fs/ceph/dir.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 #include <linux/ceph/ceph_debug.h>
   3
   4 #include <linux/spinlock.h>
   5 #include <linux/fs_struct.h>
   6 #include <linux/namei.h>
   7 #include <linux/slab.h>
   8 #include <linux/sched.h>
   9 #include <linux/xattr.h>
  10
  11 #include "super.h"
  12 #include "mds_client.h"
  13
  14 /*
  15  * Directory operations: readdir, lookup, create, link, unlink,
  16  * rename, etc.
  17  */
  18
  19 /*
  20  * Ceph MDS operations are specified in terms of a base ino and
  21  * relative path.  Thus, the client can specify an operation on a
  22  * specific inode (e.g., a getattr due to fstat(2)), or as a path
  23  * relative to, say, the root directory.
  24  *
  25  * Normally, we limit ourselves to strict inode ops (no path component)
  26  * or dentry operations (a single path component relative to an ino).  The
  27  * exception to this is open_root_dentry(), which will open the mount
  28  * point by name.
  29  */
  30
  31 const struct dentry_operations ceph_dentry_ops;
  32
  33 /*
  34  * Initialize ceph dentry state.
  35  */
  36 static int ceph_d_init(struct dentry *dentry)
  37 {
  38         struct ceph_dentry_info *di;
  39
  40         di = kmem_cache_zalloc(ceph_dentry_cachep, GFP_KERNEL);
  41         if (!di)
  42                 return -ENOMEM;          /* oh well */
  43
  44         di->dentry = dentry;
  45         di->lease_session = NULL;
  46         di->time = jiffies;
  47         dentry->d_fsdata = di;
  48         ceph_dentry_lru_add(dentry);
  49         return 0;
  50 }
  51
  52 /*
  53  * for f_pos for readdir:
  54  * - hash order:
  55  *      (0xff << 52) | ((24 bits hash) << 28) |
  56  *      (the nth entry has hash collision);
  57  * - frag+name order;
  58  *      ((frag value) << 28) | (the nth entry in frag);
  59  */
  60 #define OFFSET_BITS     28
  61 #define OFFSET_MASK     ((1 << OFFSET_BITS) - 1)
  62 #define HASH_ORDER      (0xffull << (OFFSET_BITS + 24))
  63 loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order)
  64 {
  65         loff_t fpos = ((loff_t)high << 28) | (loff_t)off;
  66         if (hash_order)
  67                 fpos |= HASH_ORDER;
  68         return fpos;
  69 }
  70
  71 static bool is_hash_order(loff_t p)
  72 {
  73         return (p & HASH_ORDER) == HASH_ORDER;
  74 }
  75
  76 static unsigned fpos_frag(loff_t p)
  77 {
  78         return p >> OFFSET_BITS;
  79 }
  80
  81 static unsigned fpos_hash(loff_t p)
  82 {
  83         return ceph_frag_value(fpos_frag(p));
  84 }
  85
  86 static unsigned fpos_off(loff_t p)
  87 {
  88         return p & OFFSET_MASK;
  89 }
  90
  91 static int fpos_cmp(loff_t l, loff_t r)
  92 {
  93         int v = ceph_frag_compare(fpos_frag(l), fpos_frag(r));
  94         if (v)
  95                 return v;
  96         return (int)(fpos_off(l) - fpos_off(r));
  97 }
  98
  99 /*
 100  * make note of the last dentry we read, so we can
 101  * continue at the same lexicographical point,
 102  * regardless of what dir changes take place on the
 103  * server.
 104  */
 105 static int note_last_dentry(struct ceph_file_info *fi, const char *name,
 106                             int len, unsigned next_offset)
 107 {
 108         char *buf = kmalloc(len+1, GFP_KERNEL);
 109         if (!buf)
 110                 return -ENOMEM;
 111         kfree(fi->last_name);
 112         fi->last_name = buf;
 113         memcpy(fi->last_name, name, len);
 114         fi->last_name[len] = 0;
 115         fi->next_offset = next_offset;
 116         dout("note_last_dentry '%s'\n", fi->last_name);
 117         return 0;
 118 }
 119
 120
 121 static struct dentry *
 122 __dcache_find_get_entry(struct dentry *parent, u64 idx,
 123                         struct ceph_readdir_cache_control *cache_ctl)
 124 {
 125         struct inode *dir = d_inode(parent);
 126         struct dentry *dentry;
 127         unsigned idx_mask = (PAGE_SIZE / sizeof(struct dentry *)) - 1;
 128         loff_t ptr_pos = idx * sizeof(struct dentry *);
 129         pgoff_t ptr_pgoff = ptr_pos >> PAGE_SHIFT;
 130
 131         if (ptr_pos >= i_size_read(dir))
 132                 return NULL;
 133
 134         if (!cache_ctl->page || ptr_pgoff != page_index(cache_ctl->page)) {
 135                 ceph_readdir_cache_release(cache_ctl);
 136                 cache_ctl->page = find_lock_page(&dir->i_data, ptr_pgoff);
 137                 if (!cache_ctl->page) {
 138                         dout(" page %lu not found\n", ptr_pgoff);
 139                         return ERR_PTR(-EAGAIN);
 140                 }
 141                 /* reading/filling the cache are serialized by
 142                    i_mutex, no need to use page lock */
 143                 unlock_page(cache_ctl->page);
 144                 cache_ctl->dentries = kmap(cache_ctl->page);
 145         }
 146
 147         cache_ctl->index = idx & idx_mask;
 148
 149         rcu_read_lock();
 150         spin_lock(&parent->d_lock);
 151         /* check i_size again here, because empty directory can be
 152          * marked as complete while not holding the i_mutex. */
 153         if (ceph_dir_is_complete_ordered(dir) && ptr_pos < i_size_read(dir))
 154                 dentry = cache_ctl->dentries[cache_ctl->index];
 155         else
 156                 dentry = NULL;
 157         spin_unlock(&parent->d_lock);
 158         if (dentry && !lockref_get_not_dead(&dentry->d_lockref))
 159                 dentry = NULL;
 160         rcu_read_unlock();
 161         return dentry ? : ERR_PTR(-EAGAIN);
 162 }
 163
 164 /*
 165  * When possible, we try to satisfy a readdir by peeking at the
 166  * dcache.  We make this work by carefully ordering dentries on
 167  * d_child when we initially get results back from the MDS, and
 168  * falling back to a "normal" sync readdir if any dentries in the dir
 169  * are dropped.
 170  *
 171  * Complete dir indicates that we have all dentries in the dir.  It is
 172  * defined IFF we hold CEPH_CAP_FILE_SHARED (which will be revoked by
 173  * the MDS if/when the directory is modified).
