fs/nfs/pnfs_nfs.c

   1 /*
   2  * Common NFS I/O  operations for the pnfs file based
   3  * layout drivers.
   4  *
   5  * Copyright (c) 2014, Primary Data, Inc. All rights reserved.
   6  *
   7  * Tom Haynes <loghyr@primarydata.com>
   8  */
   9
  10 #include <linux/nfs_fs.h>
  11 #include <linux/nfs_page.h>
  12 #include <linux/sunrpc/addr.h>
  13 #include <linux/module.h>
  14
  15 #include "nfs4session.h"
  16 #include "internal.h"
  17 #include "pnfs.h"
  18
  19 #define NFSDBG_FACILITY         NFSDBG_PNFS
  20
  21 void pnfs_generic_rw_release(void *data)
  22 {
  23         struct nfs_pgio_header *hdr = data;
  24
  25         nfs_put_client(hdr->ds_clp);
  26         hdr->mds_ops->rpc_release(data);
  27 }
  28 EXPORT_SYMBOL_GPL(pnfs_generic_rw_release);
  29
  30 /* Fake up some data that will cause nfs_commit_release to retry the writes. */
  31 void pnfs_generic_prepare_to_resend_writes(struct nfs_commit_data *data)
  32 {
  33         struct nfs_page *first = nfs_list_entry(data->pages.next);
  34
  35         data->task.tk_status = 0;
  36         memcpy(&data->verf.verifier, &first->wb_verf,
  37                sizeof(data->verf.verifier));
  38         data->verf.verifier.data[0]++; /* ensure verifier mismatch */
  39 }
  40 EXPORT_SYMBOL_GPL(pnfs_generic_prepare_to_resend_writes);
  41
  42 void pnfs_generic_write_commit_done(struct rpc_task *task, void *data)
  43 {
  44         struct nfs_commit_data *wdata = data;
  45
  46         /* Note this may cause RPC to be resent */
  47         wdata->mds_ops->rpc_call_done(task, data);
  48 }
  49 EXPORT_SYMBOL_GPL(pnfs_generic_write_commit_done);
  50
  51 void pnfs_generic_commit_release(void *calldata)
  52 {
  53         struct nfs_commit_data *data = calldata;
  54
  55         data->completion_ops->completion(data);
  56         pnfs_put_lseg(data->lseg);
  57         nfs_put_client(data->ds_clp);
  58         nfs_commitdata_release(data);
  59 }
  60 EXPORT_SYMBOL_GPL(pnfs_generic_commit_release);
  61
  62 /* The generic layer is about to remove the req from the commit list.
  63  * If this will make the bucket empty, it will need to put the lseg reference.
  64  * Note this must be called holding the inode (/cinfo) lock
  65  */
  66 void
  67 pnfs_generic_clear_request_commit(struct nfs_page *req,
  68                                   struct nfs_commit_info *cinfo)
  69 {
  70         struct pnfs_layout_segment *freeme = NULL;
  71
  72         if (!test_and_clear_bit(PG_COMMIT_TO_DS, &req->wb_flags))
  73                 goto out;
  74         cinfo->ds->nwritten--;
  75         if (list_is_singular(&req->wb_list)) {
  76                 struct pnfs_commit_bucket *bucket;
  77
  78                 bucket = list_first_entry(&req->wb_list,
  79                                           struct pnfs_commit_bucket,
  80                                           written);
  81                 freeme = bucket->wlseg;
  82                 bucket->wlseg = NULL;
  83         }
  84 out:
  85         nfs_request_remove_commit_list(req, cinfo);
  86         pnfs_put_lseg_locked(freeme);
  87 }
  88 EXPORT_SYMBOL_GPL(pnfs_generic_clear_request_commit);
  89
  90 static int
  91 pnfs_generic_transfer_commit_list(struct list_head *src, struct list_head *dst,
  92                                   struct nfs_commit_info *cinfo, int max)
  93 {
  94         struct nfs_page *req, *tmp;
  95         int ret = 0;
  96
  97         list_for_each_entry_safe(req, tmp, src, wb_list) {
  98                 if (!nfs_lock_request(req))
  99                         continue;
 100                 kref_get(&req->wb_kref);
 101                 if (cond_resched_lock(cinfo->lock))
 102                         list_safe_reset_next(req, tmp, wb_list);
 103                 nfs_request_remove_commit_list(req, cinfo);
 104                 clear_bit(PG_COMMIT_TO_DS, &req->wb_flags);
 105                 nfs_list_add_request(req, dst);
 106                 ret++;
 107                 if ((ret == max) && !cinfo->dreq)
 108                         break;
 109         }
 110         return ret;
 111 }
 112
 113 static int
 114 pnfs_generic_scan_ds_commit_list(struct pnfs_commit_bucket *bucket,
 115                                  struct nfs_commit_info *cinfo,
 116                                  int max)
 117 {
 118         struct list_head *src = &bucket->written;
 119         struct list_head *dst = &bucket->committing;
 120         int ret;
 121
 122         lockdep_assert_held(cinfo->lock);
 123         ret = pnfs_generic_transfer_commit_list(src, dst, cinfo, max);
 124         if (ret) {
 125                 cinfo->ds->nwritten -= ret;
 126                 cinfo->ds->ncommitting += ret;
 127                 if (bucket->clseg == NULL)
 128                         bucket->clseg = pnfs_get_lseg(bucket->wlseg);
 129                 if (list_empty(src)) {
 130                         pnfs_put_lseg_locked(bucket->wlseg);
 131                         bucket->wlseg = NULL;
 132                 }
 133         }
 134         return ret;
 135 }
 136
 137 /* Move reqs from written to committing lists, returning count
 138  * of number moved.
