mds_client *mdsc = session->s_mdsc;
2372         struct ceph_mds_request *req;
2373         struct ceph_mds_reply_head *head = msg->front.iov_base;
2374         struct ceph_mds_reply_info_parsed *rinfo;  /* parsed reply info */
2375         struct ceph_snap_realm *realm;
2376         u64 tid;
2377         int err, result;
2378         int mds = session->s_mds;
2379
2380         if (msg->front.iov_len < sizeof(*head)) {
2381                 pr_err("mdsc_handle_reply got corrupt (short) reply\n");
2382                 ceph_msg_dump(msg);
2383                 return;
2384         }
2385
2386         /* get request, session */
2387         tid = le64_to_cpu(msg->hdr.tid);
2388         mutex_lock(&mdsc->mutex);
2389         req = lookup_get_request(mdsc, tid);
2390         if (!req) {
2391                 dout("handle_reply on unknown tid %llu\n", tid);
2392                 mutex_unlock(&mdsc->mutex);
2393                 return;
2394         }
2395         dout("handle_reply %p\n", req);
2396
2397         /* correct session? */
2398         if (req->r_session != session) {
2399                 pr_err("mdsc_handle_reply got %llu on session mds%d"
2400                        " not mds%d\n", tid, session->s_mds,
2401                        req->r_session ? req->r_session->s_mds : -1);
2402                 mutex_unlock(&mdsc->mutex);
2403                 goto out;
2404         }
2405
2406         /* dup? */
2407         if ((req->r_got_unsafe && !head->safe) ||
2408             (req->r_got_safe && head->safe)) {
2409                 pr_warn("got a dup %s reply on %llu from mds%d\n",
2410                            head->safe ? "safe" : "unsafe", tid, mds);
2411                 mutex_unlock(&mdsc->mutex);
2412                 goto out;
2413         }
2414         if (req->r_got_safe) {
2415                 pr_warn("got unsafe after safe on %llu from mds%d\n",
2416                            tid, mds);
2417                 mutex_unlock(&mdsc->mutex);
2418                 goto out;
2419         }
2420
2421         result = le32_to_cpu(head->result);
2422
2423         /*
2424          * Handle an ESTALE
2425          * if we're not talking to the authority, send to them
2426          * if the authority has changed while we weren't looking,
2427          * send to new authority
2428          * Otherwise we just have to return an ESTALE
2429          */
2430         if (result == -ESTALE) {
2431                 dout("got ESTALE on request %llu", req->r_tid);
2432                 req->r_resend_mds = -1;
2433                 if (req->r_direct_mode != USE_AUTH_MDS) {
2434                         dout("not using auth, setting for that now");
2435                         req->r_direct_mode = USE_AUTH_MDS;
2436                         __do_request(mdsc, req);
2437                         mutex_unlock(&mdsc->mutex);
2438                         goto out;
2439                 } else  {
2440                         int mds = __choose_mds(mdsc, req);
2441                         if (mds >= 0 && mds != req->r_session->s_mds) {
2442                                 dout("but auth changed, so resending");
2443                                 __do_request(mdsc, req);
2444                                 mutex_unlock(&mdsc->mutex);
2445                                 goto out;
2446                         }
2447                 }
2448                 dout("have to return ESTALE on request %llu", req->r_tid);
2449         }
2450
2451
2452         if (head->safe) {
2453                 req->r_got_safe = true;
2454                 __unregister_request(mdsc, req);
2455
2456                 if (req->r_got_unsafe) {
2457                         /*
2458                          * We already handled the unsafe response, now do the
2459                          * cleanup.  No need to examine the response; the MDS
2460                          * doesn't include any result info in the safe
2461                          * response.  And even if it did, there is nothing
2462                          * useful we could do with a revised return value.
2463                          */
2464                         dout("got safe reply %llu, mds%d\n", tid, mds);
2465                         list_del_init(&req->r_unsafe_item);
2466
2467                         /* last unsafe request during umount? */
2468                         if (mdsc->stopping && !__get_oldest_req(mdsc))
2469                                 complete_all(&mdsc->safe_umount_waiters);
2470                         mutex_unlock(&mdsc->mutex);
2471                         goto out;
2472                 }
2473         } else {
2474                 req->r_got_unsafe = true;
2475                 list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe);
2476                 if (req->r_unsafe_dir) {
2477                         struct ceph_inode_info *ci =
2478                                         ceph_inode(req->r_unsafe_dir);
2479                         spin_lock(&ci->i_unsafe_lock);
2480                         list_add_tail(&req->r_unsafe_dir_item,
2481                                       &ci->i_unsafe_dirops);
2482                         spin_unlock(&ci->i_unsafe_lock);
2483                 }
2484         }
2485
2486         dout("handle_reply tid %lld result %d\n", tid, result);
2487         rinfo = &req->r_reply_info;
2488         err = parse_reply_info(msg, rinfo, session->s_con.peer_features);
2489         mutex_unlock(&mdsc->mutex);
2490
2491         mutex_lock(&session->s_mutex);
2492         if (err < 0) {
2493                 pr_err("mdsc_handle_reply got corrupt reply mds%d(tid:%lld)\n", mds, tid);
2494                 ceph_msg_dump(msg);
2495                 goto out_err;
2496         }
2497
2498         /* snap trace */
2499         realm = NULL;
2500         if (rinfo->snapblob_len) {
2501                 down_write(&mdsc->snap_rwsem);
2502                 ceph_update_snap_trace(mdsc, rinfo->snapblob,
2503                                 rinfo->snapblob + rinfo->snapblob_len,
2504                                 le32_to_cpu(head->op) == CEPH_MDS_OP_RMSNAP,
2505                                 &realm);
2506                 downgrade_write(&mdsc->snap_rwsem);
2507         } else {
2508                 down_read(&mdsc->snap_rwsem);
2509         }
2510
2511         /* insert trace into our cache */
2512         mutex_lock(&req->r_fill_mutex);
2513         current->journal_info = req;
2514         err = ceph_fill_trace(mdsc->fsc->sb, req, req->r_session);
2515         if (err == 0) {
2516                 if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
2517                                     req->r_op == CEPH_MDS_OP_LSSNAP))
2518                         ceph_readdir_prepopulate(req, req->r_session);
2519                 ceph_unreserve_caps(mdsc, &req->r_caps_reservation);
2520         }
2521         current->journal_info = NULL;
2522         mutex_unlock(&req->r_fill_mutex);
2523
2524         up_read(&mdsc->snap_rwsem);
2525         if (realm)
2526                 ceph_put_snap_realm(mdsc, realm);
2527
2528         if (err == 0 && req->r_got_unsafe && req->r_target_inode) {
2529                 struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
2530                 spin_lock(&ci->i_unsafe_lock);
2531                 list_add_tail(&req->r_unsafe_target_item, &ci->i_unsafe_iops);
2532                 spin_unlock(&ci->i_unsafe_lock);
2533         }
2534 out_err:
2535         mutex_lock(&mdsc->mutex);
2536         if (!req->r_aborted) {
2537                 if (err) {
2538                         req->r_err = err;
2539                 } else {
2540                         req->r_reply =  ceph_msg_get(msg);
2541                         req->r_got_result = true;
2542                 }
2543         } else {
2544                 dout("reply arrived after request %lld was aborted\n", tid);
2545         }
2546         mutex_unlock(&mdsc->mutex);
2547
2548         mutex_unlock(&session->s_mutex);
2549
2550         /* kick calling process */
2551         complete_request(mdsc, req);
2552 out:
2553         ceph_mdsc_put_request(req);
2554         return;
2555 }
2556
2557
2558
2559 /*
2560  * handle mds notification that our request has been forwarded.