 174  */
 175 static int __dcache_readdir(struct file *file,  struct dir_context *ctx,
 176                             int shared_gen)
 177 {
 178         struct ceph_file_info *fi = file->private_data;
 179         struct dentry *parent = file->f_path.dentry;
 180         struct inode *dir = d_inode(parent);
 181         struct dentry *dentry, *last = NULL;
 182         struct ceph_dentry_info *di;
 183         struct ceph_readdir_cache_control cache_ctl = {};
 184         u64 idx = 0;
 185         int err = 0;
 186
 187         dout("__dcache_readdir %p v%u at %llx\n", dir, (unsigned)shared_gen, ctx->pos);
 188
 189         /* search start position */
 190         if (ctx->pos > 2) {
 191                 u64 count = div_u64(i_size_read(dir), sizeof(struct dentry *));
 192                 while (count > 0) {
 193                         u64 step = count >> 1;
 194                         dentry = __dcache_find_get_entry(parent, idx + step,
 195                                                          &cache_ctl);
 196                         if (!dentry) {
 197                                 /* use linar search */
 198                                 idx = 0;
 199                                 break;
 200                         }
 201                         if (IS_ERR(dentry)) {
 202                                 err = PTR_ERR(dentry);
 203                                 goto out;
 204                         }
 205                         di = ceph_dentry(dentry);
 206                         spin_lock(&dentry->d_lock);
 207                         if (fpos_cmp(di->offset, ctx->pos) < 0) {
 208                                 idx += step + 1;
 209                                 count -= step + 1;
 210                         } else {
 211                                 count = step;
 212                         }
 213                         spin_unlock(&dentry->d_lock);
 214                         dput(dentry);
 215                 }
 216
 217                 dout("__dcache_readdir %p cache idx %llu\n", dir, idx);
 218         }
 219
 220
 221         for (;;) {
 222                 bool emit_dentry = false;
 223                 dentry = __dcache_find_get_entry(parent, idx++, &cache_ctl);
 224                 if (!dentry) {
 225                         fi->flags |= CEPH_F_ATEND;
 226                         err = 0;
 227                         break;
 228                 }
 229                 if (IS_ERR(dentry)) {
 230                         err = PTR_ERR(dentry);
 231                         goto out;
 232                 }
 233
 234                 spin_lock(&dentry->d_lock);
 235                 di = ceph_dentry(dentry);
 236                 if (d_unhashed(dentry) ||
 237                     d_really_is_negative(dentry) ||
 238                     di->lease_shared_gen != shared_gen) {
 239                         spin_unlock(&dentry->d_lock);
 240                         dput(dentry);
 241                         err = -EAGAIN;
 242                         goto out;
 243                 }
 244                 if (fpos_cmp(ctx->pos, di->offset) <= 0) {
 245                         emit_dentry = true;
 246                 }
 247                 spin_unlock(&dentry->d_lock);
 248
 249                 if (emit_dentry) {
 250                         dout(" %llx dentry %p %pd %p\n", di->offset,
 251                              dentry, dentry, d_inode(dentry));
 252                         ctx->pos = di->offset;
 253                         if (!dir_emit(ctx, dentry->d_name.name,
 254                                       dentry->d_name.len,
 255                                       ceph_translate_ino(dentry->d_sb,
 256                                                          d_inode(dentry)->i_ino),
 257                                       d_inode(dentry)->i_mode >> 12)) {
 258                                 dput(dentry);
 259                                 err = 0;
 260                                 break;
 261                         }
 262                         ctx->pos++;
 263
 264                         if (last)
 265                                 dput(last);
 266                         last = dentry;
 267                 } else {
 268                         dput(dentry);
 269                 }
 270         }
 271 out:
 272         ceph_readdir_cache_release(&cache_ctl);
 273         if (last) {
 274                 int ret;
 275                 di = ceph_dentry(last);
 276                 ret = note_last_dentry(fi, last->d_name.name, last->d_name.len,
 277                                        fpos_off(di->offset) + 1);
 278                 if (ret < 0)
 279                         err = ret;
 280                 dput(last);
 281                 /* last_name no longer match cache index */
 282                 if (fi->readdir_cache_idx >= 0) {
 283                         fi->readdir_cache_idx = -1;
 284                         fi->dir_release_count = 0;
 285                 }
 286         }
 287         return err;
 288 }
 289
 290 static bool need_send_readdir(struct ceph_file_info *fi, loff_t pos)
 291 {
 292         if (!fi->last_readdir)
 293                 return true;
 294         if (is_hash_order(pos))
 295                 return !ceph_frag_contains_value(fi->frag, fpos_hash(pos));
 296         else
 297                 return fi->frag != fpos_frag(pos);
 298 }
 299
 300 static int ceph_readdir(struct file *file, struct dir_context *ctx)
 301 {
 302         struct ceph_file_info *fi = file->private_data;
 303         struct inode *inode = file_inode(file);
 304         struct ceph_inode_info *ci = ceph_inode(inode);
 305         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 306         struct ceph_mds_client *mdsc = fsc->mdsc;
 307         int i;
 308         int err;
 309         unsigned frag = -1;
 310         struct ceph_mds_reply_info_parsed *rinfo;
 311
 312         dout("readdir %p file %p pos %llx\n", inode, file, ctx->pos);
 313         if (fi->flags & CEPH_F_ATEND)
 314                 return 0;
 315
 316         /* always start with . and .. */
 317         if (ctx->pos == 0) {
 318                 dout("readdir off 0 -> '.'\n");
 319                 if (!dir_emit(ctx, ".", 1,
 320                             ceph_translate_ino(inode->i_sb, inode->i_ino),
 321                             inode->i_mode >> 12))
 322                         return 0;
 323                 ctx->pos = 1;
 324         }
 325         if (ctx->pos == 1) {
 326                 ino_t ino = parent_ino(file->f_path.dentry);
 327                 dout("readdir off 1 -> '..'\n");
 328                 if (!dir_emit(ctx, "..", 2,
 329                             ceph_translate_ino(inode->i_sb, ino),
 330                             inode->i_mode >> 12))
 331                         return 0;
 332                 ctx->pos = 2;
 333         }
 334
 335         /* can we use the dcache? */
 336         spin_lock(&ci->i_ceph_lock);
 337         if (ceph_test_mount_opt(fsc, DCACHE) &&
 338             !ceph_test_mount_opt(fsc, NOASYNCREADDIR) &&
 339             ceph_snap(inode) != CEPH_SNAPDIR &&
 340             __ceph_dir_is_complete_ordered(ci) &&
 341             __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1)) {
 342                 int shared_gen = atomic_read(&ci->i_shared_gen);
 343                 spin_unlock(&ci->i_ceph_lock);
 344                 err = __dcache_readdir(file, ctx, shared_gen);
 345                 if (err != -EAGAIN)
 346                         return err;
 347         } else {
 348                 spin_unlock(&ci->i_ceph_lock);
 349         }
 350
 351         /* proceed with a normal readdir */
 352 more:
 353         /* do we have the correct frag content buffered? */
 354         if (need_send_readdir(fi, ctx->pos)) {
 355                 struct ceph_mds_request *req;
 356                 int op = ceph_snap(inode) == CEPH_SNAPDIR ?