 139  */
 140 int pnfs_generic_scan_commit_lists(struct nfs_commit_info *cinfo,
 141                                    int max)
 142 {
 143         int i, rv = 0, cnt;
 144
 145         lockdep_assert_held(cinfo->lock);
 146         for (i = 0; i < cinfo->ds->nbuckets && max != 0; i++) {
 147                 cnt = pnfs_generic_scan_ds_commit_list(&cinfo->ds->buckets[i],
 148                                                        cinfo, max);
 149                 max -= cnt;
 150                 rv += cnt;
 151         }
 152         return rv;
 153 }
 154 EXPORT_SYMBOL_GPL(pnfs_generic_scan_commit_lists);
 155
 156 /* Pull everything off the committing lists and dump into @dst.  */
 157 void pnfs_generic_recover_commit_reqs(struct list_head *dst,
 158                                       struct nfs_commit_info *cinfo)
 159 {
 160         struct pnfs_commit_bucket *b;
 161         struct pnfs_layout_segment *freeme;
 162         int i;
 163
 164         lockdep_assert_held(cinfo->lock);
 165 restart:
 166         for (i = 0, b = cinfo->ds->buckets; i < cinfo->ds->nbuckets; i++, b++) {
 167                 if (pnfs_generic_transfer_commit_list(&b->written, dst,
 168                                                       cinfo, 0)) {
 169                         freeme = b->wlseg;
 170                         b->wlseg = NULL;
 171                         spin_unlock(cinfo->lock);
 172                         pnfs_put_lseg(freeme);
 173                         spin_lock(cinfo->lock);
 174                         goto restart;
 175                 }
 176         }
 177         cinfo->ds->nwritten = 0;
 178 }
 179 EXPORT_SYMBOL_GPL(pnfs_generic_recover_commit_reqs);
 180
 181 static void pnfs_generic_retry_commit(struct nfs_commit_info *cinfo, int idx)
 182 {
 183         struct pnfs_ds_commit_info *fl_cinfo = cinfo->ds;
 184         struct pnfs_commit_bucket *bucket;
 185         struct pnfs_layout_segment *freeme;
 186         LIST_HEAD(pages);
 187         int i;
 188
 189         spin_lock(cinfo->lock);
 190         for (i = idx; i < fl_cinfo->nbuckets; i++) {
 191                 bucket = &fl_cinfo->buckets[i];
 192                 if (list_empty(&bucket->committing))
 193                         continue;
 194                 freeme = bucket->clseg;
 195                 bucket->clseg = NULL;
 196                 list_splice_init(&bucket->committing, &pages);
 197                 spin_unlock(cinfo->lock);
 198                 nfs_retry_commit(&pages, freeme, cinfo, i);
 199                 pnfs_put_lseg(freeme);
 200                 spin_lock(cinfo->lock);
 201         }
 202         spin_unlock(cinfo->lock);
 203 }
 204
 205 static unsigned int
 206 pnfs_generic_alloc_ds_commits(struct nfs_commit_info *cinfo,
 207                               struct list_head *list)
 208 {
 209         struct pnfs_ds_commit_info *fl_cinfo;
 210         struct pnfs_commit_bucket *bucket;
 211         struct nfs_commit_data *data;
 212         int i;
 213         unsigned int nreq = 0;
 214
 215         fl_cinfo = cinfo->ds;
 216         bucket = fl_cinfo->buckets;
 217         for (i = 0; i < fl_cinfo->nbuckets; i++, bucket++) {
 218                 if (list_empty(&bucket->committing))
 219                         continue;
 220                 data = nfs_commitdata_alloc();
 221                 if (!data)
 222                         break;
 223                 data->ds_commit_index = i;