2561  */
2562 static void handle_forward(struct ceph_mds_client *mdsc,
2563                            struct ceph_mds_session *session,
2564                            struct ceph_msg *msg)
2565 {
2566         struct ceph_mds_request *req;
2567         u64 tid = le64_to_cpu(msg->hdr.tid);
2568         u32 next_mds;
2569         u32 fwd_seq;
2570         int err = -EINVAL;
2571         void *p = msg->front.iov_base;
2572         void *end = p + msg->front.iov_len;
2573
2574         ceph_decode_need(&p, end, 2*sizeof(u32), bad);
2575         next_mds = ceph_decode_32(&p);
2576         fwd_seq = ceph_decode_32(&p);
2577
2578         mutex_lock(&mdsc->mutex);
2579         req = lookup_get_request(mdsc, tid);
2580         if (!req) {
2581                 dout("forward tid %llu to mds%d - req dne\n", tid, next_mds);
2582                 goto out;  /* dup reply? */
2583         }
2584
2585         if (req->r_aborted) {
2586                 dout("forward tid %llu aborted, unregistering\n", tid);
2587                 __unregister_request(mdsc, req);
2588         } else if (fwd_seq <= req->r_num_fwd) {
2589                 dout("forward tid %llu to mds%d - old seq %d <= %d\n",
2590                      tid, next_mds, req->r_num_fwd, fwd_seq);
2591         } else {
2592                 /* resend. forward race not possible; mds would drop */
2593                 dout("forward tid %llu to mds%d (we resend)\n", tid, next_mds);
2594                 BUG_ON(req->r_err);
2595                 BUG_ON(req->r_got_result);
2596                 req->r_attempts = 0;
2597                 req->r_num_fwd = fwd_seq;
2598                 req->r_resend_mds = next_mds;
2599                 put_request_session(req);
2600                 __do_request(mdsc, req);
2601         }
2602         ceph_mdsc_put_request(req);
2603 out:
2604         mutex_unlock(&mdsc->mutex);
2605         return;
2606
2607 bad:
2608         pr_err("mdsc_handle_forward decode error err=%d\n", err);
2609 }
2610
2611 /*
2612  * handle a mds session control message
2613  */
2614 static void handle_session(struct ceph_mds_session *session,
2615                            struct ceph_msg *msg)
2616 {
2617         struct ceph_mds_client *mdsc = session->s_mdsc;
2618         u32 op;
2619         u64 seq;
2620         int mds = session->s_mds;
2621         struct ceph_mds_session_head *h = msg->front.iov_base;
2622         int wake = 0;
2623
2624         /* decode */
2625         if (msg->front.iov_len != sizeof(*h))
2626                 goto bad;
2627         op = le32_to_cpu(h->op);
2628         seq = le64_to_cpu(h->seq);
2629
2630         mutex_lock(&mdsc->mutex);
2631         if (op == CEPH_SESSION_CLOSE)
2632                 __unregister_session(mdsc, session);
2633         /* FIXME: this ttl calculation is generous */
2634         session->s_ttl = jiffies + HZ*mdsc->mdsmap->m_session_autoclose;
2635         mutex_unlock(&mdsc->mutex);
2636
2637         mutex_lock(&session->s_mutex);
2638
2639         dout("handle_session mds%d %s %p state %s seq %llu\n",
2640              mds, ceph_session_op_name(op), session,
2641              ceph_session_state_name(session->s_state), seq);
2642
2643         if (session->s_state == CEPH_MDS_SESSION_HUNG) {
2644                 session->s_state = CEPH_MDS_SESSION_OPEN;
2645                 pr_info("mds%d came back\n", session->s_mds);
2646         }
2647
2648         switch (op) {
2649         case CEPH_SESSION_OPEN:
2650                 if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
2651                         pr_info("mds%d reconnect success\n", session->s_mds);
2652                 session->s_state = CEPH_MDS_SESSION_OPEN;
2653                 renewed_caps(mdsc, session, 0);
2654                 wake = 1;
2655                 if (mdsc->stopping)
2656                         __close_session(mdsc, session);
2657                 break;
2658
2659         case CEPH_SESSION_RENEWCAPS:
2660                 if (session->s_renew_seq == seq)
2661                         renewed_caps(mdsc, session, 1);
2662                 break;
2663
2664         case CEPH_SESSION_CLOSE:
2665                 if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
2666                         pr_info("mds%d reconnect denied\n", session->s_mds);
2667                 cleanup_session_requests(mdsc, session);
2668                 remove_session_caps(session);
2669                 wake = 2; /* for good measure */
2670                 wake_up_all(&mdsc->session_close_wq);
2671                 break;
2672
2673         case CEPH_SESSION_STALE:
2674                 pr_info("mds%d caps went stale, renewing\n",
2675                         session->s_mds);
2676                 spin_lock(&session->s_gen_ttl_lock);
2677                 session->s_cap_gen++;
2678                 session->s_cap_ttl = jiffies - 1;
2679                 spin_unlock(&session->s_gen_ttl_lock);
2680                 send_renew_caps(mdsc, session);
2681                 break;
2682
2683         case CEPH_SESSION_RECALL_STATE:
2684                 trim_caps(mdsc, session, le32_to_cpu(h->max_caps));
2685                 break;
2686
2687         case CEPH_SESSION_FLUSHMSG:
2688                 send_flushmsg_ack(mdsc, session, seq);
2689                 break;
2690
2691         case CEPH_SESSION_FORCE_RO:
2692                 dout("force_session_readonly %p\n", session);
2693                 spin_lock(&session->s_cap_lock);
2694                 session->s_readonly = true;
2695                 spin_unlock(&session->s_cap_lock);
2696                 wake_up_session_caps(session, 0);
2697                 break;
2698
2699         default:
2700                 pr_err("mdsc_handle_session bad op %d mds%d\n", op, mds);
2701                 WARN_ON(1);
2702         }
2703
2704         mutex_unlock(&session->s_mutex);
2705         if (wake) {
2706                 mutex_lock(&mdsc->mutex);
2707                 __wake_requests(mdsc, &session->s_waiting);
2708                 if (wake == 2)
2709                         kick_requests(mdsc, mds);
2710                 mutex_unlock(&mdsc->mutex);
2711         }
2712         return;
2713
2714 bad:
2715         pr_err("mdsc_handle_session corrupt message mds%d len %d\n", mds,
2716                (int)msg->front.iov_len);
2717         ceph_msg_dump(msg);
2718         return;
2719 }
2720
2721
2722 /*
2723  * called under session->mutex.
2724  */
2725 static void replay_unsafe_requests(struct ceph_mds_client *mdsc,
2726                                    struct ceph_mds_session *session)
2727 {
2728         struct ceph_mds_request *req, *nreq;
2729         struct rb_node *p;
2730         int err;
2731
2732         dout("replay_unsafe_requests mds%d\n", session->s_mds);
2733
2734         mutex_lock(&mdsc->mutex);
2735         list_for_each_entry_safe(req, nreq, &session->s_unsafe, r_unsafe_item) {
2736                 err = __prepare_send_request(mdsc, req, session->s_mds, true);
2737                 if (!err) {
2738                         ceph_msg_get(req->r_request);
2739                         ceph_con_send(&session->s_con, req->r_request);
2740                 }
2741         }
2742
2743         /*
2744          * also re-send old requests when MDS enters reconnect stage. So that MDS
2745          * can process completed request in clientreplay stage.
2746          */
2747         p = rb_first(&mdsc->request_tree);
2748         while (p) {
2749                 req = rb_entry(p, struct ceph_mds_request, r_node);
2750                 p = rb_next(p);
2751                 if (req->r_got_unsafe)
2752                         continue;
2753                 if (req->r_attempts == 0)
2754                         continue; /* only old requests */
2755                 if (req->r_session &&
2756                     req->r_session->s_mds == session->s_mds) {
2757                         err = __prepare_send_request(mdsc, req,
2758                                                      session->s_mds, true);
2759                         if (!err) {
2760                                 ceph_msg_get(req->r_request);
2761                                 ceph_con_send(&session->s_con, req->r_request);
2762                         }
2763                 }
2764         }
2765         mutex_unlock(&mdsc->mutex);
2766 }
2767
2768 /*
2769  * Encode information about a cap for a reconnect with the MDS.
