fs/ceph/file.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 #include <linux/ceph/ceph_debug.h>
   3 #include <linux/ceph/striper.h>
   4
   5 #include <linux/module.h>
   6 #include <linux/sched.h>
   7 #include <linux/slab.h>
   8 #include <linux/file.h>
   9 #include <linux/mount.h>
  10 #include <linux/namei.h>
  11 #include <linux/writeback.h>
  12 #include <linux/falloc.h>
  13 #include <linux/iversion.h>
  14 #include <linux/ktime.h>
  15
  16 #include "super.h"
  17 #include "mds_client.h"
  18 #include "cache.h"
  19 #include "io.h"
  20 #include "metric.h"
  21
  22 static __le32 ceph_flags_sys2wire(u32 flags)
  23 {
  24         u32 wire_flags = 0;
  25
  26         switch (flags & O_ACCMODE) {
  27         case O_RDONLY:
  28                 wire_flags |= CEPH_O_RDONLY;
  29                 break;
  30         case O_WRONLY:
  31                 wire_flags |= CEPH_O_WRONLY;
  32                 break;
  33         case O_RDWR:
  34                 wire_flags |= CEPH_O_RDWR;
  35                 break;
  36         }
  37
  38         flags &= ~O_ACCMODE;
  39
  40 #define ceph_sys2wire(a) if (flags & a) { wire_flags |= CEPH_##a; flags &= ~a; }
  41
  42         ceph_sys2wire(O_CREAT);
  43         ceph_sys2wire(O_EXCL);
  44         ceph_sys2wire(O_TRUNC);
  45         ceph_sys2wire(O_DIRECTORY);
  46         ceph_sys2wire(O_NOFOLLOW);
  47
  48 #undef ceph_sys2wire
  49
  50         if (flags)
  51                 dout("unused open flags: %x\n", flags);
  52
  53         return cpu_to_le32(wire_flags);
  54 }
  55
  56 /*
  57  * Ceph file operations
  58  *
  59  * Implement basic open/close functionality, and implement
  60  * read/write.
  61  *
  62  * We implement three modes of file I/O:
  63  *  - buffered uses the generic_file_aio_{read,write} helpers
  64  *
  65  *  - synchronous is used when there is multi-client read/write
  66  *    sharing, avoids the page cache, and synchronously waits for an
  67  *    ack from the OSD.
  68  *
  69  *  - direct io takes the variant of the sync path that references
  70  *    user pages directly.
  71  *
  72  * fsync() flushes and waits on dirty pages, but just queues metadata
  73  * for writeback: since the MDS can recover size and mtime there is no
  74  * need to wait for MDS acknowledgement.
  75  */
  76
  77 /*
  78  * How many pages to get in one call to iov_iter_get_pages().  This
  79  * determines the size of the on-stack array used as a buffer.
  80  */
  81 #define ITER_GET_BVECS_PAGES    64
  82
  83 static ssize_t __iter_get_bvecs(struct iov_iter *iter, size_t maxsize,
  84                                 struct bio_vec *bvecs)
  85 {
  86         size_t size = 0;
  87         int bvec_idx = 0;
  88
  89         if (maxsize > iov_iter_count(iter))
  90                 maxsize = iov_iter_count(iter);
  91
  92         while (size < maxsize) {
  93                 struct page *pages[ITER_GET_BVECS_PAGES];
  94                 ssize_t bytes;
  95                 size_t start;
  96                 int idx = 0;
  97
  98                 bytes = iov_iter_get_pages(iter, pages, maxsize - size,
  99                                            ITER_GET_BVECS_PAGES, &start);
 100                 if (bytes < 0)
 101                         return size ?: bytes;
 102
 103                 iov_iter_advance(iter, bytes);
 104                 size += bytes;
 105
 106                 for ( ; bytes; idx++, bvec_idx++) {
 107                         struct bio_vec bv = {
 108                                 .bv_page = pages[idx],
 109                                 .bv_len = min_t(int, bytes, PAGE_SIZE - start),
 110                                 .bv_offset = start,
 111                         };
 112
 113                         bvecs[bvec_idx] = bv;
 114                         bytes -= bv.bv_len;
 115                         start = 0;
 116                 }
 117         }
 118
 119         return size;
 120 }
 121
 122 /*
 123  * iov_iter_get_pages() only considers one iov_iter segment, no matter
 124  * what maxsize or maxpages are given.  For ITER_BVEC that is a single
 125  * page.
 126  *
 127  * Attempt to get up to @maxsize bytes worth of pages from @iter.
 128  * Return the number of bytes in the created bio_vec array, or an error.
 129  */
 130 static ssize_t iter_get_bvecs_alloc(struct iov_iter *iter, size_t maxsize,
 131                                     struct bio_vec **bvecs, int *num_bvecs)
 132 {
 133         struct bio_vec *bv;
 134         size_t orig_count = iov_iter_count(iter);
 135         ssize_t bytes;
 136         int npages;
 137
 138         iov_iter_truncate(iter, maxsize);
 139         npages = iov_iter_npages(iter, INT_MAX);
 140         iov_iter_reexpand(iter, orig_count);
 141
 142         /*
 143          * __iter_get_bvecs() may populate only part of the array -- zero it
 144          * out.
 145          */
 146         bv = kvmalloc_array(npages, sizeof(*bv), GFP_KERNEL | __GFP_ZERO);
 147         if (!bv)
 148                 return -ENOMEM;
 149
 150         bytes = __iter_get_bvecs(iter, maxsize, bv);
 151         if (bytes < 0) {
 152                 /*
 153                  * No pages were pinned -- just free the array.
 154                  */
 155                 kvfree(bv);
 156                 return bytes;
 157         }
 158
 159         *bvecs = bv;
 160         *num_bvecs = npages;
 161         return bytes;
 162 }
 163
 164 static void put_bvecs(struct bio_vec *bvecs, int num_bvecs, bool should_dirty)
 165 {
 166         int i;
 167
 168         for (i = 0; i < num_bvecs; i++) {
 169                 if (bvecs[i].bv_page) {
 170                         if (should_dirty)
 171                                 set_page_dirty_lock(bvecs[i].bv_page);
 172                         put_page(bvecs[i].bv_page);
 173                 }
 174         }
 175         kvfree(bvecs);
 176 }
 177
 178 /*
 179  * Prepare an open request.  Preallocate ceph_cap to avoid an
 180  * inopportune ENOMEM later.