 224                 list_add(&data->pages, list);
 225                 nreq++;
 226         }
 227
 228         /* Clean up on error */
 229         pnfs_generic_retry_commit(cinfo, i);
 230         return nreq;
 231 }
 232
 233 static inline
 234 void pnfs_fetch_commit_bucket_list(struct list_head *pages,
 235                 struct nfs_commit_data *data,
 236                 struct nfs_commit_info *cinfo)
 237 {
 238         struct pnfs_commit_bucket *bucket;
 239
 240         bucket = &cinfo->ds->buckets[data->ds_commit_index];
 241         spin_lock(cinfo->lock);
 242         list_splice_init(&bucket->committing, pages);
 243         data->lseg = bucket->clseg;
 244         bucket->clseg = NULL;
 245         spin_unlock(cinfo->lock);
 246
 247 }
 248
 249 /* This follows nfs_commit_list pretty closely */
 250 int
 251 pnfs_generic_commit_pagelist(struct inode *inode, struct list_head *mds_pages,
 252                              int how, struct nfs_commit_info *cinfo,
 253                              int (*initiate_commit)(struct nfs_commit_data *data,
 254                                                     int how))
 255 {
 256         struct nfs_commit_data *data, *tmp;
 257         LIST_HEAD(list);
 258         unsigned int nreq = 0;
 259
 260         if (!list_empty(mds_pages)) {
 261                 data = nfs_commitdata_alloc();
 262                 if (data != NULL) {
 263                         data->ds_commit_index = -1;
 264                         list_add(&data->pages, &list);
 265                         nreq++;
 266                 } else {
 267                         nfs_retry_commit(mds_pages, NULL, cinfo, 0);
 268                         pnfs_generic_retry_commit(cinfo, 0);
 269                         cinfo->completion_ops->error_cleanup(NFS_I(inode));
 270                         return -ENOMEM;
 271                 }
 272         }
 273
 274         nreq += pnfs_generic_alloc_ds_commits(cinfo, &list);
 275
 276         if (nreq == 0) {
 277                 cinfo->completion_ops->error_cleanup(NFS_I(inode));
 278                 goto out;
 279         }
 280
 281         atomic_add(nreq, &cinfo->mds->rpcs_out);
 282
 283         list_for_each_entry_safe(data, tmp, &list, pages) {
 284                 list_del_init(&data->pages);
 285                 if (data->ds_commit_index < 0) {
 286                         nfs_init_commit(data, mds_pages, NULL, cinfo);
 287                         nfs_initiate_commit(NFS_CLIENT(inode), data,
 288                                             NFS_PROTO(data->inode),
 289                                             data->mds_ops, how, 0);
 290                 } else {
 291                         LIST_HEAD(pages);
 292
 293                         pnfs_fetch_commit_bucket_list(&pages, data, cinfo);
 294                         nfs_init_commit(data, &pages, data->lseg, cinfo);
 295                         initiate_commit(data, how);
 296                 }
 297         }
 298 out:
 299         cinfo->ds->ncommitting = 0;
 300         return PNFS_ATTEMPTED;
 301 }
 302 EXPORT_SYMBOL_GPL(pnfs_generic_commit_pagelist);
 303
 304 /*
 305  * Data server cache
 306  *
 307  * Data servers can be mapped to different device ids.
 308  * nfs4_pnfs_ds reference counting
 309  *   - set to 1 on allocation
 310  *   - incremented when a device id maps a data server already in the cache.
 311  *   - decremented when deviceid is removed from the cache.