2770  */
2771 static int encode_caps_cb(struct inode *inode, struct ceph_cap *cap,
2772                           void *arg)
2773 {
2774         union {
2775                 struct ceph_mds_cap_reconnect v2;
2776                 struct ceph_mds_cap_reconnect_v1 v1;
2777         } rec;
2778         size_t reclen;
2779         struct ceph_inode_info *ci;
2780         struct ceph_reconnect_state *recon_state = arg;
2781         struct ceph_pagelist *pagelist = recon_state->pagelist;
2782         char *path;
2783         int pathlen, err;
2784         u64 pathbase;
2785         struct dentry *dentry;
2786
2787         ci = cap->ci;
2788
2789         dout(" adding %p ino %llx.%llx cap %p %lld %s\n",
2790              inode, ceph_vinop(inode), cap, cap->cap_id,
2791              ceph_cap_string(cap->issued));
2792         err = ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
2793         if (err)
2794                 return err;
2795
2796         dentry = d_find_alias(inode);
2797         if (dentry) {
2798                 path = ceph_mdsc_build_path(dentry, &pathlen, &pathbase, 0);
2799                 if (IS_ERR(path)) {
2800                         err = PTR_ERR(path);
2801                         goto out_dput;
2802                 }
2803         } else {
2804                 path = NULL;
2805                 pathlen = 0;
2806         }
2807         err = ceph_pagelist_encode_string(pagelist, path, pathlen);
2808         if (err)
2809                 goto out_free;
2810
2811         spin_lock(&ci->i_ceph_lock);
2812         cap->seq = 0;        /* reset cap seq */
2813         cap->issue_seq = 0;  /* and issue_seq */
2814         cap->mseq = 0;       /* and migrate_seq */
2815         cap->cap_gen = cap->session->s_cap_gen;
2816
2817         if (recon_state->flock) {
2818                 rec.v2.cap_id = cpu_to_le64(cap->cap_id);
2819                 rec.v2.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
2820                 rec.v2.issued = cpu_to_le32(cap->issued);
2821                 rec.v2.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
2822                 rec.v2.pathbase = cpu_to_le64(pathbase);
2823                 rec.v2.flock_len = 0;
2824                 reclen = sizeof(rec.v2);
2825         } else {
2826                 rec.v1.cap_id = cpu_to_le64(cap->cap_id);
2827                 rec.v1.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
2828                 rec.v1.issued = cpu_to_le32(cap->issued);
2829                 rec.v1.size = cpu_to_le64(inode->i_size);
2830                 ceph_encode_timespec(&rec.v1.mtime, &inode->i_mtime);
2831                 ceph_encode_timespec(&rec.v1.atime, &inode->i_atime);
2832                 rec.v1.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
2833                 rec.v1.pathbase = cpu_to_le64(pathbase);
2834                 reclen = sizeof(rec.v1);
2835         }
2836         spin_unlock(&ci->i_ceph_lock);
2837
2838         if (recon_state->flock) {
2839                 int num_fcntl_locks, num_flock_locks;
2840                 struct ceph_filelock *flocks;
2841
2842 encode_again:
2843                 ceph_count_locks(inode, &num_fcntl_locks, &num_flock_locks);
2844                 flocks = kmalloc((num_fcntl_locks+num_flock_locks) *
2845                                  sizeof(struct ceph_filelock), GFP_NOFS);
2846                 if (!flocks) {
2847                         err = -ENOMEM;
2848                         goto out_free;
2849                 }
2850                 err = ceph_encode_locks_to_buffer(inode, flocks,
2851                                                   num_fcntl_locks,
2852                                                   num_flock_locks);
2853                 if (err) {
2854                         kfree(flocks);
2855                         if (err == -ENOSPC)
2856                                 goto encode_again;
2857                         goto out_free;
2858                 }
2859                 /*
2860                  * number of encoded locks is stable, so copy to pagelist
2861                  */
2862                 rec.v2.flock_len = cpu_to_le32(2*sizeof(u32) +
2863                                     (num_fcntl_locks+num_flock_locks) *
2864                                     sizeof(struct ceph_filelock));
2865                 err = ceph_pagelist_append(pagelist, &rec, reclen);
2866                 if (!err)
2867                         err = ceph_locks_to_pagelist(flocks, pagelist,
2868                                                      num_fcntl_locks,
2869                                                      num_flock_locks);
2870                 kfree(flocks);
2871         } else {
2872                 err = ceph_pagelist_append(pagelist, &rec, reclen);
2873         }
2874
2875         recon_state->nr_caps++;
2876 out_free:
2877         kfree(path);
2878 out_dput:
2879         dput(dentry);
2880         return err;
2881 }
2882
2883
2884 /*
2885  * If an MDS fails and recovers, clients need to reconnect in order to
2886  * reestablish shared state.  This includes all caps issued through
2887  * this session _and_ the snap_realm hierarchy.  Because it's not
2888  * clear which snap realms the mds cares about, we send everything we
2889  * know about.. that ensures we'll then get any new info the
2890  * recovering MDS might have.
2891  *
2892  * This is a relatively heavyweight operation, but it's rare.
2893  *
2894  * called with mdsc->mutex held.
2895  */
2896 static void send_mds_reconnect(struct ceph_mds_client *mdsc,
2897                                struct ceph_mds_session *session)
2898 {
2899         struct ceph_msg *reply;
2900         struct rb_node *p;
2901         int mds = session->s_mds;
2902         int err = -ENOMEM;
2903         int s_nr_caps;
2904         struct ceph_pagelist *pagelist;
2905         struct ceph_reconnect_state recon_state;
2906
2907         pr_info("mds%d reconnect start\n", mds);
2908
2909         pagelist = kmalloc(sizeof(*pagelist), GFP_NOFS);
2910         if (!pagelist)
2911                 goto fail_nopagelist;
2912         ceph_pagelist_init(pagelist);
2913
2914         reply = ceph_msg_new(CEPH_MSG_CLIENT_RECONNECT, 0, GFP_NOFS, false);
2915         if (!reply)
2916                 goto fail_nomsg;
2917
2918         mutex_lock(&session->s_mutex);
2919         session->s_state = CEPH_MDS_SESSION_RECONNECTING;
2920         session->s_seq = 0;
2921
2922         dout("session %p state %s\n", session,
2923              ceph_session_state_name(session->s_state));
2924
2925         spin_lock(&session->s_gen_ttl_lock);
2926         session->s_cap_gen++;
2927         spin_unlock(&session->s_gen_ttl_lock);
2928
2929         spin_lock(&session->s_cap_lock);
2930         /* don't know if session is readonly */
2931         session->s_readonly = 0;
2932         /*
2933          * notify __ceph_remove_cap() that we are composing cap reconnect.
2934          * If a cap get released before being added to the cap reconnect,
2935          * __ceph_remove_cap() should skip queuing cap release.
2936          */
2937         session->s_cap_reconnect = 1;
2938         /* drop old cap expires; we're about to reestablish that state */
2939         cleanup_cap_releases(mdsc, session);
2940
2941         /* trim unused caps to reduce MDS's cache rejoin time */
2942         if (mdsc->fsc->sb->s_root)
2943                 shrink_dcache_parent(mdsc->fsc->sb->s_root);
2944
2945         ceph_con_close(&session->s_con);
2946         ceph_con_open(&session->s_con,
2947                       CEPH_ENTITY_TYPE_MDS, mds,
2948                       ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
2949
2950         /* replay unsafe requests */
2951         replay_unsafe_requests(mdsc, session);
2952
2953         down_read(&mdsc->snap_rwsem);
2954
2955         /* traverse this session's caps */
2956         s_nr_caps = session->s_nr_caps;
2957         err = ceph_pagelist_encode_32(pagelist, s_nr_caps);
2958         if (err)
2959                 goto fail;
2960
2961         recon_state.nr_caps = 0;
2962         recon_state.pagelist = pagelist;
2963         recon_state.flock = session->s_con.peer_features & CEPH_FEATURE_FLOCK;
2964         err = iterate_session_caps(session, encode_caps_cb, &recon_state);
2965         if (err < 0)
2966                 goto fail;
2967
2968         spin_lock(&session->s_cap_lock);
2969         session->s_cap_reconnect = 0;
2970         spin_unlock(&session->s_cap_lock);
2971
2972         /*
2973          * snaprealms.  we provide mds with the ino, seq (version), and
2974          * parent for all of our realms.  If the mds has any newer info,
2975          * it will tell us.