 181  */
 182 static struct ceph_mds_request *
 183 prepare_open_request(struct super_block *sb, int flags, int create_mode)
 184 {
 185         struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(sb);
 186         struct ceph_mds_request *req;
 187         int want_auth = USE_ANY_MDS;
 188         int op = (flags & O_CREAT) ? CEPH_MDS_OP_CREATE : CEPH_MDS_OP_OPEN;
 189
 190         if (flags & (O_WRONLY|O_RDWR|O_CREAT|O_TRUNC))
 191                 want_auth = USE_AUTH_MDS;
 192
 193         req = ceph_mdsc_create_request(mdsc, op, want_auth);
 194         if (IS_ERR(req))
 195                 goto out;
 196         req->r_fmode = ceph_flags_to_mode(flags);
 197         req->r_args.open.flags = ceph_flags_sys2wire(flags);
 198         req->r_args.open.mode = cpu_to_le32(create_mode);
 199 out:
 200         return req;
 201 }
 202
 203 static int ceph_init_file_info(struct inode *inode, struct file *file,
 204                                         int fmode, bool isdir)
 205 {
 206         struct ceph_inode_info *ci = ceph_inode(inode);
 207         struct ceph_file_info *fi;
 208
 209         dout("%s %p %p 0%o (%s)\n", __func__, inode, file,
 210                         inode->i_mode, isdir ? "dir" : "regular");
 211         BUG_ON(inode->i_fop->release != ceph_release);
 212
 213         if (isdir) {
 214                 struct ceph_dir_file_info *dfi =
 215                         kmem_cache_zalloc(ceph_dir_file_cachep, GFP_KERNEL);
 216                 if (!dfi)
 217                         return -ENOMEM;
 218
 219                 file->private_data = dfi;
 220                 fi = &dfi->file_info;
 221                 dfi->next_offset = 2;
 222                 dfi->readdir_cache_idx = -1;
 223         } else {
 224                 fi = kmem_cache_zalloc(ceph_file_cachep, GFP_KERNEL);
 225                 if (!fi)
 226                         return -ENOMEM;
 227
 228                 file->private_data = fi;
 229         }
 230
 231         ceph_get_fmode(ci, fmode, 1);
 232         fi->fmode = fmode;
 233
 234         spin_lock_init(&fi->rw_contexts_lock);
 235         INIT_LIST_HEAD(&fi->rw_contexts);
 236         fi->meta_err = errseq_sample(&ci->i_meta_err);
 237         fi->filp_gen = READ_ONCE(ceph_inode_to_client(inode)->filp_gen);
 238
 239         return 0;
 240 }
 241
 242 /*
 243  * initialize private struct file data.
 244  * if we fail, clean up by dropping fmode reference on the ceph_inode
 245  */
 246 static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
 247 {
 248         int ret = 0;
 249
 250         switch (inode->i_mode & S_IFMT) {
 251         case S_IFREG:
 252                 ceph_fscache_register_inode_cookie(inode);
 253                 ceph_fscache_file_set_cookie(inode, file);
 254                 fallthrough;
 255         case S_IFDIR:
 256                 ret = ceph_init_file_info(inode, file, fmode,
 257                                                 S_ISDIR(inode->i_mode));
 258                 break;
 259
 260         case S_IFLNK:
 261                 dout("init_file %p %p 0%o (symlink)\n", inode, file,
 262                      inode->i_mode);
 263                 break;
 264
 265         default:
 266                 dout("init_file %p %p 0%o (special)\n", inode, file,
 267                      inode->i_mode);
 268                 /*
 269                  * we need to drop the open ref now, since we don't
 270                  * have .release set to ceph_release.
 271                  */
 272                 BUG_ON(inode->i_fop->release == ceph_release);
 273
 274                 /* call the proper open fop */
 275                 ret = inode->i_fop->open(inode, file);
 276         }
 277         return ret;
 278 }
 279
 280 /*
 281  * try renew caps after session gets killed.
 282  */
 283 int ceph_renew_caps(struct inode *inode, int fmode)
 284 {
 285         struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
 286         struct ceph_inode_info *ci = ceph_inode(inode);
 287         struct ceph_mds_request *req;
 288         int err, flags, wanted;
 289
 290         spin_lock(&ci->i_ceph_lock);
 291         __ceph_touch_fmode(ci, mdsc, fmode);
 292         wanted = __ceph_caps_file_wanted(ci);
 293         if (__ceph_is_any_real_caps(ci) &&
 294             (!(wanted & CEPH_CAP_ANY_WR) || ci->i_auth_cap)) {
 295                 int issued = __ceph_caps_issued(ci, NULL);
 296                 spin_unlock(&ci->i_ceph_lock);
 297                 dout("renew caps %p want %s issued %s updating mds_wanted\n",
 298                      inode, ceph_cap_string(wanted), ceph_cap_string(issued));
 299                 ceph_check_caps(ci, 0, NULL);
 300                 return 0;
 301         }
 302         spin_unlock(&ci->i_ceph_lock);
 303
 304         flags = 0;
 305         if ((wanted & CEPH_CAP_FILE_RD) && (wanted & CEPH_CAP_FILE_WR))
 306                 flags = O_RDWR;
 307         else if (wanted & CEPH_CAP_FILE_RD)
 308                 flags = O_RDONLY;
 309         else if (wanted & CEPH_CAP_FILE_WR)
 310                 flags = O_WRONLY;
 311 #ifdef O_LAZY
 312         if (wanted & CEPH_CAP_FILE_LAZYIO)
 313                 flags |= O_LAZY;
 314 #endif
 315
 316         req = prepare_open_request(inode->i_sb, flags, 0);
 317         if (IS_ERR(req)) {
 318                 err = PTR_ERR(req);
 319                 goto out;
 320         }
 321
 322         req->r_inode = inode;
 323         ihold(inode);
 324         req->r_num_caps = 1;
 325
 326         err = ceph_mdsc_do_request(mdsc, NULL, req);
 327         ceph_mdsc_put_request(req);
 328 out:
 329         dout("renew caps %p open result=%d\n", inode, err);
 330         return err < 0 ? err : 0;
 331 }
 332
 333 /*
 334  * If we already have the requisite capabilities, we can satisfy
 335  * the open request locally (no need to request new caps from the
 336  * MDS).  We do, however, need to inform the MDS (asynchronously)
 337  * if our wanted caps set expands.