 312  */
 313 static DEFINE_SPINLOCK(nfs4_ds_cache_lock);
 314 static LIST_HEAD(nfs4_data_server_cache);
 315
 316 /* Debug routines */
 317 static void
 318 print_ds(struct nfs4_pnfs_ds *ds)
 319 {
 320         if (ds == NULL) {
 321                 printk(KERN_WARNING "%s NULL device\n", __func__);
 322                 return;
 323         }
 324         printk(KERN_WARNING "        ds %s\n"
 325                 "        ref count %d\n"
 326                 "        client %p\n"
 327                 "        cl_exchange_flags %x\n",
 328                 ds->ds_remotestr,
 329                 atomic_read(&ds->ds_count), ds->ds_clp,
 330                 ds->ds_clp ? ds->ds_clp->cl_exchange_flags : 0);
 331 }
 332
 333 static bool
 334 same_sockaddr(struct sockaddr *addr1, struct sockaddr *addr2)
 335 {
 336         struct sockaddr_in *a, *b;
 337         struct sockaddr_in6 *a6, *b6;
 338
 339         if (addr1->sa_family != addr2->sa_family)
 340                 return false;
 341
 342         switch (addr1->sa_family) {
 343         case AF_INET:
 344                 a = (struct sockaddr_in *)addr1;
 345                 b = (struct sockaddr_in *)addr2;
 346
 347                 if (a->sin_addr.s_addr == b->sin_addr.s_addr &&
 348                     a->sin_port == b->sin_port)
 349                         return true;
 350                 break;
 351
 352         case AF_INET6:
 353                 a6 = (struct sockaddr_in6 *)addr1;
 354                 b6 = (struct sockaddr_in6 *)addr2;
 355
 356                 /* LINKLOCAL addresses must have matching scope_id */
 357                 if (ipv6_addr_src_scope(&a6->sin6_addr) ==
 358                     IPV6_ADDR_SCOPE_LINKLOCAL &&
 359                     a6->sin6_scope_id != b6->sin6_scope_id)
 360                         return false;
 361
 362                 if (ipv6_addr_equal(&a6->sin6_addr, &b6->sin6_addr) &&
 363                     a6->sin6_port == b6->sin6_port)
 364                         return true;
 365                 break;
 366
 367         default:
 368                 dprintk("%s: unhandled address family: %u\n",
 369                         __func__, addr1->sa_family);
 370                 return false;
 371         }
 372
 373         return false;
 374 }
 375
 376 /*
 377  * Checks if 'dsaddrs1' contains a subset of 'dsaddrs2'. If it does,
 378  * declare a match.
 379  */
 380 static bool
 381 _same_data_server_addrs_locked(const struct list_head *dsaddrs1,
 382                                const struct list_head *dsaddrs2)
 383 {
 384         struct nfs4_pnfs_ds_addr *da1, *da2;
 385         struct sockaddr *sa1, *sa2;
 386         bool match = false;
 387
 388         list_for_each_entry(da1, dsaddrs1, da_node) {
 389                 sa1 = (struct sockaddr *)&da1->da_addr;
 390                 match = false;
 391                 list_for_each_entry(da2, dsaddrs2, da_node) {
 392                         sa2 = (struct sockaddr *)&da2->da_addr;
 393                         match = same_sockaddr(sa1, sa2);
 394                         if (match)
 395                                 break;
 396                 }
 397                 if (!match)
 398                         break;
 399         }
 400         return match;
 401 }
 402
 403 /*
 404  * Lookup DS by addresses.  nfs4_ds_cache_lock is held
 405  */
 406 static struct nfs4_pnfs_ds *
 407 _data_server_lookup_locked(const struct list_head *dsaddrs)
 408 {
 409         struct nfs4_pnfs_ds *ds;
 410
 411         list_for_each_entry(ds, &nfs4_data_server_cache, ds_node)
 412                 if (_same_data_server_addrs_locked(&ds->ds_addrs, dsaddrs))
 413                         return ds;
 414         return NULL;
 415 }
 416
 417 static void destroy_ds(struct nfs4_pnfs_ds *ds)
 418 {
 419         struct nfs4_pnfs_ds_addr *da;
 420
 421         dprintk("--> %s\n", __func__);
 422         ifdebug(FACILITY)
 423                 print_ds(ds);
 424
 425         nfs_put_client(ds->ds_clp);
 426
 427         while (!list_empty(&ds->ds_addrs)) {
 428                 da = list_first_entry(&ds->ds_addrs,
 429                                       struct nfs4_pnfs_ds_addr,
 430                                       da_node);
 431                 list_del_init(&da->da_node);
 432                 kfree(da->da_remotestr);
 433                 kfree(da);
 434         }
 435
 436         kfree(ds->ds_remotestr);
 437         kfree(ds);
 438 }
 439
 440 void nfs4_pnfs_ds_put(struct nfs4_pnfs_ds *ds)
 441 {
 442         if (atomic_dec_and_lock(&ds->ds_count,
 443                                 &nfs4_ds_cache_lock)) {
 444                 list_del_init(&ds->ds_node);
 445                 spin_unlock(&nfs4_ds_cache_lock);
 446                 destroy_ds(ds);
 447         }
 448 }
 449 EXPORT_SYMBOL_GPL(nfs4_pnfs_ds_put);
 450
 451 /*
 452  * Create a string with a human readable address and port to avoid
 453  * complicated setup around many dprinks.