 357                         CEPH_MDS_OP_LSSNAP : CEPH_MDS_OP_READDIR;
 358
 359                 /* discard old result, if any */
 360                 if (fi->last_readdir) {
 361                         ceph_mdsc_put_request(fi->last_readdir);
 362                         fi->last_readdir = NULL;
 363                 }
 364
 365                 if (is_hash_order(ctx->pos)) {
 366                         /* fragtree isn't always accurate. choose frag
 367                          * based on previous reply when possible. */
 368                         if (frag == (unsigned)-1)
 369                                 frag = ceph_choose_frag(ci, fpos_hash(ctx->pos),
 370                                                         NULL, NULL);
 371                 } else {
 372                         frag = fpos_frag(ctx->pos);
 373                 }
 374
 375                 dout("readdir fetching %llx.%llx frag %x offset '%s'\n",
 376                      ceph_vinop(inode), frag, fi->last_name);
 377                 req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
 378                 if (IS_ERR(req))
 379                         return PTR_ERR(req);
 380                 err = ceph_alloc_readdir_reply_buffer(req, inode);
 381                 if (err) {
 382                         ceph_mdsc_put_request(req);
 383                         return err;
 384                 }
 385                 /* hints to request -> mds selection code */
 386                 req->r_direct_mode = USE_AUTH_MDS;
 387                 if (op == CEPH_MDS_OP_READDIR) {
 388                         req->r_direct_hash = ceph_frag_value(frag);
 389                         __set_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
 390                         req->r_inode_drop = CEPH_CAP_FILE_EXCL;
 391                 }
 392                 if (fi->last_name) {
 393                         req->r_path2 = kstrdup(fi->last_name, GFP_KERNEL);
 394                         if (!req->r_path2) {
 395                                 ceph_mdsc_put_request(req);
 396                                 return -ENOMEM;
 397                         }
 398                 } else if (is_hash_order(ctx->pos)) {
 399                         req->r_args.readdir.offset_hash =
 400                                 cpu_to_le32(fpos_hash(ctx->pos));
 401                 }
 402
 403                 req->r_dir_release_cnt = fi->dir_release_count;
 404                 req->r_dir_ordered_cnt = fi->dir_ordered_count;
 405                 req->r_readdir_cache_idx = fi->readdir_cache_idx;
 406                 req->r_readdir_offset = fi->next_offset;
 407                 req->r_args.readdir.frag = cpu_to_le32(frag);
 408                 req->r_args.readdir.flags =
 409                                 cpu_to_le16(CEPH_READDIR_REPLY_BITFLAGS);
 410
 411                 req->r_inode = inode;
 412                 ihold(inode);
 413                 req->r_dentry = dget(file->f_path.dentry);
 414                 err = ceph_mdsc_do_request(mdsc, NULL, req);
 415                 if (err < 0) {
 416                         ceph_mdsc_put_request(req);
 417                         return err;
 418                 }
 419                 dout("readdir got and parsed readdir result=%d on "
 420                      "frag %x, end=%d, complete=%d, hash_order=%d\n",
 421                      err, frag,
 422                      (int)req->r_reply_info.dir_end,
 423                      (int)req->r_reply_info.dir_complete,
 424                      (int)req->r_reply_info.hash_order);
 425
 426                 rinfo = &req->r_reply_info;
 427                 if (le32_to_cpu(rinfo->dir_dir->frag) != frag) {
 428                         frag = le32_to_cpu(rinfo->dir_dir->frag);
 429                         if (!rinfo->hash_order) {
 430                                 fi->next_offset = req->r_readdir_offset;
 431                                 /* adjust ctx->pos to beginning of frag */
 432                                 ctx->pos = ceph_make_fpos(frag,
 433                                                           fi->next_offset,
 434                                                           false);
 435                         }
 436                 }
 437
 438                 fi->frag = frag;
 439                 fi->last_readdir = req;
 440
 441                 if (test_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags)) {
 442                         fi->readdir_cache_idx = req->r_readdir_cache_idx;
 443                         if (fi->readdir_cache_idx < 0) {
 444                                 /* preclude from marking dir ordered */
 445                                 fi->dir_ordered_count = 0;
 446                         } else if (ceph_frag_is_leftmost(frag) &&
 447                                    fi->next_offset == 2) {
 448                                 /* note dir version at start of readdir so
 449                                  * we can tell if any dentries get dropped */
 450                                 fi->dir_release_count = req->r_dir_release_cnt;
 451                                 fi->dir_ordered_count = req->r_dir_ordered_cnt;
 452                         }
 453                 } else {
 454                         dout("readdir !did_prepopulate");
 455                         /* disable readdir cache */
 456                         fi->readdir_cache_idx = -1;
 457                         /* preclude from marking dir complete */
 458                         fi->dir_release_count = 0;
 459                 }
 460
 461                 /* note next offset and last dentry name */
 462                 if (rinfo->dir_nr > 0) {
 463                         struct ceph_mds_reply_dir_entry *rde =
 464                                         rinfo->dir_entries + (rinfo->dir_nr-1);
 465                         unsigned next_offset = req->r_reply_info.dir_end ?
 466                                         2 : (fpos_off(rde->offset) + 1);
 467                         err = note_last_dentry(fi, rde->name, rde->name_len,
 468                                                next_offset);
 469                         if (err)
 470                                 return err;
 471                 } else if (req->r_reply_info.dir_end) {
 472                         fi->next_offset = 2;
 473                         /* keep last name */
 474                 }
 475         }
 476
 477         rinfo = &fi->last_readdir->r_reply_info;
 478         dout("readdir frag %x num %d pos %llx chunk first %llx\n",
 479              fi->frag, rinfo->dir_nr, ctx->pos,
 480              rinfo->dir_nr ? rinfo->dir_entries[0].offset : 0LL);
 481
 482         i = 0;
 483         /* search start position */
 484         if (rinfo->dir_nr > 0) {
 485                 int step, nr = rinfo->dir_nr;
 486                 while (nr > 0) {
 487                         step = nr >> 1;
 488                         if (rinfo->dir_entries[i + step].offset < ctx->pos) {
 489                                 i +=  step + 1;
 490                                 nr -= step + 1;
 491                         } else {
 492                                 nr = step;
 493                         }
 494                 }
 495         }
 496         for (; i < rinfo->dir_nr; i++) {
 497                 struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
 498                 struct ceph_vino vino;
 499                 ino_t ino;
 500                 u32 ftype;
 501
 502                 BUG_ON(rde->offset < ctx->pos);
 503
 504                 ctx->pos = rde->offset;
 505                 dout("readdir (%d/%d) -> %llx '%.*s' %p\n",
 506                      i, rinfo->dir_nr, ctx->pos,
 507                      rde->name_len, rde->name, &rde->inode.in);
 508
 509                 BUG_ON(!rde->inode.in);
 510                 ftype = le32_to_cpu(rde->inode.in->mode) >> 12;
 511                 vino.ino = le64_to_cpu(rde->inode.in->ino);
 512                 vino.snap = le64_to_cpu(rde->inode.in->snapid);
 513                 ino = ceph_vino_to_ino(vino);
 514
 515                 if (!dir_emit(ctx, rde->name, rde->name_len,
 516                               ceph_translate_ino(inode->i_sb, ino), ftype)) {
 517                         dout("filldir stopping us...\n");
 518                         return 0;
 519                 }
 520                 ctx->pos++;
 521         }
 522
 523         ceph_mdsc_put_request(fi->last_readdir);
 524         fi->last_readdir = NULL;
 525
 526         if (fi->next_offset > 2) {
 527                 frag = fi->frag;
 528                 goto more;
 529         }
 530
 531         /* more frags? */
 532         if (!ceph_frag_is_rightmost(fi->frag)) {
 533                 frag = ceph_frag_next(fi->frag);
 534                 if (is_hash_order(ctx->pos)) {
 535                         loff_t new_pos = ceph_make_fpos(ceph_frag_value(frag),
 536                                                         fi->next_offset, true);
 537                         if (new_pos > ctx->pos)
 538                                 ctx->pos = new_pos;
 539                         /* keep last_name */
 540                 } else {
 541                         ctx->pos = ceph_make_fpos(frag, fi->next_offset, false);
 542                         kfree(fi->last_name);
 543                         fi->last_name = NULL;
 544                 }
 545                 dout("readdir next frag is %x\n", frag);
 546                 goto more;
 547         }
 548         fi->flags |= CEPH_F_ATEND;
 549
 550         /*
 551          * if dir_release_count still matches the dir, no dentries
 552          * were released during the whole readdir, and we should have
 553          * the complete dir contents in our cache.