 338  */
 339 int ceph_open(struct inode *inode, struct file *file)
 340 {
 341         struct ceph_inode_info *ci = ceph_inode(inode);
 342         struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
 343         struct ceph_mds_client *mdsc = fsc->mdsc;
 344         struct ceph_mds_request *req;
 345         struct ceph_file_info *fi = file->private_data;
 346         int err;
 347         int flags, fmode, wanted;
 348
 349         if (fi) {
 350                 dout("open file %p is already opened\n", file);
 351                 return 0;
 352         }
 353
 354         /* filter out O_CREAT|O_EXCL; vfs did that already.  yuck. */
 355         flags = file->f_flags & ~(O_CREAT|O_EXCL);
 356         if (S_ISDIR(inode->i_mode))
 357                 flags = O_DIRECTORY;  /* mds likes to know */
 358
 359         dout("open inode %p ino %llx.%llx file %p flags %d (%d)\n", inode,
 360              ceph_vinop(inode), file, flags, file->f_flags);
 361         fmode = ceph_flags_to_mode(flags);
 362         wanted = ceph_caps_for_mode(fmode);
 363
 364         /* snapped files are read-only */
 365         if (ceph_snap(inode) != CEPH_NOSNAP && (file->f_mode & FMODE_WRITE))
 366                 return -EROFS;
 367
 368         /* trivially open snapdir */
 369         if (ceph_snap(inode) == CEPH_SNAPDIR) {
 370                 return ceph_init_file(inode, file, fmode);
 371         }
 372
 373         /*
 374          * No need to block if we have caps on the auth MDS (for
 375          * write) or any MDS (for read).  Update wanted set
 376          * asynchronously.
 377          */
 378         spin_lock(&ci->i_ceph_lock);
 379         if (__ceph_is_any_real_caps(ci) &&
 380             (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) {
 381                 int mds_wanted = __ceph_caps_mds_wanted(ci, true);
 382                 int issued = __ceph_caps_issued(ci, NULL);
 383
 384                 dout("open %p fmode %d want %s issued %s using existing\n",
 385                      inode, fmode, ceph_cap_string(wanted),
 386                      ceph_cap_string(issued));
 387                 __ceph_touch_fmode(ci, mdsc, fmode);
 388                 spin_unlock(&ci->i_ceph_lock);
 389
 390                 /* adjust wanted? */
 391                 if ((issued & wanted) != wanted &&
 392                     (mds_wanted & wanted) != wanted &&
 393                     ceph_snap(inode) != CEPH_SNAPDIR)
 394                         ceph_check_caps(ci, 0, NULL);
 395
 396                 return ceph_init_file(inode, file, fmode);
 397         } else if (ceph_snap(inode) != CEPH_NOSNAP &&
 398                    (ci->i_snap_caps & wanted) == wanted) {
 399                 __ceph_touch_fmode(ci, mdsc, fmode);
 400                 spin_unlock(&ci->i_ceph_lock);
 401                 return ceph_init_file(inode, file, fmode);
 402         }
 403
 404         spin_unlock(&ci->i_ceph_lock);
 405
 406         dout("open fmode %d wants %s\n", fmode, ceph_cap_string(wanted));
 407         req = prepare_open_request(inode->i_sb, flags, 0);
 408         if (IS_ERR(req)) {
 409                 err = PTR_ERR(req);
 410                 goto out;
 411         }
 412         req->r_inode = inode;
 413         ihold(inode);
 414
 415         req->r_num_caps = 1;
 416         err = ceph_mdsc_do_request(mdsc, NULL, req);
 417         if (!err)
 418                 err = ceph_init_file(inode, file, req->r_fmode);
 419         ceph_mdsc_put_request(req);
 420         dout("open result=%d on %llx.%llx\n", err, ceph_vinop(inode));
 421 out:
 422         return err;
 423 }
 424
 425 /* Clone the layout from a synchronous create, if the dir now has Dc caps */
 426 static void
 427 cache_file_layout(struct inode *dst, struct inode *src)
 428 {
 429         struct ceph_inode_info *cdst = ceph_inode(dst);
 430         struct ceph_inode_info *csrc = ceph_inode(src);
 431
 432         spin_lock(&cdst->i_ceph_lock);
 433         if ((__ceph_caps_issued(cdst, NULL) & CEPH_CAP_DIR_CREATE) &&
 434             !ceph_file_layout_is_valid(&cdst->i_cached_layout)) {
 435                 memcpy(&cdst->i_cached_layout, &csrc->i_layout,
 436                         sizeof(cdst->i_cached_layout));
 437                 rcu_assign_pointer(cdst->i_cached_layout.pool_ns,
 438                                    ceph_try_get_string(csrc->i_layout.pool_ns));
 439         }
 440         spin_unlock(&cdst->i_ceph_lock);
 441 }
 442
 443 /*
 444  * Try to set up an async create. We need caps, a file layout, and inode number,
 445  * and either a lease on the dentry or complete dir info. If any of those
 446  * criteria are not satisfied, then return false and the caller can go
 447  * synchronous.
 448  */
 449 static int try_prep_async_create(struct inode *dir, struct dentry *dentry,
 450                                  struct ceph_file_layout *lo, u64 *pino)
 451 {
 452         struct ceph_inode_info *ci = ceph_inode(dir);
 453         struct ceph_dentry_info *di = ceph_dentry(dentry);
 454         int got = 0, want = CEPH_CAP_FILE_EXCL | CEPH_CAP_DIR_CREATE;
 455         u64 ino;
 456
 457         spin_lock(&ci->i_ceph_lock);
 458         /* No auth cap means no chance for Dc caps */
 459         if (!ci->i_auth_cap)
 460                 goto no_async;
 461
 462         /* Any delegated inos? */
 463         if (xa_empty(&ci->i_auth_cap->session->s_delegated_inos))
 464                 goto no_async;
 465
 466         if (!ceph_file_layout_is_valid(&ci->i_cached_layout))
 467                 goto no_async;
 468
 469         if ((__ceph_caps_issued(ci, NULL) & want) != want)
 470                 goto no_async;
 471
 472         if (d_in_lookup(dentry)) {
 473                 if (!__ceph_dir_is_complete(ci))
 474                         goto no_async;
 475                 spin_lock(&dentry->d_lock);
 476                 di->lease_shared_gen = atomic_read(&ci->i_shared_gen);
 477                 spin_unlock(&dentry->d_lock);
 478         } else if (atomic_read(&ci->i_shared_gen) !=
 479                    READ_ONCE(di->lease_shared_gen)) {
 480                 goto no_async;
 481         }
 482
 483         ino = ceph_get_deleg_ino(ci->i_auth_cap->session);
 484         if (!ino)
 485                 goto no_async;
 486
 487         *pino = ino;
 488         ceph_take_cap_refs(ci, want, false);
 489         memcpy(lo, &ci->i_cached_layout, sizeof(*lo));
 490         rcu_assign_pointer(lo->pool_ns,
 491                            ceph_try_get_string(ci->i_cached_layout.pool_ns));
 492         got = want;
 493 no_async:
 494         spin_unlock(&ci->i_ceph_lock);