 454  */
 455 static char *
 456 nfs4_pnfs_remotestr(struct list_head *dsaddrs, gfp_t gfp_flags)
 457 {
 458         struct nfs4_pnfs_ds_addr *da;
 459         char *remotestr;
 460         size_t len;
 461         char *p;
 462
 463         len = 3;        /* '{', '}' and eol */
 464         list_for_each_entry(da, dsaddrs, da_node) {
 465                 len += strlen(da->da_remotestr) + 1;    /* string plus comma */
 466         }
 467
 468         remotestr = kzalloc(len, gfp_flags);
 469         if (!remotestr)
 470                 return NULL;
 471
 472         p = remotestr;
 473         *(p++) = '{';
 474         len--;
 475         list_for_each_entry(da, dsaddrs, da_node) {
 476                 size_t ll = strlen(da->da_remotestr);
 477
 478                 if (ll > len)
 479                         goto out_err;
 480
 481                 memcpy(p, da->da_remotestr, ll);
 482                 p += ll;
 483                 len -= ll;
 484
 485                 if (len < 1)
 486                         goto out_err;
 487                 (*p++) = ',';
 488                 len--;
 489         }
 490         if (len < 2)
 491                 goto out_err;
 492         *(p++) = '}';
 493         *p = '\0';
 494         return remotestr;
 495 out_err:
 496         kfree(remotestr);
 497         return NULL;
 498 }
 499
 500 /*
 501  * Given a list of multipath struct nfs4_pnfs_ds_addr, add it to ds cache if
 502  * uncached and return cached struct nfs4_pnfs_ds.
 503  */
 504 struct nfs4_pnfs_ds *
 505 nfs4_pnfs_ds_add(struct list_head *dsaddrs, gfp_t gfp_flags)
 506 {
 507         struct nfs4_pnfs_ds *tmp_ds, *ds = NULL;
 508         char *remotestr;
 509
 510         if (list_empty(dsaddrs)) {
 511                 dprintk("%s: no addresses defined\n", __func__);
 512                 goto out;
 513         }
 514
 515         ds = kzalloc(sizeof(*ds), gfp_flags);
 516         if (!ds)
 517                 goto out;
 518
 519         /* this is only used for debugging, so it's ok if its NULL */
 520         remotestr = nfs4_pnfs_remotestr(dsaddrs, gfp_flags);
 521
 522         spin_lock(&nfs4_ds_cache_lock);
 523         tmp_ds = _data_server_lookup_locked(dsaddrs);
 524         if (tmp_ds == NULL) {
 525                 INIT_LIST_HEAD(&ds->ds_addrs);
 526                 list_splice_init(dsaddrs, &ds->ds_addrs);
 527                 ds->ds_remotestr = remotestr;
 528                 atomic_set(&ds->ds_count, 1);
 529                 INIT_LIST_HEAD(&ds->ds_node);
 530                 ds->ds_clp = NULL;
 531                 list_add(&ds->ds_node, &nfs4_data_server_cache);
 532                 dprintk("%s add new data server %s\n", __func__,
 533                         ds->ds_remotestr);
 534         } else {
 535                 kfree(remotestr);
 536                 kfree(ds);
 537                 atomic_inc(&tmp_ds->ds_count);
 538                 dprintk("%s data server %s found, inc'ed ds_count to %d\n",
 539                         __func__, tmp_ds->ds_remotestr,
 540                         atomic_read(&tmp_ds->ds_count));
 541                 ds = tmp_ds;
 542         }
 543         spin_unlock(&nfs4_ds_cache_lock);
 544 out:
 545         return ds;
 546 }
 547 EXPORT_SYMBOL_GPL(nfs4_pnfs_ds_add);
 548
 549 static void nfs4_wait_ds_connect(struct nfs4_pnfs_ds *ds)
 550 {
 551         might_sleep();
 552         wait_on_bit(&ds->ds_state, NFS4DS_CONNECTING,
 553                         TASK_KILLABLE);
 554 }
 555
 556 static void nfs4_clear_ds_conn_bit(struct nfs4_pnfs_ds *ds)
 557 {
 558         smp_mb__before_atomic();
 559         clear_bit(NFS4DS_CONNECTING, &ds->ds_state);
 560         smp_mb__after_atomic();
 561         wake_up_bit(&ds->ds_state, NFS4DS_CONNECTING);
 562 }
 563
 564 static struct nfs_client *(*get_v3_ds_connect)(
 565                         struct nfs_client *mds_clp,
 566                         const struct sockaddr *ds_addr,
 567                         int ds_addrlen,
 568                         int ds_proto,
 569                         unsigned int ds_timeo,