 554          */
 555         if (atomic64_read(&ci->i_release_count) == fi->dir_release_count) {
 556                 spin_lock(&ci->i_ceph_lock);
 557                 if (fi->dir_ordered_count == atomic64_read(&ci->i_ordered_count)) {
 558                         dout(" marking %p complete and ordered\n", inode);
 559                         /* use i_size to track number of entries in
 560                          * readdir cache */
 561                         BUG_ON(fi->readdir_cache_idx < 0);
 562                         i_size_write(inode, fi->readdir_cache_idx *
 563                                      sizeof(struct dentry*));
 564                 } else {
 565                         dout(" marking %p complete\n", inode);
 566                 }
 567                 __ceph_dir_set_complete(ci, fi->dir_release_count,
 568                                         fi->dir_ordered_count);
 569                 spin_unlock(&ci->i_ceph_lock);
 570         }
 571
 572         dout("readdir %p file %p done.\n", inode, file);
 573         return 0;
 574 }
 575
 576 static void reset_readdir(struct ceph_file_info *fi)
 577 {
 578         if (fi->last_readdir) {
 579                 ceph_mdsc_put_request(fi->last_readdir);
 580                 fi->last_readdir = NULL;
 581         }
 582         kfree(fi->last_name);
 583         fi->last_name = NULL;
 584         fi->dir_release_count = 0;
 585         fi->readdir_cache_idx = -1;
 586         fi->next_offset = 2;  /* compensate for . and .. */
 587         fi->flags &= ~CEPH_F_ATEND;
 588 }
 589
 590 /*
 591  * discard buffered readdir content on seekdir(0), or seek to new frag,
 592  * or seek prior to current chunk
 593  */
 594 static bool need_reset_readdir(struct ceph_file_info *fi, loff_t new_pos)
 595 {
 596         struct ceph_mds_reply_info_parsed *rinfo;
 597         loff_t chunk_offset;
 598         if (new_pos == 0)
 599                 return true;
 600         if (is_hash_order(new_pos)) {
 601                 /* no need to reset last_name for a forward seek when
 602                  * dentries are sotred in hash order */
 603         } else if (fi->frag != fpos_frag(new_pos)) {
 604                 return true;
 605         }
 606         rinfo = fi->last_readdir ? &fi->last_readdir->r_reply_info : NULL;
 607         if (!rinfo || !rinfo->dir_nr)
 608                 return true;
 609         chunk_offset = rinfo->dir_entries[0].offset;
 610         return new_pos < chunk_offset ||
 611                is_hash_order(new_pos) != is_hash_order(chunk_offset);
 612 }
 613
 614 static loff_t ceph_dir_llseek(struct file *file, loff_t offset, int whence)
 615 {
 616         struct ceph_file_info *fi = file->private_data;
 617         struct inode *inode = file->f_mapping->host;
 618         loff_t retval;
 619
 620         inode_lock(inode);
 621         retval = -EINVAL;
 622         switch (whence) {
 623         case SEEK_CUR:
 624                 offset += file->f_pos;
 625         case SEEK_SET:
 626                 break;
 627         case SEEK_END:
 628                 retval = -EOPNOTSUPP;
 629         default:
 630                 goto out;
 631         }
 632
 633         if (offset >= 0) {
 634                 if (need_reset_readdir(fi, offset)) {
 635                         dout("dir_llseek dropping %p content\n", file);
 636                         reset_readdir(fi);
 637                 } else if (is_hash_order(offset) && offset > file->f_pos) {
 638                         /* for hash offset, we don't know if a forward seek
 639                          * is within same frag */
 640                         fi->dir_release_count = 0;
 641                         fi->readdir_cache_idx = -1;
 642                 }
 643
 644                 if (offset != file->f_pos) {
 645                         file->f_pos = offset;
 646                         file->f_version = 0;
 647                         fi->flags &= ~CEPH_F_ATEND;
 648                 }
 649                 retval = offset;
 650         }
 651 out:
 652         inode_unlock(inode);
 653         return retval;
 654 }
 655
 656 /*
 657  * Handle lookups for the hidden .snap directory.
 658  */
 659 int ceph_handle_snapdir(struct ceph_mds_request *req,
 660                         struct dentry *dentry, int err)
 661 {
 662         struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
 663         struct inode *parent = d_inode(dentry->d_parent); /* we hold i_mutex */
 664
 665         /* .snap dir? */
 666         if (err == -ENOENT &&
 667             ceph_snap(parent) == CEPH_NOSNAP &&
 668             strcmp(dentry->d_name.name,
 669                    fsc->mount_options->snapdir_name) == 0) {
 670                 struct inode *inode = ceph_get_snapdir(parent);
 671                 dout("ENOENT on snapdir %p '%pd', linking to snapdir %p\n",
 672                      dentry, dentry, inode);
 673                 BUG_ON(!d_unhashed(dentry));
 674                 d_add(dentry, inode);
 675                 err = 0;
 676         }
 677         return err;
 678 }
 679
 680 /*
 681  * Figure out final result of a lookup/open request.
 682  *
 683  * Mainly, make sure we return the final req->r_dentry (if it already
 684  * existed) in place of the original VFS-provided dentry when they
 685  * differ.
 686  *
 687  * Gracefully handle the case where the MDS replies with -ENOENT and
 688  * no trace (which it may do, at its discretion, e.g., if it doesn't
 689  * care to issue a lease on the negative dentry).
 690  */
 691 struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
 692                                   struct dentry *dentry, int err)
 693 {
 694         if (err == -ENOENT) {
 695                 /* no trace? */
 696                 err = 0;
 697                 if (!req->r_reply_info.head->is_dentry) {
 698                         dout("ENOENT and no trace, dentry %p inode %p\n",
 699                              dentry, d_inode(dentry));
 700                         if (d_really_is_positive(dentry)) {
 701                                 d_drop(dentry);
 702                                 err = -ENOENT;
 703                         } else {
 704                                 d_add(dentry, NULL);
 705                         }
 706                 }
 707         }
 708         if (err)
 709                 dentry = ERR_PTR(err);
 710         else if (dentry != req->r_dentry)
 711                 dentry = dget(req->r_dentry);   /* we got spliced */
 712         else
 713                 dentry = NULL;
 714         return dentry;
 715 }
 716
 717 static bool is_root_ceph_dentry(struct inode *inode, struct dentry *dentry)
 718 {
 719         return ceph_ino(inode) == CEPH_INO_ROOT &&
 720                 strncmp(dentry->d_name.name, ".ceph", 5) == 0;
 721 }
 722
 723 /*
 724  * Look up a single dir entry.  If there is a lookup intent, inform
 725  * the MDS so that it gets our 'caps wanted' value in a single op.
 726  */
 727 static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
 728                                   unsigned int flags)
 729 {
 730         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 731         struct ceph_mds_client *mdsc = fsc->mdsc;
 732         struct ceph_mds_request *req;
 733         int op;
 734         int mask;
 735         int err;
 736
 737         dout("lookup %p dentry %p '%pd'\n",
 738              dir, dentry, dentry);
 739
 740         if (dentry->d_name.len > NAME_MAX)
 741                 return ERR_PTR(-ENAMETOOLONG);
 742
 743         /* can we conclude ENOENT locally? */
 744         if (d_really_is_negative(dentry)) {
 745                 struct ceph_inode_info *ci = ceph_inode(dir);
 746                 struct ceph_dentry_info *di = ceph_dentry(dentry);
 747
 748                 spin_lock(&ci->i_ceph_lock);
 749                 dout(" dir %p flags are %d\n", dir, ci->i_ceph_flags);
 750                 if (strncmp(dentry->d_name.name,
 751                             fsc->mount_options->snapdir_name,
 752                             dentry->d_name.len) &&
 753                     !is_root_ceph_dentry(dir, dentry) &&
 754                     ceph_test_mount_opt(fsc, DCACHE) &&
 755                     __ceph_dir_is_complete(ci) &&
 756                     (__ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1))) {
 757                         spin_unlock(&ci->i_ceph_lock);
 758                         dout(" dir %p complete, -ENOENT\n", dir);
 759                         d_add(dentry, NULL);
 760                         di->lease_shared_gen = atomic_read(&ci->i_shared_gen);
 761                         return NULL;
 762                 }
 763                 spin_unlock(&ci->i_ceph_lock);
 764         }
 765
 766         op = ceph_snap(dir) == CEPH_SNAPDIR ?