 495         return got;
 496 }
 497
 498 static void restore_deleg_ino(struct inode *dir, u64 ino)
 499 {
 500         struct ceph_inode_info *ci = ceph_inode(dir);
 501         struct ceph_mds_session *s = NULL;
 502
 503         spin_lock(&ci->i_ceph_lock);
 504         if (ci->i_auth_cap)
 505                 s = ceph_get_mds_session(ci->i_auth_cap->session);
 506         spin_unlock(&ci->i_ceph_lock);
 507         if (s) {
 508                 int err = ceph_restore_deleg_ino(s, ino);
 509                 if (err)
 510                         pr_warn("ceph: unable to restore delegated ino 0x%llx to session: %d\n",
 511                                 ino, err);
 512                 ceph_put_mds_session(s);
 513         }
 514 }
 515
 516 static void ceph_async_create_cb(struct ceph_mds_client *mdsc,
 517                                  struct ceph_mds_request *req)
 518 {
 519         int result = req->r_err ? req->r_err :
 520                         le32_to_cpu(req->r_reply_info.head->result);
 521
 522         if (result == -EJUKEBOX)
 523                 goto out;
 524
 525         mapping_set_error(req->r_parent->i_mapping, result);
 526
 527         if (result) {
 528                 struct dentry *dentry = req->r_dentry;
 529                 int pathlen = 0;
 530                 u64 base = 0;
 531                 char *path = ceph_mdsc_build_path(req->r_dentry, &pathlen,
 532                                                   &base, 0);
 533
 534                 ceph_dir_clear_complete(req->r_parent);
 535                 if (!d_unhashed(dentry))
 536                         d_drop(dentry);
 537
 538                 /* FIXME: start returning I/O errors on all accesses? */
 539                 pr_warn("ceph: async create failure path=(%llx)%s result=%d!\n",
 540                         base, IS_ERR(path) ? "<<bad>>" : path, result);
 541                 ceph_mdsc_free_path(path, pathlen);
 542         }
 543
 544         if (req->r_target_inode) {
 545                 struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
 546                 u64 ino = ceph_vino(req->r_target_inode).ino;
 547
 548                 if (req->r_deleg_ino != ino)
 549                         pr_warn("%s: inode number mismatch! err=%d deleg_ino=0x%llx target=0x%llx\n",
 550                                 __func__, req->r_err, req->r_deleg_ino, ino);
 551                 mapping_set_error(req->r_target_inode->i_mapping, result);
 552
 553                 spin_lock(&ci->i_ceph_lock);
 554                 if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
 555                         ci->i_ceph_flags &= ~CEPH_I_ASYNC_CREATE;
 556                         wake_up_bit(&ci->i_ceph_flags, CEPH_ASYNC_CREATE_BIT);
 557                 }
 558                 ceph_kick_flushing_inode_caps(req->r_session, ci);
 559                 spin_unlock(&ci->i_ceph_lock);
 560         } else {
 561                 pr_warn("%s: no req->r_target_inode for 0x%llx\n", __func__,
 562                         req->r_deleg_ino);
 563         }
 564 out:
 565         ceph_mdsc_release_dir_caps(req);
 566 }
 567
 568 static int ceph_finish_async_create(struct inode *dir, struct dentry *dentry,
 569                                     struct file *file, umode_t mode,
 570                                     struct ceph_mds_request *req,
 571                                     struct ceph_acl_sec_ctx *as_ctx,
 572                                     struct ceph_file_layout *lo)
 573 {
 574         int ret;
 575         char xattr_buf[4];
 576         struct ceph_mds_reply_inode in = { };
 577         struct ceph_mds_reply_info_in iinfo = { .in = &in };
 578         struct ceph_inode_info *ci = ceph_inode(dir);
 579         struct inode *inode;
 580         struct timespec64 now;
 581         struct ceph_vino vino = { .ino = req->r_deleg_ino,
 582                                   .snap = CEPH_NOSNAP };
 583
 584         ktime_get_real_ts64(&now);
 585
 586         inode = ceph_get_inode(dentry->d_sb, vino);
 587         if (IS_ERR(inode))
 588                 return PTR_ERR(inode);
 589
 590         iinfo.inline_version = CEPH_INLINE_NONE;
 591         iinfo.change_attr = 1;
 592         ceph_encode_timespec64(&iinfo.btime, &now);
 593
 594         iinfo.xattr_len = ARRAY_SIZE(xattr_buf);
 595         iinfo.xattr_data = xattr_buf;
 596         memset(iinfo.xattr_data, 0, iinfo.xattr_len);
 597
 598         in.ino = cpu_to_le64(vino.ino);
 599         in.snapid = cpu_to_le64(CEPH_NOSNAP);
 600         in.version = cpu_to_le64(1);    // ???
 601         in.cap.caps = in.cap.wanted = cpu_to_le32(CEPH_CAP_ALL_FILE);
 602         in.cap.cap_id = cpu_to_le64(1);
 603         in.cap.realm = cpu_to_le64(ci->i_snap_realm->ino);
 604         in.cap.flags = CEPH_CAP_FLAG_AUTH;
 605         in.ctime = in.mtime = in.atime = iinfo.btime;
 606         in.mode = cpu_to_le32((u32)mode);
 607         in.truncate_seq = cpu_to_le32(1);
 608         in.truncate_size = cpu_to_le64(-1ULL);
 609         in.xattr_version = cpu_to_le64(1);
 610         in.uid = cpu_to_le32(from_kuid(&init_user_ns, current_fsuid()));
 611         in.gid = cpu_to_le32(from_kgid(&init_user_ns, dir->i_mode & S_ISGID ?
 612                                 dir->i_gid : current_fsgid()));
 613         in.nlink = cpu_to_le32(1);
 614         in.max_size = cpu_to_le64(lo->stripe_unit);
 615
 616         ceph_file_layout_to_legacy(lo, &in.layout);
 617
 618         ret = ceph_fill_inode(inode, NULL, &iinfo, NULL, req->r_session,
 619                               req->r_fmode, NULL);
 620         if (ret) {
 621                 dout("%s failed to fill inode: %d\n", __func__, ret);
 622                 ceph_dir_clear_complete(dir);
 623                 if (!d_unhashed(dentry))
 624                         d_drop(dentry);
 625                 if (inode->i_state & I_NEW)
 626                         discard_new_inode(inode);
 627         } else {
 628                 struct dentry *dn;
 629
 630                 dout("%s d_adding new inode 0x%llx to 0x%llx/%s\n", __func__,
 631                         vino.ino, ceph_ino(dir), dentry->d_name.name);
 632                 ceph_dir_clear_ordered(dir);
 633                 ceph_init_inode_acls(inode, as_ctx);
 634                 if (inode->i_state & I_NEW) {
 635                         /*
 636                          * If it's not I_NEW, then someone created this before
 637                          * we got here. Assume the server is aware of it at
 638                          * that point and don't worry about setting
 639                          * CEPH_I_ASYNC_CREATE.