 570                         unsigned int ds_retrans,
 571                         rpc_authflavor_t au_flavor);
 572
 573 static bool load_v3_ds_connect(void)
 574 {
 575         if (!get_v3_ds_connect) {
 576                 get_v3_ds_connect = symbol_request(nfs3_set_ds_client);
 577                 WARN_ON_ONCE(!get_v3_ds_connect);
 578         }
 579
 580         return(get_v3_ds_connect != NULL);
 581 }
 582
 583 void nfs4_pnfs_v3_ds_connect_unload(void)
 584 {
 585         if (get_v3_ds_connect) {
 586                 symbol_put(nfs3_set_ds_client);
 587                 get_v3_ds_connect = NULL;
 588         }
 589 }
 590 EXPORT_SYMBOL_GPL(nfs4_pnfs_v3_ds_connect_unload);
 591
 592 static int _nfs4_pnfs_v3_ds_connect(struct nfs_server *mds_srv,
 593                                  struct nfs4_pnfs_ds *ds,
 594                                  unsigned int timeo,
 595                                  unsigned int retrans,
 596                                  rpc_authflavor_t au_flavor)
 597 {
 598         struct nfs_client *clp = ERR_PTR(-EIO);
 599         struct nfs4_pnfs_ds_addr *da;
 600         int status = 0;
 601
 602         dprintk("--> %s DS %s au_flavor %d\n", __func__,
 603                 ds->ds_remotestr, au_flavor);
 604
 605         if (!load_v3_ds_connect())
 606                 goto out;
 607
 608         list_for_each_entry(da, &ds->ds_addrs, da_node) {
 609                 dprintk("%s: DS %s: trying address %s\n",
 610                         __func__, ds->ds_remotestr, da->da_remotestr);
 611
 612                 clp = get_v3_ds_connect(mds_srv->nfs_client,
 613                                         (struct sockaddr *)&da->da_addr,
 614                                         da->da_addrlen, IPPROTO_TCP,
 615                                         timeo, retrans, au_flavor);
 616                 if (!IS_ERR(clp))
 617                         break;
 618         }
 619
 620         if (IS_ERR(clp)) {
 621                 status = PTR_ERR(clp);
 622                 goto out;
 623         }
 624
 625         smp_wmb();
 626         ds->ds_clp = clp;
 627         dprintk("%s [new] addr: %s\n", __func__, ds->ds_remotestr);
 628 out:
 629         return status;
 630 }
 631
 632 static int _nfs4_pnfs_v4_ds_connect(struct nfs_server *mds_srv,
 633                                  struct nfs4_pnfs_ds *ds,
 634                                  unsigned int timeo,
 635                                  unsigned int retrans,
 636                                  u32 minor_version,
 637                                  rpc_authflavor_t au_flavor)
 638 {
 639         struct nfs_client *clp = ERR_PTR(-EIO);
 640         struct nfs4_pnfs_ds_addr *da;
 641         int status = 0;
 642
 643         dprintk("--> %s DS %s au_flavor %d\n", __func__, ds->ds_remotestr,
 644                 au_flavor);
 645
 646         list_for_each_entry(da, &ds->ds_addrs, da_node) {
 647                 dprintk("%s: DS %s: trying address %s\n",
 648                         __func__, ds->ds_remotestr, da->da_remotestr);
 649
 650                 clp = nfs4_set_ds_client(mds_srv->nfs_client,
 651                                         (struct sockaddr *)&da->da_addr,
 652                                         da->da_addrlen, IPPROTO_TCP,
 653                                         timeo, retrans, minor_version,
 654                                         au_flavor);
 655                 if (!IS_ERR(clp))
 656                         break;
 657         }
 658
 659         if (IS_ERR(clp)) {
 660                 status = PTR_ERR(clp);
 661                 goto out;
 662         }
 663
 664         status = nfs4_init_ds_session(clp, mds_srv->nfs_client->cl_lease_time);
 665         if (status)
 666                 goto out_put;
 667
 668         smp_wmb();
 669         ds->ds_clp = clp;
 670         dprintk("%s [new] addr: %s\n", __func__, ds->ds_remotestr);
 671 out:
 672         return status;
 673 out_put:
 674         nfs_put_client(clp);
 675         goto out;
 676 }
 677
 678 /*
 679  * Create an rpc connection to the nfs4_pnfs_ds data server.