 767                 CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP;
 768         req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS);
 769         if (IS_ERR(req))
 770                 return ERR_CAST(req);
 771         req->r_dentry = dget(dentry);
 772         req->r_num_caps = 2;
 773
 774         mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
 775         if (ceph_security_xattr_wanted(dir))
 776                 mask |= CEPH_CAP_XATTR_SHARED;
 777         req->r_args.getattr.mask = cpu_to_le32(mask);
 778
 779         req->r_parent = dir;
 780         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 781         err = ceph_mdsc_do_request(mdsc, NULL, req);
 782         err = ceph_handle_snapdir(req, dentry, err);
 783         dentry = ceph_finish_lookup(req, dentry, err);
 784         ceph_mdsc_put_request(req);  /* will dput(dentry) */
 785         dout("lookup result=%p\n", dentry);
 786         return dentry;
 787 }
 788
 789 /*
 790  * If we do a create but get no trace back from the MDS, follow up with
 791  * a lookup (the VFS expects us to link up the provided dentry).
 792  */
 793 int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry)
 794 {
 795         struct dentry *result = ceph_lookup(dir, dentry, 0);
 796
 797         if (result && !IS_ERR(result)) {
 798                 /*
 799                  * We created the item, then did a lookup, and found
 800                  * it was already linked to another inode we already
 801                  * had in our cache (and thus got spliced). To not
 802                  * confuse VFS (especially when inode is a directory),
 803                  * we don't link our dentry to that inode, return an
 804                  * error instead.
 805                  *
 806                  * This event should be rare and it happens only when
 807                  * we talk to old MDS. Recent MDS does not send traceless
 808                  * reply for request that creates new inode.
 809                  */
 810                 d_drop(result);
 811                 return -ESTALE;
 812         }
 813         return PTR_ERR(result);
 814 }
 815
 816 static int ceph_mknod(struct inode *dir, struct dentry *dentry,
 817                       umode_t mode, dev_t rdev)
 818 {
 819         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 820         struct ceph_mds_client *mdsc = fsc->mdsc;
 821         struct ceph_mds_request *req;
 822         struct ceph_acls_info acls = {};
 823         int err;
 824
 825         if (ceph_snap(dir) != CEPH_NOSNAP)
 826                 return -EROFS;
 827
 828         err = ceph_pre_init_acls(dir, &mode, &acls);
 829         if (err < 0)
 830                 return err;
 831
 832         dout("mknod in dir %p dentry %p mode 0%ho rdev %d\n",
 833              dir, dentry, mode, rdev);
 834         req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_MKNOD, USE_AUTH_MDS);
 835         if (IS_ERR(req)) {
 836                 err = PTR_ERR(req);
 837                 goto out;
 838         }
 839         req->r_dentry = dget(dentry);
 840         req->r_num_caps = 2;
 841         req->r_parent = dir;
 842         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 843         req->r_args.mknod.mode = cpu_to_le32(mode);
 844         req->r_args.mknod.rdev = cpu_to_le32(rdev);
 845         req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
 846         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 847         if (acls.pagelist) {
 848                 req->r_pagelist = acls.pagelist;
 849                 acls.pagelist = NULL;
 850         }
 851         err = ceph_mdsc_do_request(mdsc, dir, req);
 852         if (!err && !req->r_reply_info.head->is_dentry)
 853                 err = ceph_handle_notrace_create(dir, dentry);
 854         ceph_mdsc_put_request(req);
 855 out:
 856         if (!err)
 857                 ceph_init_inode_acls(d_inode(dentry), &acls);
 858         else
 859                 d_drop(dentry);
 860         ceph_release_acls_info(&acls);
 861         return err;
 862 }
 863
 864 static int ceph_create(struct inode *dir, struct dentry *dentry, umode_t mode,
 865                        bool excl)
 866 {
 867         return ceph_mknod(dir, dentry, mode, 0);
 868 }
 869
 870 static int ceph_symlink(struct inode *dir, struct dentry *dentry,
 871                             const char *dest)
 872 {
 873         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 874         struct ceph_mds_client *mdsc = fsc->mdsc;
 875         struct ceph_mds_request *req;
 876         int err;
 877
 878         if (ceph_snap(dir) != CEPH_NOSNAP)
 879                 return -EROFS;
 880
 881         dout("symlink in dir %p dentry %p to '%s'\n", dir, dentry, dest);
 882         req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SYMLINK, USE_AUTH_MDS);
 883         if (IS_ERR(req)) {
 884                 err = PTR_ERR(req);
 885                 goto out;
 886         }
 887         req->r_path2 = kstrdup(dest, GFP_KERNEL);
 888         if (!req->r_path2) {
 889                 err = -ENOMEM;
 890                 ceph_mdsc_put_request(req);
 891                 goto out;
 892         }
 893         req->r_parent = dir;
 894         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 895         req->r_dentry = dget(dentry);
 896         req->r_num_caps = 2;
 897         req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
 898         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 899         err = ceph_mdsc_do_request(mdsc, dir, req);
 900         if (!err && !req->r_reply_info.head->is_dentry)
 901                 err = ceph_handle_notrace_create(dir, dentry);
 902         ceph_mdsc_put_request(req);
 903 out:
 904         if (err)
 905                 d_drop(dentry);
 906         return err;
 907 }
 908
 909 static int ceph_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
 910 {
 911         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 912         struct ceph_mds_client *mdsc = fsc->mdsc;
 913         struct ceph_mds_request *req;
 914         struct ceph_acls_info acls = {};
 915         int err = -EROFS;
 916         int op;
 917
 918         if (ceph_snap(dir) == CEPH_SNAPDIR) {
 919                 /* mkdir .snap/foo is a MKSNAP */
 920                 op = CEPH_MDS_OP_MKSNAP;
 921                 dout("mksnap dir %p snap '%pd' dn %p\n", dir,
 922                      dentry, dentry);
 923         } else if (ceph_snap(dir) == CEPH_NOSNAP) {
 924                 dout("mkdir dir %p dn %p mode 0%ho\n", dir, dentry, mode);
 925                 op = CEPH_MDS_OP_MKDIR;
 926         } else {
 927                 goto out;
 928         }
 929
 930         mode |= S_IFDIR;
 931         err = ceph_pre_init_acls(dir, &mode, &acls);
 932         if (err < 0)
 933                 goto out;
 934
 935         req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
 936         if (IS_ERR(req)) {
 937                 err = PTR_ERR(req);
 938                 goto out;
 939         }
 940
 941         req->r_dentry = dget(dentry);
 942         req->r_num_caps = 2;
 943         req->r_parent = dir;
 944         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 945         req->r_args.mkdir.mode = cpu_to_le32(mode);