2976          */
2977         for (p = rb_first(&mdsc->snap_realms); p; p = rb_next(p)) {
2978                 struct ceph_snap_realm *realm =
2979                         rb_entry(p, struct ceph_snap_realm, node);
2980                 struct ceph_mds_snaprealm_reconnect sr_rec;
2981
2982                 dout(" adding snap realm %llx seq %lld parent %llx\n",
2983                      realm->ino, realm->seq, realm->parent_ino);
2984                 sr_rec.ino = cpu_to_le64(realm->ino);
2985                 sr_rec.seq = cpu_to_le64(realm->seq);
2986                 sr_rec.parent = cpu_to_le64(realm->parent_ino);
2987                 err = ceph_pagelist_append(pagelist, &sr_rec, sizeof(sr_rec));
2988                 if (err)
2989                         goto fail;
2990         }
2991
2992         if (recon_state.flock)
2993                 reply->hdr.version = cpu_to_le16(2);
2994
2995         /* raced with cap release? */
2996         if (s_nr_caps != recon_state.nr_caps) {
2997                 struct page *page = list_first_entry(&pagelist->head,
2998                                                      struct page, lru);
2999                 __le32 *addr = kmap_atomic(page);
3000                 *addr = cpu_to_le32(recon_state.nr_caps);
3001                 kunmap_atomic(addr);
3002         }
3003
3004         reply->hdr.data_len = cpu_to_le32(pagelist->length);
3005         ceph_msg_data_add_pagelist(reply, pagelist);
3006
3007         ceph_early_kick_flushing_caps(mdsc, session);
3008
3009         ceph_con_send(&session->s_con, reply);
3010
3011         mutex_unlock(&session->s_mutex);
3012
3013         mutex_lock(&mdsc->mutex);
3014         __wake_requests(mdsc, &session->s_waiting);
3015         mutex_unlock(&mdsc->mutex);
3016
3017         up_read(&mdsc->snap_rwsem);
3018         return;
3019
3020 fail:
3021         ceph_msg_put(reply);
3022         up_read(&mdsc->snap_rwsem);
3023         mutex_unlock(&session->s_mutex);
3024 fail_nomsg:
3025         ceph_pagelist_release(pagelist);
3026 fail_nopagelist:
3027         pr_err("error %d preparing reconnect for mds%d\n", err, mds);
3028         return;
3029 }
3030
3031
3032 /*
3033  * compare old and new mdsmaps, kicking requests
3034  * and closing out old connections as necessary
3035  *
3036  * called under mdsc->mutex.
3037  */
3038 static void check_new_map(struct ceph_mds_client *mdsc,
3039                           struct ceph_mdsmap *newmap,
3040                           struct ceph_mdsmap *oldmap)
3041 {
3042         int i;
3043         int oldstate, newstate;
3044         struct ceph_mds_session *s;
3045
3046         dout("check_new_map new %u old %u\n",
3047              newmap->m_epoch, oldmap->m_epoch);
3048
3049         for (i = 0; i < oldmap->m_max_mds && i < mdsc->max_sessions; i++) {
3050                 if (mdsc->sessions[i] == NULL)
3051                         continue;
3052                 s = mdsc->sessions[i];
3053                 oldstate = ceph_mdsmap_get_state(oldmap, i);
3054                 newstate = ceph_mdsmap_get_state(newmap, i);
3055
3056                 dout("check_new_map mds%d state %s%s -> %s%s (session %s)\n",
3057                      i, ceph_mds_state_name(oldstate),
3058                      ceph_mdsmap_is_laggy(oldmap, i) ? " (laggy)" : "",
3059                      ceph_mds_state_name(newstate),
3060                      ceph_mdsmap_is_laggy(newmap, i) ? " (laggy)" : "",
3061                      ceph_session_state_name(s->s_state));
3062
3063                 if (i >= newmap->m_max_mds ||
3064                     memcmp(ceph_mdsmap_get_addr(oldmap, i),
3065                            ceph_mdsmap_get_addr(newmap, i),
3066                            sizeof(struct ceph_entity_addr))) {
3067                         if (s->s_state == CEPH_MDS_SESSION_OPENING) {
3068                                 /* the session never opened, just close it
3069                                  * out now */
3070                                 __wake_requests(mdsc, &s->s_waiting);
3071                                 __unregister_session(mdsc, s);
3072                         } else {
3073                                 /* just close it */
3074                                 mutex_unlock(&mdsc->mutex);
3075                                 mutex_lock(&s->s_mutex);
3076                                 mutex_lock(&mdsc->mutex);
3077                                 ceph_con_close(&s->s_con);
3078                                 mutex_unlock(&s->s_mutex);
3079                                 s->s_state = CEPH_MDS_SESSION_RESTARTING;
3080                         }
3081                 } else if (oldstate == newstate) {
3082                         continue;  /* nothing new with this mds */
3083                 }
3084
3085                 /*
3086                  * send reconnect?
3087                  */
3088                 if (s->s_state == CEPH_MDS_SESSION_RESTARTING &&
3089                     newstate >= CEPH_MDS_STATE_RECONNECT) {
3090                         mutex_unlock(&mdsc->mutex);
3091                         send_mds_reconnect(mdsc, s);
3092                         mutex_lock(&mdsc->mutex);
3093                 }
3094
3095                 /*
3096                  * kick request on any mds that has gone active.