 640                          */
 641                         ceph_inode(inode)->i_ceph_flags = CEPH_I_ASYNC_CREATE;
 642                         unlock_new_inode(inode);
 643                 }
 644                 if (d_in_lookup(dentry) || d_really_is_negative(dentry)) {
 645                         if (!d_unhashed(dentry))
 646                                 d_drop(dentry);
 647                         dn = d_splice_alias(inode, dentry);
 648                         WARN_ON_ONCE(dn && dn != dentry);
 649                 }
 650                 file->f_mode |= FMODE_CREATED;
 651                 ret = finish_open(file, dentry, ceph_open);
 652         }
 653         return ret;
 654 }
 655
 656 /*
 657  * Do a lookup + open with a single request.  If we get a non-existent
 658  * file or symlink, return 1 so the VFS can retry.
 659  */
 660 int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
 661                      struct file *file, unsigned flags, umode_t mode)
 662 {
 663         struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 664         struct ceph_mds_client *mdsc = fsc->mdsc;
 665         struct ceph_mds_request *req;
 666         struct dentry *dn;
 667         struct ceph_acl_sec_ctx as_ctx = {};
 668         bool try_async = ceph_test_mount_opt(fsc, ASYNC_DIROPS);
 669         int mask;
 670         int err;
 671
 672         dout("atomic_open %p dentry %p '%pd' %s flags %d mode 0%o\n",
 673              dir, dentry, dentry,
 674              d_unhashed(dentry) ? "unhashed" : "hashed", flags, mode);
 675
 676         if (dentry->d_name.len > NAME_MAX)
 677                 return -ENAMETOOLONG;
 678
 679         if (flags & O_CREAT) {
 680                 if (ceph_quota_is_max_files_exceeded(dir))
 681                         return -EDQUOT;
 682                 err = ceph_pre_init_acls(dir, &mode, &as_ctx);
 683                 if (err < 0)
 684                         return err;
 685                 err = ceph_security_init_secctx(dentry, mode, &as_ctx);
 686                 if (err < 0)
 687                         goto out_ctx;
 688         } else if (!d_in_lookup(dentry)) {
 689                 /* If it's not being looked up, it's negative */
 690                 return -ENOENT;
 691         }
 692 retry:
 693         /* do the open */
 694         req = prepare_open_request(dir->i_sb, flags, mode);
 695         if (IS_ERR(req)) {
 696                 err = PTR_ERR(req);
 697                 goto out_ctx;
 698         }
 699         req->r_dentry = dget(dentry);
 700         req->r_num_caps = 2;
 701         mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
 702         if (ceph_security_xattr_wanted(dir))
 703                 mask |= CEPH_CAP_XATTR_SHARED;
 704         req->r_args.open.mask = cpu_to_le32(mask);
 705         req->r_parent = dir;
 706
 707         if (flags & O_CREAT) {
 708                 struct ceph_file_layout lo;
 709
 710                 req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
 711                 req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 712                 if (as_ctx.pagelist) {
 713                         req->r_pagelist = as_ctx.pagelist;
 714                         as_ctx.pagelist = NULL;
 715                 }
 716                 if (try_async &&
 717                     (req->r_dir_caps =
 718                       try_prep_async_create(dir, dentry, &lo,
 719                                             &req->r_deleg_ino))) {
 720                         set_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags);
 721                         req->r_args.open.flags |= cpu_to_le32(CEPH_O_EXCL);
 722                         req->r_callback = ceph_async_create_cb;
 723                         err = ceph_mdsc_submit_request(mdsc, dir, req);
 724                         if (!err) {
 725                                 err = ceph_finish_async_create(dir, dentry,
 726                                                         file, mode, req,
 727                                                         &as_ctx, &lo);
 728                         } else if (err == -EJUKEBOX) {
 729                                 restore_deleg_ino(dir, req->r_deleg_ino);
 730                                 ceph_mdsc_put_request(req);
 731                                 try_async = false;
 732                                 goto retry;
 733                         }
 734                         goto out_req;
 735                 }
 736         }
 737
 738         set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 739         err = ceph_mdsc_do_request(mdsc,
 740                                    (flags & (O_CREAT|O_TRUNC)) ? dir : NULL,
 741                                    req);
 742         err = ceph_handle_snapdir(req, dentry, err);
 743         if (err)
 744                 goto out_req;
 745
 746         if ((flags & O_CREAT) && !req->r_reply_info.head->is_dentry)
 747                 err = ceph_handle_notrace_create(dir, dentry);
 748
 749         if (d_in_lookup(dentry)) {
 750                 dn = ceph_finish_lookup(req, dentry, err);
 751                 if (IS_ERR(dn))
 752                         err = PTR_ERR(dn);
 753         } else {
 754                 /* we were given a hashed negative dentry */
 755                 dn = NULL;
 756         }
 757         if (err)
 758                 goto out_req;
 759         if (dn || d_really_is_negative(dentry) || d_is_symlink(dentry)) {
 760                 /* make vfs retry on splice, ENOENT, or symlink */
 761                 dout("atomic_open finish_no_open on dn %p\n", dn);
 762                 err = finish_no_open(file, dn);
 763         } else {
 764                 dout("atomic_open finish_open on dn %p\n", dn);
 765                 if (req->r_op == CEPH_MDS_OP_CREATE && req->r_reply_info.has_create_ino) {
 766                         struct inode *newino = d_inode(dentry);
 767
 768                         cache_file_layout(dir, newino);
 769                         ceph_init_inode_acls(newino, &as_ctx);
 770                         file->f_mode |= FMODE_CREATED;
 771                 }
 772                 err = finish_open(file, dentry, ceph_open);
 773         }
 774 out_req:
 775         ceph_mdsc_put_request(req);
 776 out_ctx:
 777         ceph_release_acl_sec_ctx(&as_ctx);
 778         dout("atomic_open result=%d\n", err);
 779         return err;
 780 }
 781
 782 int ceph_release(struct inode *inode, struct file *file)
 783 {
 784         struct ceph_inode_info *ci = ceph_inode(inode);
 785
 786         if (S_ISDIR(inode->i_mode)) {
 787                 struct ceph_dir_file_info *dfi = file->private_data;
 788                 dout("release inode %p dir file %p\n", inode, file);
 789                 WARN_ON(!list_empty(&dfi->file_info.rw_contexts));
 790
 791                 ceph_put_fmode(ci, dfi->file_info.fmode, 1);
 792
 793                 if (dfi->last_readdir)
 794                         ceph_mdsc_put_request(dfi->last_readdir);
 795                 kfree(dfi->last_name);
 796                 kfree(dfi->dir_info);
 797                 kmem_cache_free(ceph_dir_file_cachep, dfi);
 798         } else {
 799                 struct ceph_file_info *fi = file->private_data;
 800                 dout("release inode %p regular file %p\n", inode, file);
 801                 WARN_ON(!list_empty(&fi->rw_contexts));
 802
 803                 ceph_put_fmode(ci, fi->fmode, 1);
 804
 805                 kmem_cache_free(ceph_file_cachep, fi);
 806         }
 807
 808         /* wake up anyone waiting for caps on this inode */
 809         wake_up_all(&ci->i_cap_wq);
 810         return 0;
 811 }
 812
 813 enum {
 814         HAVE_RETRIED = 1,
 815         CHECK_EOF =    2,
 816         READ_INLINE =  3,
 817 };
 818
 819 /*
 820  * Completely synchronous read and write methods.  Direct from __user
 821  * buffer to osd, or directly to user pages (if O_DIRECT).