 946         req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
 947         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 948         if (acls.pagelist) {
 949                 req->r_pagelist = acls.pagelist;
 950                 acls.pagelist = NULL;
 951         }
 952         err = ceph_mdsc_do_request(mdsc, dir, req);
 953         if (!err &&
 954             !req->r_reply_info.head->is_target &&
 955             !req->r_reply_info.head->is_dentry)
 956                 err = ceph_handle_notrace_create(dir, dentry);
 957         ceph_mdsc_put_request(req);
 958 out:
 959         if (!err)
 960                 ceph_init_inode_acls(d_inode(dentry), &acls);
 961         else
 962                 d_drop(dentry);
 963         ceph_release_acls_info(&acls);
 964         return err;
 965 }
 966
 967 static int ceph_link(struct dentry *old_dentry, struct inode *dir,
 968                      struct dentry *dentry)
 969 {
 970         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 971         struct ceph_mds_client *mdsc = fsc->mdsc;
 972         struct ceph_mds_request *req;
 973         int err;
 974
 975         if (ceph_snap(dir) != CEPH_NOSNAP)
 976                 return -EROFS;
 977
 978         dout("link in dir %p old_dentry %p dentry %p\n", dir,
 979              old_dentry, dentry);
 980         req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LINK, USE_AUTH_MDS);
 981         if (IS_ERR(req)) {
 982                 d_drop(dentry);
 983                 return PTR_ERR(req);
 984         }
 985         req->r_dentry = dget(dentry);
 986         req->r_num_caps = 2;
 987         req->r_old_dentry = dget(old_dentry);
 988         req->r_parent = dir;
 989         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 990         req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
 991         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 992         /* release LINK_SHARED on source inode (mds will lock it) */
 993         req->r_old_inode_drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
 994         err = ceph_mdsc_do_request(mdsc, dir, req);
 995         if (err) {
 996                 d_drop(dentry);
 997         } else if (!req->r_reply_info.head->is_dentry) {
 998                 ihold(d_inode(old_dentry));
 999                 d_instantiate(dentry, d_inode(old_dentry));
1000         }
1001         ceph_mdsc_put_request(req);
1002         return err;
1003 }
1004
1005 /*
1006  * rmdir and unlink are differ only by the metadata op code
1007  */
1008 static int ceph_unlink(struct inode *dir, struct dentry *dentry)
1009 {
1010         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
1011         struct ceph_mds_client *mdsc = fsc->mdsc;
1012         struct inode *inode = d_inode(dentry);
1013         struct ceph_mds_request *req;
1014         int err = -EROFS;
1015         int op;
1016
1017         if (ceph_snap(dir) == CEPH_SNAPDIR) {
1018                 /* rmdir .snap/foo is RMSNAP */
1019                 dout("rmsnap dir %p '%pd' dn %p\n", dir, dentry, dentry);
1020                 op = CEPH_MDS_OP_RMSNAP;
1021         } else if (ceph_snap(dir) == CEPH_NOSNAP) {
1022                 dout("unlink/rmdir dir %p dn %p inode %p\n",
1023                      dir, dentry, inode);
1024                 op = d_is_dir(dentry) ?
1025                         CEPH_MDS_OP_RMDIR : CEPH_MDS_OP_UNLINK;
1026         } else
1027                 goto out;
1028         req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
1029         if (IS_ERR(req)) {
1030                 err = PTR_ERR(req);
1031                 goto out;
1032         }
1033         req->r_dentry = dget(dentry);
1034         req->r_num_caps = 2;
1035         req->r_parent = dir;
1036         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
1037         req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
1038         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
1039         req->r_inode_drop = ceph_drop_caps_for_unlink(inode);
1040         err = ceph_mdsc_do_request(mdsc, dir, req);
1041         if (!err && !req->r_reply_info.head->is_dentry)
1042                 d_delete(dentry);
1043         ceph_mdsc_put_request(req);
1044 out:
1045         return err;
1046 }
1047
1048 static int ceph_rename(struct inode *old_dir, struct dentry *old_dentry,
1049                        struct inode *new_dir, struct dentry *new_dentry,
1050                        unsigned int flags)
1051 {
1052         struct ceph_fs_client *fsc = ceph_sb_to_client(old_dir->i_sb);
1053         struct ceph_mds_client *mdsc = fsc->mdsc;
1054         struct ceph_mds_request *req;
1055         int op = CEPH_MDS_OP_RENAME;
1056         int err;
1057
1058         if (flags)
1059                 return -EINVAL;
1060
1061         if (ceph_snap(old_dir) != ceph_snap(new_dir))
1062                 return -EXDEV;
1063         if (ceph_snap(old_dir) != CEPH_NOSNAP) {
1064                 if (old_dir == new_dir && ceph_snap(old_dir) == CEPH_SNAPDIR)
1065                         op = CEPH_MDS_OP_RENAMESNAP;
1066                 else
1067                         return -EROFS;
1068         }
1069         dout("rename dir %p dentry %p to dir %p dentry %p\n",
1070              old_dir, old_dentry, new_dir, new_dentry);
1071         req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
1072         if (IS_ERR(req))
1073                 return PTR_ERR(req);
1074         ihold(old_dir);
1075         req->r_dentry = dget(new_dentry);
1076         req->r_num_caps = 2;
1077         req->r_old_dentry = dget(old_dentry);
1078         req->r_old_dentry_dir = old_dir;
1079         req->r_parent = new_dir;
1080         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
1081         req->r_old_dentry_drop = CEPH_CAP_FILE_SHARED;
1082         req->r_old_dentry_unless = CEPH_CAP_FILE_EXCL;
1083         req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
1084         req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
1085         /* release LINK_RDCACHE on source inode (mds will lock it) */
1086         req->r_old_inode_drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
1087         if (d_really_is_positive(new_dentry)) {
1088                 req->r_inode_drop =
1089                         ceph_drop_caps_for_unlink(d_inode(new_dentry));
1090         }
1091         err = ceph_mdsc_do_request(mdsc, old_dir, req);
1092         if (!err && !req->r_reply_info.head->is_dentry) {
1093                 /*
1094                  * Normally d_move() is done by fill_trace (called by
1095                  * do_request, above).  If there is no trace, we need
1096                  * to do it here.
1097                  */
1098                 d_move(old_dentry, new_dentry);
1099         }
1100         ceph_mdsc_put_request(req);
1101         return err;
1102 }
1103
1104 /*
1105  * Ensure a dentry lease will no longer revalidate.
1106  */
1107 void ceph_invalidate_dentry_lease(struct dentry *dentry)
1108 {
1109         spin_lock(&dentry->d_lock);
1110         ceph_dentry(dentry)->time = jiffies;
1111         ceph_dentry(dentry)->lease_shared_gen = 0;
1112         spin_unlock(&dentry->d_lock);
1113 }
1114
1115 /*
1116  * Check if dentry lease is valid.  If not, delete the lease.  Try to
1117  * renew if the least is more than half up.