3097                  */
3098                 if (oldstate < CEPH_MDS_STATE_ACTIVE &&
3099                     newstate >= CEPH_MDS_STATE_ACTIVE) {
3100                         if (oldstate != CEPH_MDS_STATE_CREATING &&
3101                             oldstate != CEPH_MDS_STATE_STARTING)
3102                                 pr_info("mds%d recovery completed\n", s->s_mds);
3103                         kick_requests(mdsc, i);
3104                         ceph_kick_flushing_caps(mdsc, s);
3105                         wake_up_session_caps(s, 1);
3106                 }
3107         }
3108
3109         for (i = 0; i < newmap->m_max_mds && i < mdsc->max_sessions; i++) {
3110                 s = mdsc->sessions[i];
3111                 if (!s)
3112                         continue;
3113                 if (!ceph_mdsmap_is_laggy(newmap, i))
3114                         continue;
3115                 if (s->s_state == CEPH_MDS_SESSION_OPEN ||
3116                     s->s_state == CEPH_MDS_SESSION_HUNG ||
3117                     s->s_state == CEPH_MDS_SESSION_CLOSING) {
3118                         dout(" connecting to export targets of laggy mds%d\n",
3119                              i);
3120                         __open_export_target_sessions(mdsc, s);
3121                 }
3122         }
3123 }
3124
3125
3126
3127 /*
3128  * leases
3129  */
3130
3131 /*
3132  * caller must hold session s_mutex, dentry->d_lock
3133  */
3134 void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry)
3135 {
3136         struct ceph_dentry_info *di = ceph_dentry(dentry);
3137
3138         ceph_put_mds_session(di->lease_session);
3139         di->lease_session = NULL;
3140 }
3141
3142 static void handle_lease(struct ceph_mds_client *mdsc,
3143                          struct ceph_mds_session *session,
3144                          struct ceph_msg *msg)
3145 {
3146         struct super_block *sb = mdsc->fsc->sb;
3147         struct inode *inode;
3148         struct dentry *parent, *dentry;
3149         struct ceph_dentry_info *di;
3150         int mds = session->s_mds;
3151         struct ceph_mds_lease *h = msg->front.iov_base;
3152         u32 seq;
3153         struct ceph_vino vino;
3154         struct qstr dname;
3155         int release = 0;
3156
3157         dout("handle_lease from mds%d\n", mds);
3158
3159         /* decode */
3160         if (msg->front.iov_len < sizeof(*h) + sizeof(u32))
3161                 goto bad;
3162         vino.ino = le64_to_cpu(h->ino);
3163         vino.snap = CEPH_NOSNAP;
3164         seq = le32_to_cpu(h->seq);
3165         dname.name = (void *)h + sizeof(*h) + sizeof(u32);
3166         dname.len = msg->front.iov_len - sizeof(*h) - sizeof(u32);
3167         if (dname.len != get_unaligned_le32(h+1))
3168                 goto bad;
3169
3170         /* lookup inode */
3171         inode = ceph_find_inode(sb, vino);
3172         dout("handle_lease %s, ino %llx %p %.*s\n",
3173              ceph_lease_op_name(h->action), vino.ino, inode,
3174              dname.len, dname.name);
3175
3176         mutex_lock(&session->s_mutex);
3177         session->s_seq++;
3178
3179         if (inode == NULL) {
3180                 dout("handle_lease no inode %llx\n", vino.ino);
3181                 goto release;
3182         }
3183
3184         /* dentry */
3185         parent = d_find_alias(inode);
3186         if (!parent) {
3187                 dout("no parent dentry on inode %p\n", inode);
3188                 WARN_ON(1);
3189                 goto release;  /* hrm... */
3190         }
3191         dname.hash = full_name_hash(dname.name, dname.len);
3192         dentry = d_lookup(parent, &dname);
3193         dput(parent);
3194         if (!dentry)
3195                 goto release;
3196
3197         spin_lock(&dentry->d_lock);
3198         di = ceph_dentry(dentry);
3199         switch (h->action) {
3200         case CEPH_MDS_LEASE_REVOKE:
3201                 if (di->lease_session == session) {
3202                         if (ceph_seq_cmp(di->lease_seq, seq) > 0)
3203                                 h->seq = cpu_to_le32(di->lease_seq);
3204                         __ceph_mdsc_drop_dentry_lease(dentry);
3205                 }
3206                 release = 1;
3207                 break;
3208
3209         case CEPH_MDS_LEASE_RENEW:
3210                 if (di->lease_session == session &&
3211                     di->lease_gen == session->s_cap_gen &&
3212                     di->lease_renew_from &&
3213                     di->lease_renew_after == 0) {
3214                         unsigned long duration =
3215                                 msecs_to_jiffies(le32_to_cpu(h->duration_ms));
3216
3217                         di->lease_seq = seq;
3218                         dentry->d_time = di->lease_renew_from + duration;
3219                         di->lease_renew_after = di->lease_renew_from +
3220                                 (duration >> 1);
3221                         di->lease_renew_from = 0;
3222                 }
3223                 break;
3224         }
3225         spin_unlock(&dentry->d_lock);
3226         dput(dentry);
3227
3228         if (!release)
3229                 goto out;
3230
3231 release:
3232         /* let's just reuse the same message */
3233         h->action = CEPH_MDS_LEASE_REVOKE_ACK;
3234         ceph_msg_get(msg);
3235         ceph_con_send(&session->s_con, msg);
3236
3237 out:
3238         iput(inode);
3239         mutex_unlock(&session->s_mutex);
3240         return;
3241
3242 bad:
3243         pr_err("corrupt lease message\n");
3244         ceph_msg_dump(msg);
3245 }
3246
3247 void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
3248                               struct inode *inode,
3249                               struct dentry *dentry, char action,
3250                               u32 seq)
3251 {
3252         struct ceph_msg *msg;
3253         struct ceph_mds_lease *lease;
3254         int len = sizeof(*lease) + sizeof(u32);
3255         int dnamelen = 0;
3256
3257         dout("lease_send_msg inode %p dentry %p %s to mds%d\n",
3258              inode, dentry, ceph_lease_op_name(action), session->s_mds);
3259         dnamelen = dentry->d_name.len;
3260         len += dnamelen;
3261
3262         msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS, false);
3263         if (!msg)
3264                 return;
3265         lease = msg->front.iov_base;
3266         lease->action = action;
3267         lease->ino = cpu_to_le64(ceph_vino(inode).ino);
3268         lease->first = lease->last = cpu_to_le64(ceph_vino(inode).snap);
3269         lease->seq = cpu_to_le32(seq);
3270         put_unaligned_le32(dnamelen, lease + 1);
3271         memcpy((void *)(lease + 1) + 4, dentry->d_name.name, dnamelen);
3272
3273         /*
3274          * if this is a preemptive lease RELEASE, no need to
3275          * flush request stream, since the actual request will
3276          * soon follow.
3277          */
3278         msg->more_to_follow = (action == CEPH_MDS_LEASE_RELEASE);
3279
3280         ceph_con_send(&session->s_con, msg);
3281 }
3282
3283 /*
3284  * Preemptively release a lease we expect to invalidate anyway.
3285  * Pass @inode always, @dentry is optional.
3286  */
3287 void ceph_mdsc_lease_release(struct ceph_mds_client *mdsc, struct inode *inode,
3288                              struct dentry *dentry)
3289 {
3290         struct ceph_dentry_info *di;
3291         struct ceph_mds_session *session;
3292         u32 seq;
3293
3294         BUG_ON(inode == NULL);
3295         BUG_ON(dentry == NULL);
3296
3297         /* is dentry lease valid? */
3298         spin_lock(&dentry->d_lock);
3299         di = ceph_dentry(dentry);
3300         if (!di || !di->lease_session ||
3301             di->lease_session->s_mds < 0 ||
3302             di->lease_gen != di->lease_session->s_cap_gen ||
3303             !time_before(jiffies, dentry->d_time)) {
3304                 dout("lease_release inode %p dentry %p -- "
3305                      "no lease\n",
3306                      inode, dentry);
3307                 spin_unlock(&dentry->d_lock);
3308                 return;
3309         }
3310
3311         /* we do have a lease on this dentry; note mds and seq */