 822  *
 823  * If the read spans object boundary, just do multiple reads.  (That's not
 824  * atomic, but good enough for now.)
 825  *
 826  * If we get a short result from the OSD, check against i_size; we need to
 827  * only return a short read to the caller if we hit EOF.
 828  */
 829 static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *to,
 830                               int *retry_op)
 831 {
 832         struct file *file = iocb->ki_filp;
 833         struct inode *inode = file_inode(file);
 834         struct ceph_inode_info *ci = ceph_inode(inode);
 835         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 836         struct ceph_osd_client *osdc = &fsc->client->osdc;
 837         ssize_t ret;
 838         u64 off = iocb->ki_pos;
 839         u64 len = iov_iter_count(to);
 840
 841         dout("sync_read on file %p %llu~%u %s\n", file, off, (unsigned)len,
 842              (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
 843
 844         if (!len)
 845                 return 0;
 846         /*
 847          * flush any page cache pages in this range.  this
 848          * will make concurrent normal and sync io slow,
 849          * but it will at least behave sensibly when they are
 850          * in sequence.
 851          */
 852         ret = filemap_write_and_wait_range(inode->i_mapping,
 853                                            off, off + len - 1);
 854         if (ret < 0)
 855                 return ret;
 856
 857         ret = 0;
 858         while ((len = iov_iter_count(to)) > 0) {
 859                 struct ceph_osd_request *req;
 860                 struct page **pages;
 861                 int num_pages;
 862                 size_t page_off;
 863                 u64 i_size;
 864                 bool more;
 865                 int idx;
 866                 size_t left;
 867
 868                 req = ceph_osdc_new_request(osdc, &ci->i_layout,
 869                                         ci->i_vino, off, &len, 0, 1,
 870                                         CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
 871                                         NULL, ci->i_truncate_seq,
 872                                         ci->i_truncate_size, false);
 873                 if (IS_ERR(req)) {
 874                         ret = PTR_ERR(req);
 875                         break;
 876                 }
 877
 878                 more = len < iov_iter_count(to);
 879
 880                 num_pages = calc_pages_for(off, len);
 881                 page_off = off & ~PAGE_MASK;
 882                 pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
 883                 if (IS_ERR(pages)) {
 884                         ceph_osdc_put_request(req);
 885                         ret = PTR_ERR(pages);
 886                         break;
 887                 }
 888
 889                 osd_req_op_extent_osd_data_pages(req, 0, pages, len, page_off,
 890                                                  false, false);
 891                 ret = ceph_osdc_start_request(osdc, req, false);
 892                 if (!ret)
 893                         ret = ceph_osdc_wait_request(osdc, req);
 894
 895                 ceph_update_read_latency(&fsc->mdsc->metric,
 896                                          req->r_start_latency,
 897                                          req->r_end_latency,
 898                                          ret);
 899
 900                 ceph_osdc_put_request(req);
 901
 902                 i_size = i_size_read(inode);
 903                 dout("sync_read %llu~%llu got %zd i_size %llu%s\n",
 904                      off, len, ret, i_size, (more ? " MORE" : ""));
 905
 906                 if (ret == -ENOENT)
 907                         ret = 0;
 908                 if (ret >= 0 && ret < len && (off + ret < i_size)) {
 909                         int zlen = min(len - ret, i_size - off - ret);
 910                         int zoff = page_off + ret;
 911                         dout("sync_read zero gap %llu~%llu\n",
 912                              off + ret, off + ret + zlen);
 913                         ceph_zero_page_vector_range(zoff, zlen, pages);
 914                         ret += zlen;
 915                 }
 916
 917                 idx = 0;
 918                 left = ret > 0 ? ret : 0;
 919                 while (left > 0) {
 920                         size_t len, copied;
 921                         page_off = off & ~PAGE_MASK;
 922                         len = min_t(size_t, left, PAGE_SIZE - page_off);
 923                         SetPageUptodate(pages[idx]);
 924                         copied = copy_page_to_iter(pages[idx++],
 925                                                    page_off, len, to);
 926                         off += copied;
 927                         left -= copied;
 928                         if (copied < len) {
 929                                 ret = -EFAULT;
 930                                 break;
 931                         }
 932                 }
 933                 ceph_release_page_vector(pages, num_pages);
 934
 935                 if (ret < 0) {
 936                         if (ret == -EBLOCKLISTED)
 937                                 fsc->blocklisted = true;
 938                         break;
 939                 }
 940
 941                 if (off >= i_size || !more)
 942                         break;
 943         }
 944
 945         if (off > iocb->ki_pos) {
 946                 if (ret >= 0 &&
 947                     iov_iter_count(to) > 0 && off >= i_size_read(inode))
 948                         *retry_op = CHECK_EOF;
 949                 ret = off - iocb->ki_pos;
 950                 iocb->ki_pos = off;
 951         }
 952
 953         dout("sync_read result %zd retry_op %d\n", ret, *retry_op);
 954         return ret;
 955 }
 956
 957 struct ceph_aio_request {
 958         struct kiocb *iocb;
 959         size_t total_len;
 960         bool write;
 961         bool should_dirty;
 962         int error;
 963         struct list_head osd_reqs;
 964         unsigned num_reqs;
 965         atomic_t pending_reqs;
 966         struct timespec64 mtime;
 967         struct ceph_cap_flush *prealloc_cf;
 968 };
 969
 970 struct ceph_aio_work {
 971         struct work_struct work;
 972         struct ceph_osd_request *req;
 973 };
 974
 975 static void ceph_aio_retry_work(struct work_struct *work);
 976
 977 static void ceph_aio_complete(struct inode *inode,
 978                               struct ceph_aio_request *aio_req)
 979 {
 980         struct ceph_inode_info *ci = ceph_inode(inode);
 981         int ret;
 982
 983         if (!atomic_dec_and_test(&aio_req->pending_reqs))
 984                 return;
 985
 986         if (aio_req->iocb->ki_flags & IOCB_DIRECT)
 987                 inode_dio_end(inode);
 988
 989         ret = aio_req->error;
 990         if (!ret)
 991                 ret = aio_req->total_len;
 992
 993         dout("ceph_aio_complete %p rc %d\n", inode, ret);
 994
 995         if (ret >= 0 && aio_req->write) {
 996                 int dirty;
 997
 998                 loff_t endoff = aio_req->iocb->ki_pos + aio_req->total_len;