1118  */
1119 static int dentry_lease_is_valid(struct dentry *dentry, unsigned int flags,
1120                                  struct inode *dir)
1121 {
1122         struct ceph_dentry_info *di;
1123         struct ceph_mds_session *s;
1124         int valid = 0;
1125         u32 gen;
1126         unsigned long ttl;
1127         struct ceph_mds_session *session = NULL;
1128         u32 seq = 0;
1129
1130         spin_lock(&dentry->d_lock);
1131         di = ceph_dentry(dentry);
1132         if (di && di->lease_session) {
1133                 s = di->lease_session;
1134                 spin_lock(&s->s_gen_ttl_lock);
1135                 gen = s->s_cap_gen;
1136                 ttl = s->s_cap_ttl;
1137                 spin_unlock(&s->s_gen_ttl_lock);
1138
1139                 if (di->lease_gen == gen &&
1140                     time_before(jiffies, di->time) &&
1141                     time_before(jiffies, ttl)) {
1142                         valid = 1;
1143                         if (di->lease_renew_after &&
1144                             time_after(jiffies, di->lease_renew_after)) {
1145                                 /*
1146                                  * We should renew. If we're in RCU walk mode
1147                                  * though, we can't do that so just return
1148                                  * -ECHILD.
1149                                  */
1150                                 if (flags & LOOKUP_RCU) {
1151                                         valid = -ECHILD;
1152                                 } else {
1153                                         session = ceph_get_mds_session(s);
1154                                         seq = di->lease_seq;
1155                                         di->lease_renew_after = 0;
1156                                         di->lease_renew_from = jiffies;
1157                                 }
1158                         }
1159                 }
1160         }
1161         spin_unlock(&dentry->d_lock);
1162
1163         if (session) {
1164                 ceph_mdsc_lease_send_msg(session, dir, dentry,
1165                                          CEPH_MDS_LEASE_RENEW, seq);
1166                 ceph_put_mds_session(session);
1167         }
1168         dout("dentry_lease_is_valid - dentry %p = %d\n", dentry, valid);
1169         return valid;
1170 }
1171
1172 /*
1173  * Check if directory-wide content lease/cap is valid.
1174  */
1175 static int dir_lease_is_valid(struct inode *dir, struct dentry *dentry)
1176 {
1177         struct ceph_inode_info *ci = ceph_inode(dir);
1178         struct ceph_dentry_info *di = ceph_dentry(dentry);
1179         int valid = 0;
1180
1181         spin_lock(&ci->i_ceph_lock);
1182         if (atomic_read(&ci->i_shared_gen) == di->lease_shared_gen)
1183                 valid = __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1);
1184         spin_unlock(&ci->i_ceph_lock);
1185         dout("dir_lease_is_valid dir %p v%u dentry %p v%u = %d\n",
1186              dir, (unsigned)atomic_read(&ci->i_shared_gen),
1187              dentry, (unsigned)di->lease_shared_gen, valid);
1188         return valid;
1189 }
1190
1191 /*
1192  * Check if cached dentry can be trusted.
1193  */
1194 static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags)
1195 {
1196         int valid = 0;
1197         struct dentry *parent;
1198         struct inode *dir;
1199
1200         if (flags & LOOKUP_RCU) {
1201                 parent = READ_ONCE(dentry->d_parent);
1202                 dir = d_inode_rcu(parent);
1203                 if (!dir)
1204                         return -ECHILD;
1205         } else {
1206                 parent = dget_parent(dentry);
1207                 dir = d_inode(parent);
1208         }
1209
1210         dout("d_revalidate %p '%pd' inode %p offset %lld\n", dentry,
1211              dentry, d_inode(dentry), ceph_dentry(dentry)->offset);
1212
1213         /* always trust cached snapped dentries, snapdir dentry */
1214         if (ceph_snap(dir) != CEPH_NOSNAP) {
1215                 dout("d_revalidate %p '%pd' inode %p is SNAPPED\n", dentry,
1216                      dentry, d_inode(dentry));
1217                 valid = 1;
1218         } else if (d_really_is_positive(dentry) &&
1219                    ceph_snap(d_inode(dentry)) == CEPH_SNAPDIR) {
1220                 valid = 1;
1221         } else {
1222                 valid = dentry_lease_is_valid(dentry, flags, dir);
1223                 if (valid == -ECHILD)
1224                         return valid;
1225                 if (valid || dir_lease_is_valid(dir, dentry)) {
1226                         if (d_really_is_positive(dentry))
1227                                 valid = ceph_is_any_caps(d_inode(dentry));
1228                         else
1229                                 valid = 1;
1230                 }
1231         }
1232
1233         if (!valid) {
1234                 struct ceph_mds_client *mdsc =
1235                         ceph_sb_to_client(dir->i_sb)->mdsc;
1236                 struct ceph_mds_request *req;
1237                 int op, err;
1238                 u32 mask;
1239
1240                 if (flags & LOOKUP_RCU)
1241                         return -ECHILD;
1242
1243                 op = ceph_snap(dir) == CEPH_SNAPDIR ?
1244                         CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP;
1245                 req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS);
1246                 if (!IS_ERR(req)) {
1247                         req->r_dentry = dget(dentry);
1248                         req->r_num_caps = 2;
1249                         req->r_parent = dir;
1250
1251                         mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
1252                         if (ceph_security_xattr_wanted(dir))
1253                                 mask |= CEPH_CAP_XATTR_SHARED;
1254                         req->r_args.getattr.mask = cpu_to_le32(mask);
1255
1256                         err = ceph_mdsc_do_request(mdsc, NULL, req);
1257                         switch (err) {
1258                         case 0:
1259                                 if (d_really_is_positive(dentry) &&
1260                                     d_inode(dentry) == req->r_target_inode)
1261                                         valid = 1;
1262                                 break;
1263                         case -ENOENT:
1264                                 if (d_really_is_negative(dentry))
1265                                         valid = 1;
1266                                 /* Fallthrough */
1267                         default:
1268                                 break;
1269                         }
1270                         ceph_mdsc_put_request(req);
1271                         dout("d_revalidate %p lookup result=%d\n",
1272                              dentry, err);
1273                 }
1274         }
1275
1276         dout("d_revalidate %p %s\n", dentry, valid ? "valid" : "invalid");
1277         if (valid) {
1278                 ceph_dentry_lru_touch(dentry);
1279         } else {
1280                 ceph_dir_clear_complete(dir);
1281         }
1282
1283         if (!(flags & LOOKUP_RCU))
1284                 dput(parent);
1285         return valid;
1286 }
1287
1288 /*
1289  * Release our ceph_dentry_info.
1290  */
1291 static void ceph_d_release(struct dentry *dentry)
1292 {
1293         struct ceph_dentry_info *di = ceph_dentry(dentry);
1294
1295         dout("d_release %p\n", dentry);
1296         ceph_dentry_lru_del(dentry);
1297
1298         spin_lock(&dentry->d_lock);
1299         dentry->d_fsdata = NULL;
1300         spin_unlock(&dentry->d_lock);
1301
1302         if (di->lease_session)
1303                 ceph_put_mds_session(di->lease_session);
1304         kmem_cache_free(ceph_dentry_cachep, di);
1305 }
1306
1307 /*
1308  * When the VFS prunes a dentry from the cache, we need to clear the
1309  * complete flag on the parent directory.
1310  *
1311  * Called under dentry->d_lock.