 680  * Currently only supports IPv4 and IPv6 addresses.
 681  * If connection fails, make devid unavailable.
 682  */
 683 void nfs4_pnfs_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds,
 684                           struct nfs4_deviceid_node *devid, unsigned int timeo,
 685                           unsigned int retrans, u32 version,
 686                           u32 minor_version, rpc_authflavor_t au_flavor)
 687 {
 688         if (test_and_set_bit(NFS4DS_CONNECTING, &ds->ds_state) == 0) {
 689                 int err = 0;
 690
 691                 if (version == 3) {
 692                         err = _nfs4_pnfs_v3_ds_connect(mds_srv, ds, timeo,
 693                                                        retrans, au_flavor);
 694                 } else if (version == 4) {
 695                         err = _nfs4_pnfs_v4_ds_connect(mds_srv, ds, timeo,
 696                                                        retrans, minor_version,
 697                                                        au_flavor);
 698                 } else {
 699                         dprintk("%s: unsupported DS version %d\n", __func__,
 700                                 version);
 701                         err = -EPROTONOSUPPORT;
 702                 }
 703
 704                 if (err)
 705                         nfs4_mark_deviceid_unavailable(devid);
 706                 nfs4_clear_ds_conn_bit(ds);
 707         } else {
 708                 nfs4_wait_ds_connect(ds);
 709         }
 710 }
 711 EXPORT_SYMBOL_GPL(nfs4_pnfs_ds_connect);
 712
 713 /*
 714  * Currently only supports ipv4, ipv6 and one multi-path address.
 715  */
 716 struct nfs4_pnfs_ds_addr *
 717 nfs4_decode_mp_ds_addr(struct net *net, struct xdr_stream *xdr, gfp_t gfp_flags)
 718 {
 719         struct nfs4_pnfs_ds_addr *da = NULL;
 720         char *buf, *portstr;
 721         __be16 port;
 722         int nlen, rlen;
 723         int tmp[2];
 724         __be32 *p;
 725         char *netid, *match_netid;
 726         size_t len, match_netid_len;
 727         char *startsep = "";
 728         char *endsep = "";
 729
 730
 731         /* r_netid */
 732         p = xdr_inline_decode(xdr, 4);
 733         if (unlikely(!p))
 734                 goto out_err;
 735         nlen = be32_to_cpup(p++);
 736
 737         p = xdr_inline_decode(xdr, nlen);
 738         if (unlikely(!p))
 739                 goto out_err;
 740
 741         netid = kmalloc(nlen+1, gfp_flags);
 742         if (unlikely(!netid))
 743                 goto out_err;
 744
 745         netid[nlen] = '\0';
 746         memcpy(netid, p, nlen);
 747
 748         /* r_addr: ip/ip6addr with port in dec octets - see RFC 5665 */
 749         p = xdr_inline_decode(xdr, 4);
 750         if (unlikely(!p))
 751                 goto out_free_netid;
 752         rlen = be32_to_cpup(p);
 753
 754         p = xdr_inline_decode(xdr, rlen);
 755         if (unlikely(!p))
 756                 goto out_free_netid;
 757
 758         /* port is ".ABC.DEF", 8 chars max */
 759         if (rlen > INET6_ADDRSTRLEN + IPV6_SCOPE_ID_LEN + 8) {
 760                 dprintk("%s: Invalid address, length %d\n", __func__,
 761                         rlen);
 762                 goto out_free_netid;
 763         }
 764         buf = kmalloc(rlen + 1, gfp_flags);
 765         if (!buf) {
 766                 dprintk("%s: Not enough memory\n", __func__);
 767                 goto out_free_netid;
 768         }
 769         buf[rlen] = '\0';
 770         memcpy(buf, p, rlen);
 771
 772         /* replace port '.' with '-' */
 773         portstr = strrchr(buf, '.');
 774         if (!portstr) {
 775                 dprintk("%s: Failed finding expected dot in port\n",
 776                         __func__);
 777                 goto out_free_buf;
 778         }
 779         *portstr = '-';
 780
 781         /* find '.' between address and port */
 782         portstr = strrchr(buf, '.');
 783         if (!portstr) {
 784                 dprintk("%s: Failed finding expected dot between address and "
 785                         "port\n", __func__);
 786                 goto out_free_buf;
 787         }
 788         *portstr = '\0';
 789
 790         da = kzalloc(sizeof(*da), gfp_flags);