 999                 if (endoff > i_size_read(inode)) {
1000                         if (ceph_inode_set_size(inode, endoff))
1001                                 ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
1002                 }
1003
1004                 spin_lock(&ci->i_ceph_lock);
1005                 ci->i_inline_version = CEPH_INLINE_NONE;
1006                 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
1007                                                &aio_req->prealloc_cf);
1008                 spin_unlock(&ci->i_ceph_lock);
1009                 if (dirty)
1010                         __mark_inode_dirty(inode, dirty);
1011
1012         }
1013
1014         ceph_put_cap_refs(ci, (aio_req->write ? CEPH_CAP_FILE_WR :
1015                                                 CEPH_CAP_FILE_RD));
1016
1017         aio_req->iocb->ki_complete(aio_req->iocb, ret, 0);
1018
1019         ceph_free_cap_flush(aio_req->prealloc_cf);
1020         kfree(aio_req);
1021 }
1022
1023 static void ceph_aio_complete_req(struct ceph_osd_request *req)
1024 {
1025         int rc = req->r_result;
1026         struct inode *inode = req->r_inode;
1027         struct ceph_aio_request *aio_req = req->r_priv;
1028         struct ceph_osd_data *osd_data = osd_req_op_extent_osd_data(req, 0);
1029         struct ceph_client_metric *metric = &ceph_sb_to_mdsc(inode->i_sb)->metric;
1030
1031         BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_BVECS);
1032         BUG_ON(!osd_data->num_bvecs);
1033
1034         dout("ceph_aio_complete_req %p rc %d bytes %u\n",
1035              inode, rc, osd_data->bvec_pos.iter.bi_size);
1036
1037         /* r_start_latency == 0 means the request was not submitted */
1038         if (req->r_start_latency) {
1039                 if (aio_req->write)
1040                         ceph_update_write_latency(metric, req->r_start_latency,
1041                                                   req->r_end_latency, rc);
1042                 else
1043                         ceph_update_read_latency(metric, req->r_start_latency,
1044                                                  req->r_end_latency, rc);
1045         }
1046
1047         if (rc == -EOLDSNAPC) {
1048                 struct ceph_aio_work *aio_work;
1049                 BUG_ON(!aio_req->write);
1050
1051                 aio_work = kmalloc(sizeof(*aio_work), GFP_NOFS);
1052                 if (aio_work) {
1053                         INIT_WORK(&aio_work->work, ceph_aio_retry_work);
1054                         aio_work->req = req;
1055                         queue_work(ceph_inode_to_client(inode)->inode_wq,
1056                                    &aio_work->work);
1057                         return;
1058                 }
1059                 rc = -ENOMEM;
1060         } else if (!aio_req->write) {
1061                 if (rc == -ENOENT)
1062                         rc = 0;
1063                 if (rc >= 0 && osd_data->bvec_pos.iter.bi_size > rc) {
1064                         struct iov_iter i;
1065                         int zlen = osd_data->bvec_pos.iter.bi_size - rc;
1066
1067                         /*
1068                          * If read is satisfied by single OSD request,
1069                          * it can pass EOF. Otherwise read is within
1070                          * i_size.
1071                          */
1072                         if (aio_req->num_reqs == 1) {
1073                                 loff_t i_size = i_size_read(inode);
1074                                 loff_t endoff = aio_req->iocb->ki_pos + rc;
1075                                 if (endoff < i_size)
1076                                         zlen = min_t(size_t, zlen,
1077                                                      i_size - endoff);
1078                                 aio_req->total_len = rc + zlen;
1079                         }
1080
1081                         iov_iter_bvec(&i, READ, osd_data->bvec_pos.bvecs,
1082                                       osd_data->num_bvecs,
1083                                       osd_data->bvec_pos.iter.bi_size);
1084                         iov_iter_advance(&i, rc);
1085                         iov_iter_zero(zlen, &i);
1086                 }
1087         }
1088
1089         put_bvecs(osd_data->bvec_pos.bvecs, osd_data->num_bvecs,
1090                   aio_req->should_dirty);
1091         ceph_osdc_put_request(req);
1092
1093         if (rc < 0)
1094                 cmpxchg(&aio_req->error, 0, rc);
1095
1096         ceph_aio_complete(inode, aio_req);
1097         return;
1098 }
1099
1100 static void ceph_aio_retry_work(struct work_struct *work)
1101 {
1102         struct ceph_aio_work *aio_work =
1103                 container_of(work, struct ceph_aio_work, work);
1104         struct ceph_osd_request *orig_req = aio_work->req;
1105         struct ceph_aio_request *aio_req = orig_req->r_priv;
1106         struct inode *inode = orig_req->r_inode;
1107         struct ceph_inode_info *ci = ceph_inode(inode);
1108         struct ceph_snap_context *snapc;
1109         struct ceph_osd_request *req;
1110         int ret;
1111
1112         spin_lock(&ci->i_ceph_lock);
1113         if (__ceph_have_pending_cap_snap(ci)) {
1114                 struct ceph_cap_snap *capsnap =
1115                         list_last_entry(&ci->i_cap_snaps,
1116                                         struct ceph_cap_snap,
1117                                         ci_item);
1118                 snapc = ceph_get_snap_context(capsnap->context);
1119         } else {
1120                 BUG_ON(!ci->i_head_snapc);
1121                 snapc = ceph_get_snap_context(ci->i_head_snapc);
1122         }
1123         spin_unlock(&ci->i_ceph_lock);
1124
1125         req = ceph_osdc_alloc_request(orig_req->r_osdc, snapc, 1,
1126                         false, GFP_NOFS);
1127         if (!req) {
1128                 ret = -ENOMEM;
1129                 req = orig_req;
1130                 goto out;
1131         }
1132
1133         req->r_flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
1134         ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc);
1135         ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid);
1136
1137         req->r_ops[0] = orig_req->r_ops[0];
1138
1139         req->r_mtime = aio_req->mtime;
1140         req->r_data_offset = req->r_ops[0].extent.offset;
1141
1142         ret = ceph_osdc_alloc_messages(req, GFP_NOFS);
1143         if (ret) {
1144                 ceph_osdc_put_request(req);
1145                 req = orig_req;
1146                 goto out;
1147         }
1148
1149         ceph_osdc_put_request(orig_req);
1150
1151         req->r_callback = ceph_aio_complete_req;
1152         req->r_inode = inode;
1153         req->r_priv = aio_req;
1154
1155         ret = ceph_osdc_start_request(req->r_osdc, req, false);
1156 out:
1157         if (ret < 0) {
1158                 req->r_result = ret;
1159                 ceph_aio_complete_req(req);
1160         }
1161
1162         ceph_put_snap_context(snapc);
1163         kfree(aio_work);
1164 }
1165
1166 static ssize_t
1167 ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
1168                        struct ceph_snap_context *snapc,