1312  */
1313 static void ceph_d_prune(struct dentry *dentry)
1314 {
1315         struct ceph_inode_info *dir_ci;
1316         struct ceph_dentry_info *di;
1317
1318         dout("ceph_d_prune %pd %p\n", dentry, dentry);
1319
1320         /* do we have a valid parent? */
1321         if (IS_ROOT(dentry))
1322                 return;
1323
1324         /* we hold d_lock, so d_parent is stable */
1325         dir_ci = ceph_inode(d_inode(dentry->d_parent));
1326         if (dir_ci->i_vino.snap == CEPH_SNAPDIR)
1327                 return;
1328
1329         /* who calls d_delete() should also disable dcache readdir */
1330         if (d_really_is_negative(dentry))
1331                 return;
1332
1333         /* d_fsdata does not get cleared until d_release */
1334         if (!d_unhashed(dentry)) {
1335                 __ceph_dir_clear_complete(dir_ci);
1336                 return;
1337         }
1338
1339         /* Disable dcache readdir just in case that someone called d_drop()
1340          * or d_invalidate(), but MDS didn't revoke CEPH_CAP_FILE_SHARED
1341          * properly (dcache readdir is still enabled) */
1342         di = ceph_dentry(dentry);
1343         if (di->offset > 0 &&
1344             di->lease_shared_gen == atomic_read(&dir_ci->i_shared_gen))
1345                 __ceph_dir_clear_ordered(dir_ci);
1346 }
1347
1348 /*
1349  * read() on a dir.  This weird interface hack only works if mounted
1350  * with '-o dirstat'.
1351  */
1352 static ssize_t ceph_read_dir(struct file *file, char __user *buf, size_t size,
1353                              loff_t *ppos)
1354 {
1355         struct ceph_file_info *cf = file->private_data;
1356         struct inode *inode = file_inode(file);
1357         struct ceph_inode_info *ci = ceph_inode(inode);
1358         int left;
1359         const int bufsize = 1024;
1360
1361         if (!ceph_test_mount_opt(ceph_sb_to_client(inode->i_sb), DIRSTAT))
1362                 return -EISDIR;
1363
1364         if (!cf->dir_info) {
1365                 cf->dir_info = kmalloc(bufsize, GFP_KERNEL);
1366                 if (!cf->dir_info)
1367                         return -ENOMEM;
1368                 cf->dir_info_len =
1369                         snprintf(cf->dir_info, bufsize,
1370                                 "entries:   %20lld\n"
1371                                 " files:    %20lld\n"
1372                                 " subdirs:  %20lld\n"
1373                                 "rentries:  %20lld\n"
1374                                 " rfiles:   %20lld\n"
1375                                 " rsubdirs: %20lld\n"
1376                                 "rbytes:    %20lld\n"
1377                                 "rctime:    %10ld.%09ld\n",
1378                                 ci->i_files + ci->i_subdirs,
1379                                 ci->i_files,
1380                                 ci->i_subdirs,
1381                                 ci->i_rfiles + ci->i_rsubdirs,
1382                                 ci->i_rfiles,
1383                                 ci->i_rsubdirs,
1384                                 ci->i_rbytes,
1385                                 (long)ci->i_rctime.tv_sec,
1386                                 (long)ci->i_rctime.tv_nsec);
1387         }
1388
1389         if (*ppos >= cf->dir_info_len)
1390                 return 0;
1391         size = min_t(unsigned, size, cf->dir_info_len-*ppos);
1392         left = copy_to_user(buf, cf->dir_info + *ppos, size);
1393         if (left == size)
1394                 return -EFAULT;
1395         *ppos += (size - left);
1396         return size - left;
1397 }
1398
1399 /*
1400  * We maintain a private dentry LRU.
1401  *
1402  * FIXME: this needs to be changed to a per-mds lru to be useful.
1403  */
1404 void ceph_dentry_lru_add(struct dentry *dn)
1405 {
1406         struct ceph_dentry_info *di = ceph_dentry(dn);
1407         struct ceph_mds_client *mdsc;
1408
1409         dout("dentry_lru_add %p %p '%pd'\n", di, dn, dn);
1410         mdsc = ceph_sb_to_client(dn->d_sb)->mdsc;
1411         spin_lock(&mdsc->dentry_lru_lock);
1412         list_add_tail(&di->lru, &mdsc->dentry_lru);
1413         mdsc->num_dentry++;
1414         spin_unlock(&mdsc->dentry_lru_lock);
1415 }
1416
1417 void ceph_dentry_lru_touch(struct dentry *dn)
1418 {
1419         struct ceph_dentry_info *di = ceph_dentry(dn);
1420         struct ceph_mds_client *mdsc;
1421
1422         dout("dentry_lru_touch %p %p '%pd' (offset %lld)\n", di, dn, dn,
1423              di->offset);
1424         mdsc = ceph_sb_to_client(dn->d_sb)->mdsc;
1425         spin_lock(&mdsc->dentry_lru_lock);
1426         list_move_tail(&di->lru, &mdsc->dentry_lru);
1427         spin_unlock(&mdsc->dentry_lru_lock);
1428 }
1429
1430 void ceph_dentry_lru_del(struct dentry *dn)
1431 {
1432         struct ceph_dentry_info *di = ceph_dentry(dn);
1433         struct ceph_mds_client *mdsc;
1434
1435         dout("dentry_lru_del %p %p '%pd'\n", di, dn, dn);
1436         mdsc = ceph_sb_to_client(dn->d_sb)->mdsc;
1437         spin_lock(&mdsc->dentry_lru_lock);
1438         list_del_init(&di->lru);
1439         mdsc->num_dentry--;
1440         spin_unlock(&mdsc->dentry_lru_lock);
1441 }
1442
1443 /*
1444  * Return name hash for a given dentry.  This is dependent on
1445  * the parent directory's hash function.
1446  */
1447 unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn)
1448 {
1449         struct ceph_inode_info *dci = ceph_inode(dir);
1450
1451         switch (dci->i_dir_layout.dl_dir_hash) {
1452         case 0: /* for backward compat */
1453         case CEPH_STR_HASH_LINUX:
1454                 return dn->d_name.hash;
1455
1456         default:
1457                 return ceph_str_hash(dci->i_dir_layout.dl_dir_hash,
1458                                      dn->d_name.name, dn->d_name.len);
1459         }
1460 }
1461
1462 const struct file_operations ceph_dir_fops = {
1463         .read = ceph_read_dir,
1464         .iterate = ceph_readdir,
1465         .llseek = ceph_dir_llseek,
1466         .open = ceph_open,
1467         .release = ceph_release,
1468         .unlocked_ioctl = ceph_ioctl,
1469         .fsync = ceph_fsync,
1470 };
1471
1472 const struct file_operations ceph_snapdir_fops = {
1473         .iterate = ceph_readdir,
1474         .llseek = ceph_dir_llseek,
1475         .open = ceph_open,
1476         .release = ceph_release,
1477 };
1478
1479 const struct inode_operations ceph_dir_iops = {
1480         .lookup = ceph_lookup,
1481         .permission = ceph_permission,
1482         .getattr = ceph_getattr,
1483         .setattr = ceph_setattr,
1484         .listxattr = ceph_listxattr,
1485         .get_acl = ceph_get_acl,
1486         .set_acl = ceph_set_acl,
1487         .mknod = ceph_mknod,
1488         .symlink = ceph_symlink,
1489         .mkdir = ceph_mkdir,
1490         .link = ceph_link,
1491         .unlink = ceph_unlink,
1492         .rmdir = ceph_unlink,
1493         .rename = ceph_rename,
1494         .create = ceph_create,
1495         .atomic_open = ceph_atomic_open,
1496 };
1497
1498 const struct inode_operations ceph_snapdir_iops = {
1499         .lookup = ceph_lookup,
1500         .permission = ceph_permission,
1501         .getattr = ceph_getattr,
1502         .mkdir = ceph_mkdir,
1503         .rmdir = ceph_unlink,
1504         .rename = ceph_rename,
1505 };
1506
1507 const struct dentry_operations ceph_dentry_ops = {
1508         .d_revalidate = ceph_d_revalidate,
1509         .d_release = ceph_d_release,
1510         .d_prune = ceph_d_prune,
1511         .d_init = ceph_d_init,
1512 };