 791         if (unlikely(!da))
 792                 goto out_free_buf;
 793
 794         INIT_LIST_HEAD(&da->da_node);
 795
 796         if (!rpc_pton(net, buf, portstr-buf, (struct sockaddr *)&da->da_addr,
 797                       sizeof(da->da_addr))) {
 798                 dprintk("%s: error parsing address %s\n", __func__, buf);
 799                 goto out_free_da;
 800         }
 801
 802         portstr++;
 803         sscanf(portstr, "%d-%d", &tmp[0], &tmp[1]);
 804         port = htons((tmp[0] << 8) | (tmp[1]));
 805
 806         switch (da->da_addr.ss_family) {
 807         case AF_INET:
 808                 ((struct sockaddr_in *)&da->da_addr)->sin_port = port;
 809                 da->da_addrlen = sizeof(struct sockaddr_in);
 810                 match_netid = "tcp";
 811                 match_netid_len = 3;
 812                 break;
 813
 814         case AF_INET6:
 815                 ((struct sockaddr_in6 *)&da->da_addr)->sin6_port = port;
 816                 da->da_addrlen = sizeof(struct sockaddr_in6);
 817                 match_netid = "tcp6";
 818                 match_netid_len = 4;
 819                 startsep = "[";
 820                 endsep = "]";
 821                 break;
 822
 823         default:
 824                 dprintk("%s: unsupported address family: %u\n",
 825                         __func__, da->da_addr.ss_family);
 826                 goto out_free_da;
 827         }
 828
 829         if (nlen != match_netid_len || strncmp(netid, match_netid, nlen)) {
 830                 dprintk("%s: ERROR: r_netid \"%s\" != \"%s\"\n",
 831                         __func__, netid, match_netid);
 832                 goto out_free_da;
 833         }
 834
 835         /* save human readable address */
 836         len = strlen(startsep) + strlen(buf) + strlen(endsep) + 7;
 837         da->da_remotestr = kzalloc(len, gfp_flags);
 838
 839         /* NULL is ok, only used for dprintk */
 840         if (da->da_remotestr)
 841                 snprintf(da->da_remotestr, len, "%s%s%s:%u", startsep,
 842                          buf, endsep, ntohs(port));
 843
 844         dprintk("%s: Parsed DS addr %s\n", __func__, da->da_remotestr);
 845         kfree(buf);
 846         kfree(netid);
 847         return da;
 848
 849 out_free_da:
 850         kfree(da);
 851 out_free_buf:
 852         dprintk("%s: Error parsing DS addr: %s\n", __func__, buf);
 853         kfree(buf);
 854 out_free_netid:
 855         kfree(netid);
 856 out_err:
 857         return NULL;
 858 }
 859 EXPORT_SYMBOL_GPL(nfs4_decode_mp_ds_addr);
 860
 861 void
 862 pnfs_layout_mark_request_commit(struct nfs_page *req,
 863                                 struct pnfs_layout_segment *lseg,
 864                                 struct nfs_commit_info *cinfo,
 865                                 u32 ds_commit_idx)
 866 {
 867         struct list_head *list;
 868         struct pnfs_commit_bucket *buckets;
 869
 870         spin_lock(cinfo->lock);
 871         buckets = cinfo->ds->buckets;
 872         list = &buckets[ds_commit_idx].written;
 873         if (list_empty(list)) {
 874                 /* Non-empty buckets hold a reference on the lseg.  That ref
 875                  * is normally transferred to the COMMIT call and released
 876                  * there.  It could also be released if the last req is pulled
 877                  * off due to a rewrite, in which case it will be done in
 878                  * pnfs_common_clear_request_commit
 879                  */
 880                 WARN_ON_ONCE(buckets[ds_commit_idx].wlseg != NULL);
 881                 buckets[ds_commit_idx].wlseg = pnfs_get_lseg(lseg);
 882         }
 883         set_bit(PG_COMMIT_TO_DS, &req->wb_flags);
 884         cinfo->ds->nwritten++;
 885
 886         nfs_request_add_commit_list_locked(req, list, cinfo);
 887         spin_unlock(cinfo->lock);
 888         nfs_mark_page_unstable(req->wb_page, cinfo);
 889 }
 890 EXPORT_SYMBOL_GPL(pnfs_layout_mark_request_commit);
 891
 892 int
 893 pnfs_nfs_generic_sync(struct inode *inode, bool datasync)
 894 {
 895         if (datasync)
 896                 return 0;
 897         return pnfs_layoutcommit_inode(inode, true);
 898 }
 899 EXPORT_SYMBOL_GPL(pnfs_nfs_generic_sync);
 900