3312         session = ceph_get_mds_session(di->lease_session);
3313         seq = di->lease_seq;
3314         __ceph_mdsc_drop_dentry_lease(dentry);
3315         spin_unlock(&dentry->d_lock);
3316
3317         dout("lease_release inode %p dentry %p to mds%d\n",
3318              inode, dentry, session->s_mds);
3319         ceph_mdsc_lease_send_msg(session, inode, dentry,
3320                                  CEPH_MDS_LEASE_RELEASE, seq);
3321         ceph_put_mds_session(session);
3322 }
3323
3324 /*
3325  * drop all leases (and dentry refs) in preparation for umount
3326  */
3327 static void drop_leases(struct ceph_mds_client *mdsc)
3328 {
3329         int i;
3330
3331         dout("drop_leases\n");
3332         mutex_lock(&mdsc->mutex);
3333         for (i = 0; i < mdsc->max_sessions; i++) {
3334                 struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
3335                 if (!s)
3336                         continue;
3337                 mutex_unlock(&mdsc->mutex);
3338                 mutex_lock(&s->s_mutex);
3339                 mutex_unlock(&s->s_mutex);
3340                 ceph_put_mds_session(s);
3341                 mutex_lock(&mdsc->mutex);
3342         }
3343         mutex_unlock(&mdsc->mutex);
3344 }
3345
3346
3347
3348 /*
3349  * delayed work -- periodically trim expired leases, renew caps with mds
3350  */
3351 static void schedule_delayed(struct ceph_mds_client *mdsc)
3352 {
3353         int delay = 5;
3354         unsigned hz = round_jiffies_relative(HZ * delay);
3355         schedule_delayed_work(&mdsc->delayed_work, hz);
3356 }
3357
3358 static void delayed_work(struct work_struct *work)
3359 {
3360         int i;
3361         struct ceph_mds_client *mdsc =
3362                 container_of(work, struct ceph_mds_client, delayed_work.work);
3363         int renew_interval;
3364         int renew_caps;
3365
3366         dout("mdsc delayed_work\n");
3367         ceph_check_delayed_caps(mdsc);
3368
3369         mutex_lock(&mdsc->mutex);
3370         renew_interval = mdsc->mdsmap->m_session_timeout >> 2;
3371         renew_caps = time_after_eq(jiffies, HZ*renew_interval +
3372                                    mdsc->last_renew_caps);
3373         if (renew_caps)
3374                 mdsc->last_renew_caps = jiffies;
3375
3376         for (i = 0; i < mdsc->max_sessions; i++) {
3377                 struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
3378                 if (s == NULL)
3379                         continue;
3380                 if (s->s_state == CEPH_MDS_SESSION_CLOSING) {
3381                         dout("resending session close request for mds%d\n",
3382                              s->s_mds);
3383                         request_close_session(mdsc, s);
3384                         ceph_put_mds_session(s);
3385                         continue;
3386                 }
3387                 if (s->s_ttl && time_after(jiffies, s->s_ttl)) {
3388                         if (s->s_state == CEPH_MDS_SESSION_OPEN) {
3389                                 s->s_state = CEPH_MDS_SESSION_HUNG;
3390                                 pr_info("mds%d hung\n", s->s_mds);
3391                         }
3392                 }
3393                 if (s->s_state < CEPH_MDS_SESSION_OPEN) {
3394                         /* this mds is failed or recovering, just wait */
3395                         ceph_put_mds_session(s);
3396                         continue;
3397                 }
3398                 mutex_unlock(&mdsc->mutex);
3399
3400                 mutex_lock(&s->s_mutex);
3401                 if (renew_caps)
3402                         send_renew_caps(mdsc, s);
3403                 else
3404                         ceph_con_keepalive(&s->s_con);
3405                 if (s->s_state == CEPH_MDS_SESSION_OPEN ||
3406                     s->s_state == CEPH_MDS_SESSION_HUNG)
3407                         ceph_send_cap_releases(mdsc, s);
3408                 mutex_unlock(&s->s_mutex);
3409                 ceph_put_mds_session(s);
3410
3411                 mutex_lock(&mdsc->mutex);
3412         }
3413         mutex_unlock(&mdsc->mutex);
3414
3415         schedule_delayed(mdsc);
3416 }
3417
3418 int ceph_mdsc_init(struct ceph_fs_client *fsc)
3419
3420 {
3421         struct ceph_mds_client *mdsc;
3422
3423         mdsc = kzalloc(sizeof(struct ceph_mds_client), GFP_NOFS);
3424         if (!mdsc)
3425                 return -ENOMEM;
3426         mdsc->fsc = fsc;
3427         fsc->mdsc = mdsc;
3428         mutex_init(&mdsc->mutex);
3429         mdsc->mdsmap = kzalloc(sizeof(*mdsc->mdsmap), GFP_NOFS);
3430         if (mdsc->mdsmap == NULL) {
3431                 kfree(mdsc);
3432                 return -ENOMEM;
3433         }
3434
3435         init_completion(&mdsc->safe_umount_waiters);
3436         init_waitqueue_head(&mdsc->session_close_wq);
3437         INIT_LIST_HEAD(&mdsc->waiting_for_map);
3438         mdsc->sessions = NULL;
3439         atomic_set(&mdsc->num_sessions, 0);
3440         mdsc->max_sessions = 0;
3441         mdsc->stopping = 0;
3442         mdsc->last_snap_seq = 0;
3443         init_rwsem(&mdsc->snap_rwsem);
3444         mdsc->snap_realms = RB_ROOT;
3445         INIT_LIST_HEAD(&mdsc->snap_empty);
3446         spin_lock_init(&mdsc->snap_empty_lock);
3447         mdsc->last_tid = 0;
3448         mdsc->oldest_tid = 0;
3449         mdsc->request_tree = RB_ROOT;
3450         INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
3451         mdsc->last_renew_caps = jiffies;
3452         INIT_LIST_HEAD(&mdsc->cap_delay_list);
3453         spin_lock_init(&mdsc->cap_delay_lock);
3454         INIT_LIST_HEAD(&mdsc->snap_flush_list);
3455         spin_lock_init(&mdsc->snap_flush_lock);
3456         mdsc->last_cap_flush_tid = 1;
3457         mdsc->cap_flush_tree = RB_ROOT;
3458         INIT_LIST_HEAD(&mdsc->cap_dirty);
3459         INIT_LIST_HEAD(&mdsc->cap_dirty_migrating);
3460         mdsc->num_cap_flushing = 0;
3461         spin_lock_init(&mdsc->cap_dirty_lock);
3462         init_waitqueue_head(&mdsc->cap_flushing_wq);
3463         spin_lock_init(&mdsc->dentry_lru_lock);
3464         INIT_LIST_HEAD(&mdsc->dentry_lru);
3465
3466         ceph_caps_init(mdsc);
3467         ceph_adjust_min_caps(mdsc, fsc->min_caps);
3468
3469         init_rwsem(&mdsc->pool_perm_rwsem);
3470         mdsc->pool_perm_tree = RB_ROOT;
3471
3472         return 0;
3473 }
3474
3475 /*
3476  * Wait for safe replies on open mds requests.  If we time out, drop
3477  * all requests from the tree to avoid dangling dentry refs.
3478  */
3479 static void wait_requests(struct ceph_mds_client *mdsc)
3480 {
3481         struct ceph_options *opts = mdsc->fsc->client->options;
3482         struct ceph_mds_request *req;
3483
3484         mutex_lock(&mdsc->mutex);
3485         if (__get_oldest_req(mdsc)) {
3486                 mutex_unlock(&mdsc->mutex);
3487
3488                 dout("wait_requests waiting for requests\n");
3489                 wait_for_completion_timeout(&mdsc->safe_umount_waiters,
3490                                     ceph_timeout_jiffies(opts->mount_timeout));
3491
3492                 /* tear down remaining requests */
3493                 mutex_lock(&mdsc->mutex);
3494                 while ((req = __get_oldest_req(mdsc))) {
3495                         dout("wait_requests timed out on tid %llu\n",
3496                              req->r_tid);
3497                         __unregister_request(mdsc, req);
3498                 }
3499         }
3500         mutex_unlock(&mdsc->mutex);
3501         dout("wait_requests done\n");
3502 }
3503
3504 /*
3505  * called before mount is ro, and before dentries are torn down.
3506  * (hmm, does this still race with new lookups?)
3507  */
3508 void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
3509 {
3510         dout("pre_umount\n");
3511         mdsc->stopping = 1;
3512
3513         drop_leases(mdsc);
3514         ceph_flush_dirty_caps(mdsc);
3515         wait_requests(mdsc);
3516
3517         /*
3518          * wait for reply handlers to drop their request refs and
3519          * their inode/dcache refs
3520          */
3521         ceph_msgr_flush();
3522 }
3523
3524 /*
3525  * wait for all write mds requests to flush.
3526  */
3527 static void wait_unsafe_requests(struct ceph_mds_client *mdsc, u64 want_tid)
3528 {
3529         struct ceph_mds_request *req = NULL, *nextreq;
3530         struct rb_node *n;
3531
3532         mutex_lock(&mdsc->mutex);
3533         dout("wait_unsafe_requests want %lld\n", want_tid);
3534 restart:
3535         req = __get_oldest_req(mdsc);
3536         while (req && req->r_tid <= want_tid) {
3537                 /* find next request */
3538                 n = rb_next(&req->r_node);
3539                 if (n)
3540                         nextreq = rb_entry(n, struct ceph_mds_request, r_node);
3541                 else
3542                         nextreq = NULL;
3543                 if (req->r_op != CEPH_MDS_OP_SETFILELOCK &&
3544                     (req->r_op & CEPH_MDS_OP_WRITE)) {
3545                         /* write op */
3546                         ceph_mdsc_get_request(req);
3547                         if (nextreq)
3548                                 ceph_mdsc_get_request(nextreq);
3549                         mutex_unlock(&mdsc->mutex);
3550                         dout("wait_unsafe_requests  wait on %llu (want %llu)\n",
3551                              req->r_tid, want_tid);
3552                         wait_for_completion(&req->r_safe_completion);
3553                         mutex_lock(&mdsc->mutex);
3554                         ceph_mdsc_put_request(req);
3555                         if (!nextreq)
3556                                 break;  /* next dne before, so we're done! */
3557                         if (RB_EMPTY_NODE(&nextreq->r_node)) {
3558                                 /* next request was removed from tree */
3559                                 ceph_mdsc_put_request(nextreq);
3560                                 goto restart;
3561                         }
3562                         ceph_mdsc_put_request(nextreq);  /* won't go away */
3563                 }
3564                 req = nextreq;
3565         }
3566         mutex_unlock(&mdsc->mutex);
3567         dout("wait_unsafe_requests done\n");
3568 }
3569
3570 void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
3571 {
3572         u64 want_tid, want_flush, want_snap;
3573
3574         if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
3575                 return;
3576
3577         dout("sync\n");
3578         mutex_lock(&mdsc->mutex);
3579         want_tid = mdsc->last_tid;
3580         mutex_unlock(&mdsc->mutex);
3581
3582         ceph_flush_dirty_caps(mdsc);
3583         spin_lock(&mdsc->cap_dirty_lock);
3584         want_flush = mdsc->last_cap_flush_tid;
3585         spin_unlock(&mdsc->cap_dirty_lock);
3586
3587         down_read(&mdsc->snap_rwsem);
3588         want_snap = mdsc->last_snap_seq;
3589         up_read(&mdsc->snap_rwsem);
3590
3591         dout("sync want tid %lld flush_seq %lld snap_seq %lld\n",
3592              want_tid, want_flush, want_snap);
3593
3594         wait_unsafe_requests(mdsc, want_tid);
3595         wait_caps_flush(mdsc, want_flush, want_snap);
3596 }
3597
3598 /*
3599  * true if all sessions are closed, or we force unmount
3600  */
3601 static bool done_closing_sessions(struct ceph_mds_client *mdsc)
3602 {
3603         if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
3604                 return true;
3605         return atomic_read(&mdsc->num_sessions) == 0;
3606 }
3607
3608 /*
3609  * called after sb is ro.
3610  */
3611 void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
3612 {
3613         struct ceph_options *opts = mdsc->fsc->client->options;
3614         struct ceph_mds_session *session;
3615         int i;
3616
3617         dout("close_sessions\n");
3618
3619         /* close sessions */
3620         mutex_lock(&mdsc->mutex);
3621         for (i = 0; i < mdsc->max_sessions; i++) {
3622                 session = __ceph_lookup_mds_session(mdsc, i);
3623                 if (!session)
3624                         continue;
3625                 mutex_unlock(&mdsc->mutex);
3626                 mutex_lock(&session->s_mutex);
3627                 __close_session(mdsc, session);
3628                 mutex_unlock(&session->s_mutex);
3629                 ceph_put_mds_session(session);
3630                 mutex_lock(&mdsc->mutex);
3631         }
3632         mutex_unlock(&mdsc->mutex);
3633
3634         dout("waiting for sessions to close\n");
3635         wait_event_timeout(mdsc->session_close_wq, done_closing_sessions(mdsc),
3636                            ceph_timeout_jiffies(opts->mount_timeout));
3637
3638         /* tear down remaining sessions */
3639         mutex_lock(&mdsc->mutex);
3640         for (i = 0; i < mdsc->max_sessions; i++) {
3641                 if (mdsc->sessions[i]) {
3642                         session = get_session(mdsc->sessions[i]);
3643                         __unregister_session(mdsc, session);
3644                         mutex_unlock(&mdsc->mutex);
3645                         mutex_lock(&session->s_mutex);
3646                         remove_session_caps(session);
3647                         mutex_unlock(&session->s_mutex);
3648                         ceph_put_mds_session(session);
3649                         mutex_lock(&mdsc->mutex);
3650                 }
3651         }
3652         WARN_ON(!list_empty(&mdsc->cap_delay_list));
3653         mutex_unlock(&mdsc->mutex);
3654
3655         ceph_cleanup_empty_realms(mdsc);
3656
3657         cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
3658
3659         dout("stopped\n");
3660 }
3661
3662 void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
3663 {
3664         struct ceph_mds_session *session;
3665         int mds;
3666
3667         dout("force umount\n");
3668
3669         mutex_lock(&mdsc->mutex);
3670         for (mds = 0; mds < mdsc->max_sessions; mds++) {
3671                 session = __ceph_lookup_mds_session(mdsc, mds);
3672                 if (!session)
3673                         continue;
3674                 mutex_unlock(&mdsc->mutex);
3675                 mutex_lock(&session->s_mutex);
3676                 __close_session(mdsc, session);
3677                 if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
3678                         cleanup_session_requests(mdsc, session);
3679                         remove_session_caps(session);
3680                 }
3681                 mutex_unlock(&session->s_mutex);
3682                 ceph_put_mds_session(session);
3683                 mutex_lock(&mdsc->mutex);
3684                 kick_requests(mdsc, mds);
3685         }
3686         __wake_requests(mdsc, &mdsc->waiting_for_map);
3687         mutex_unlock(&mdsc->mutex);
3688 }
3689
3690 static void ceph_mdsc_stop(struct ceph_mds_client *mdsc)
3691 {
3692         dout("stop\n");
3693         cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
3694         if (mdsc->mdsmap)
3695                 ceph_mdsmap_destroy(mdsc->mdsmap);
3696         kfree(mdsc->sessions);
3697         ceph_caps_finalize(mdsc);
3698         ceph_pool_perm_destroy(mdsc);
3699 }
3700
3701 void ceph_mdsc_destroy(struct ceph_fs_client *fsc)
3702 {
3703         struct ceph_mds_client *mdsc = fsc->mdsc;
3704
3705         dout("mdsc_destroy %p\n", mdsc);
3706         ceph_mdsc_stop(mdsc);
3707
3708         /* flush out any connection work with references to us */
3709         ceph_msgr_flush();
3710
3711         fsc->mdsc = NULL;
3712         kfree(mdsc);
3713         dout("mdsc_destroy %p done\n", mdsc);
3714 }
3715
3716
3717 /*
3718  * handle mds map update.
3719  */
3720 void ceph_mdsc_handle_map(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
3721 {
3722         u32 epoch;
3723         u32 maplen;
3724         void *p = msg->front.iov_base;
3725         void *end = p + msg->front.iov_len;
3726         struct ceph_mdsmap *newmap, *oldmap;
3727         struct ceph_fsid fsid;
3728         int err = -EINVAL;
3729
3730         ceph_decode_need(&p, end, sizeof(fsid)+2*sizeof(u32), bad);
3731         ceph_decode_copy(&p, &fsid, sizeof(fsid));
3732         if (ceph_check_fsid(mdsc->fsc->client, &fsid) < 0)
3733                 return;
3734         epoch = ceph_decode_32(&p);
3735         maplen = ceph_decode_32(&p);
3736         dout("handle_map epoch %u len %d\n", epoch, (int)maplen);
3737
3738         /* do we need it? */
3739         mutex_lock(&mdsc->mutex);
3740         if (mdsc->mdsmap && epoch <= mdsc->mdsmap->m_epoch) {
3741                 dout("handle_map epoch %u <= our %u\n",
3742                      epoch, mdsc->mdsmap->m_epoch);
3743                 mutex_unlock(&mdsc->mutex);
3744                 return;
3745         }
3746
3747         newmap = ceph_mdsmap_decode(&p, end);
3748         if (IS_ERR(newmap)) {
3749                 err = PTR_ERR(newmap);
3750                 goto bad_unlock;
3751         }
3752
3753         /* swap into place */
3754         if (mdsc->mdsmap) {
3755                 oldmap = mdsc->mdsmap;
3756                 mdsc->mdsmap = newmap;
3757                 check_new_map(mdsc, newmap, oldmap);
3758                 ceph_mdsmap_destroy(oldmap);
3759         } else {
3760                 mdsc->mdsmap = newmap;  /* first mds map */
3761         }
3762         mdsc->fsc->sb->s_maxbytes = mdsc->mdsmap->m_max_file_size;
3763
3764         __wake_requests(mdsc, &mdsc->waiting_for_map);
3765         ceph_monc_got_map(&mdsc->fsc->client->monc, CEPH_SUB_MDSMAP,
3766                           mdsc->mdsmap->m_epoch);
3767
3768         mutex_unlock(&mdsc->mutex);
3769         schedule_delayed(mdsc);
3770         return;
3771
3772 bad_unlock:
3773         mutex_unlock(&mdsc->mutex);
3774 bad:
3775         pr_err("error decoding mdsmap %d\n", err);
3776         return;
3777 }
3778
3779 static struct ceph_connection *con_get(struct ceph_connection *con)
3780 {
3781         struct ceph_mds_session *s = con->private;
3782
3783         if (get_session(s)) {
3784                 dout("mdsc con_get %p ok (%d)\n", s, atomic_read(&s->s_ref));
3785                 return con;
3786         }
3787         dout("mdsc con_get %p FAIL\n", s);
3788         return NULL;
3789 }
3790
3791 static void con_put(struct ceph_connection *con)
3792 {
3793         struct ceph_mds_session *s = con->private;
3794
3795         dout("mdsc con_put %p (%d)\n", s, atomic_read(&s->s_ref) - 1);
3796         ceph_put_mds_session(s);
3797 }
3798
3799 /*
3800  * if the client is unresponsive for long enough, the mds will kill
3801  * the session entirely.
3802  */
3803 static void peer_reset(struct ceph_connection *con)
3804 {
3805         struct ceph_mds_session *s = con->private;
3806         struct ceph_mds_client *mdsc = s->s_mdsc;
3807
3808         pr_warn("mds%d closed our session\n", s->s_mds);
3809         send_mds_reconnect(mdsc, s);
3810 }
3811
3812 static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
3813 {
3814         struct ceph_mds_session *s = con->private;
3815         struct ceph_mds_client *mdsc = s->s_mdsc;
3816         int type = le16_to_cpu(msg->hdr.type);
3817
3818         mutex_lock(&mdsc->mutex);
3819         if (__verify_registered_session(mdsc, s) < 0) {
3820                 mutex_unlock(&mdsc->mutex);
3821                 goto out;
3822         }
3823         mutex_unlock(&mdsc->mutex);
3824
3825         switch (type) {
3826         case CEPH_MSG_MDS_MAP:
3827                 ceph_mdsc_handle_map(mdsc, msg);
3828                 break;
3829         case CEPH_MSG_CLIENT_SESSION:
3830                 handle_session(s, msg);
3831                 break;
3832         case CEPH_MSG_CLIENT_REPLY:
3833                 handle_reply(s, msg);
3834                 break;
3835         case CEPH_MSG_CLIENT_REQUEST_FORWARD:
3836                 handle_forward(mdsc, s, msg);
3837                 break;
3838         case CEPH_MSG_CLIENT_CAPS:
3839                 ceph_handle_caps(s, msg);
3840                 break;
3841         case CEPH_MSG_CLIENT_SNAP:
3842                 ceph_handle_snap(mdsc, s, msg);
3843                 break;
3844         case CEPH_MSG_CLIENT_LEASE:
3845                 handle_lease(mdsc, s, msg);
3846                 break;
3847
3848         default:
3849                 pr_err("received unknown message type %d %s\n", type,
3850                        ceph_msg_type_name(type));
3851         }
3852 out:
3853         ceph_msg_put(msg);
3854 }
3855
3856 /*
3857  * authentication
3858  */
3859
3860 /*
3861  * Note: returned pointer is the address of a structure that's
3862  * managed separately.  Caller must *not* attempt to free it.
3863  */
3864 static struct ceph_auth_handshake *get_authorizer(struct ceph_connection *con,
3865                                         int *proto, int force_new)
3866 {
3867         struct ceph_mds_session *s = con->private;
3868         struct ceph_mds_client *mdsc = s->s_mdsc;
3869         struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
3870         struct ceph_auth_handshake *auth = &s->s_auth;
3871
3872         if (force_new && auth->authorizer) {
3873                 ceph_auth_destroy_authorizer(auth->authorizer);
3874                 auth->authorizer = NULL;
3875         }
3876         if (!auth->authorizer) {
3877                 int ret = ceph_auth_create_authorizer(ac, CEPH_ENTITY_TYPE_MDS,
3878                                                       auth);
3879                 if (ret)
3880                         return ERR_PTR(ret);
3881         } else {
3882                 int ret = ceph_auth_update_authorizer(ac, CEPH_ENTITY_TYPE_MDS,
3883                                                       auth);
3884                 if (ret)
3885                         return ERR_PTR(ret);
3886         }
3887         *proto = ac->protocol;
3888
3889         return auth;
3890 }
3891
3892
3893 static int verify_authorizer_reply(struct ceph_connection *con, int len)
3894 {
3895         struct ceph_mds_session *s = con->private;
3896         struct ceph_mds_client *mdsc = s->s_mdsc;
3897         struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
3898
3899         return ceph_auth_verify_authorizer_reply(ac, s->s_auth.authorizer, len);
3900 }
3901
3902 static int invalidate_authorizer(struct ceph_connection *con)
3903 {
3904         struct ceph_mds_session *s = con->private;
3905         struct ceph_mds_client *mdsc = s->s_mdsc;
3906         struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
3907
3908         ceph_auth_invalidate_authorizer(ac, CEPH_ENTITY_TYPE_MDS);
3909
3910         return ceph_monc_validate_auth(&mdsc->fsc->client->monc);
3911 }
3912
3913 static struct ceph_msg *mds_alloc_msg(struct ceph_connection *con,
3914                                 struct ceph_msg_header *hdr, int *skip)
3915 {
3916         struct ceph_msg *msg;
3917         int type = (int) le16_to_cpu(hdr->type);
3918         int front_len = (int) le32_to_cpu(hdr->front_len);
3919
3920         if (con->in_msg)
3921                 return con->in_msg;
3922
3923         *skip = 0;
3924         msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
3925         if (!msg) {
3926                 pr_err("unable to allocate msg type %d len %d\n",
3927                        type, front_len);
3928                 return NULL;
3929         }
3930
3931         return msg;
3932 }
3933
3934 static int mds_sign_message(struct ceph_msg *msg)
3935 {
3936        struct ceph_mds_session *s = msg->con->private;
3937        struct ceph_auth_handshake *auth = &s->s_auth;
3938
3939        return ceph_auth_sign_message(auth, msg);
3940 }
3941
3942 static int mds_check_message_signature(struct ceph_msg *msg)
3943 {
3944        struct ceph_mds_session *s = msg->con->private;
3945        struct ceph_auth_handshake *auth = &s->s_auth;
3946
3947        return ceph_auth_check_message_signature(auth, msg);
3948 }
3949
3950 static const struct ceph_connection_operations mds_con_ops = {
3951         .get = con_get,
3952         .put = con_put,
3953         .dispatch = dispatch,
3954         .get_authorizer = get_authorizer,
3955         .verify_authorizer_reply = verify_authorizer_reply,
3956         .invalidate_authorizer = invalidate_authorizer,
3957         .peer_reset = peer_reset,
3958         .alloc_msg = mds_alloc_msg,
3959         .sign_message = mds_sign_message,
3960         .check_message_signature = mds_check_message_signature,
3961 };
3962
3963 /* eof */