1169                        struct ceph_cap_flush **pcf)
1170 {
1171         struct file *file = iocb->ki_filp;
1172         struct inode *inode = file_inode(file);
1173         struct ceph_inode_info *ci = ceph_inode(inode);
1174         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1175         struct ceph_client_metric *metric = &fsc->mdsc->metric;
1176         struct ceph_vino vino;
1177         struct ceph_osd_request *req;
1178         struct bio_vec *bvecs;
1179         struct ceph_aio_request *aio_req = NULL;
1180         int num_pages = 0;
1181         int flags;
1182         int ret = 0;
1183         struct timespec64 mtime = current_time(inode);
1184         size_t count = iov_iter_count(iter);
1185         loff_t pos = iocb->ki_pos;
1186         bool write = iov_iter_rw(iter) == WRITE;
1187         bool should_dirty = !write && iter_is_iovec(iter);
1188
1189         if (write && ceph_snap(file_inode(file)) != CEPH_NOSNAP)
1190                 return -EROFS;
1191
1192         dout("sync_direct_%s on file %p %lld~%u snapc %p seq %lld\n",
1193              (write ? "write" : "read"), file, pos, (unsigned)count,
1194              snapc, snapc ? snapc->seq : 0);
1195
1196         if (write) {
1197                 int ret2 = invalidate_inode_pages2_range(inode->i_mapping,
1198                                         pos >> PAGE_SHIFT,
1199                                         (pos + count - 1) >> PAGE_SHIFT);
1200                 if (ret2 < 0)
1201                         dout("invalidate_inode_pages2_range returned %d\n", ret2);
1202
1203                 flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
1204         } else {
1205                 flags = CEPH_OSD_FLAG_READ;
1206         }
1207
1208         while (iov_iter_count(iter) > 0) {
1209                 u64 size = iov_iter_count(iter);
1210                 ssize_t len;
1211
1212                 if (write)
1213                         size = min_t(u64, size, fsc->mount_options->wsize);
1214                 else
1215                         size = min_t(u64, size, fsc->mount_options->rsize);
1216
1217                 vino = ceph_vino(inode);
1218                 req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1219                                             vino, pos, &size, 0,
1220                                             1,
1221                                             write ? CEPH_OSD_OP_WRITE :
1222                                                     CEPH_OSD_OP_READ,
1223                                             flags, snapc,
1224                                             ci->i_truncate_seq,
1225                                             ci->i_truncate_size,
1226                                             false);
1227                 if (IS_ERR(req)) {
1228                         ret = PTR_ERR(req);
1229                         break;
1230                 }
1231
1232                 len = iter_get_bvecs_alloc(iter, size, &bvecs, &num_pages);
1233                 if (len < 0) {
1234                         ceph_osdc_put_request(req);
1235                         ret = len;
1236                         break;
1237                 }
1238                 if (len != size)
1239                         osd_req_op_extent_update(req, 0, len);
1240
1241                 /*
1242                  * To simplify error handling, allow AIO when IO within i_size
1243                  * or IO can be satisfied by single OSD request.
1244                  */
1245                 if (pos == iocb->ki_pos && !is_sync_kiocb(iocb) &&
1246                     (len == count || pos + count <= i_size_read(inode))) {
1247                         aio_req = kzalloc(sizeof(*aio_req), GFP_KERNEL);
1248                         if (aio_req) {
1249                                 aio_req->iocb = iocb;
1250                                 aio_req->write = write;
1251                                 aio_req->should_dirty = should_dirty;
1252                                 INIT_LIST_HEAD(&aio_req->osd_reqs);
1253                                 if (write) {
1254                                         aio_req->mtime = mtime;
1255                                         swap(aio_req->prealloc_cf, *pcf);
1256                                 }
1257                         }
1258                         /* ignore error */
1259                 }
1260
1261                 if (write) {
1262                         /*
1263                          * throw out any page cache pages in this range. this
1264                          * may block.
1265                          */
1266                         truncate_inode_pages_range(inode->i_mapping, pos,
1267                                                    PAGE_ALIGN(pos + len) - 1);
1268
1269                         req->r_mtime = mtime;
1270                 }
1271
1272                 osd_req_op_extent_osd_data_bvecs(req, 0, bvecs, num_pages, len);
1273
1274                 if (aio_req) {
1275                         aio_req->total_len += len;
1276                         aio_req->num_reqs++;
1277                         atomic_inc(&aio_req->pending_reqs);
1278
1279                         req->r_callback = ceph_aio_complete_req;
1280                         req->r_inode = inode;
1281                         req->r_priv = aio_req;
1282                         list_add_tail(&req->r_private_item, &aio_req->osd_reqs);
1283
1284                         pos += len;
1285                         continue;
1286                 }
1287
1288                 ret = ceph_osdc_start_request(req->r_osdc, req, false);
1289                 if (!ret)
1290                         ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
1291
1292                 if (write)
1293                         ceph_update_write_latency(metric, req->r_start_latency,
1294                                                   req->r_end_latency, ret);
1295                 else
1296                         ceph_update_read_latency(metric, req->r_start_latency,
1297                                                  req->r_end_latency, ret);
1298
1299                 size = i_size_read(inode);
1300                 if (!write) {
1301                         if (ret == -ENOENT)
1302                                 ret = 0;
1303                         if (ret >= 0 && ret < len && pos + ret < size) {
1304                                 struct iov_iter i;
1305                                 int zlen = min_t(size_t, len - ret,
1306                                                  size - pos - ret);
1307
1308                                 iov_iter_bvec(&i, READ, bvecs, num_pages, len);
1309                                 iov_iter_advance(&i, ret);
1310                                 iov_iter_zero(zlen, &i);
1311                                 ret += zlen;
1312                         }
1313                         if (ret >= 0)
1314                                 len = ret;
1315                 }
1316
1317                 put_bvecs(bvecs, num_pages, should_dirty);
1318                 ceph_osdc_put_request(req);
1319                 if (ret < 0)
1320                         break;
1321
1322                 pos += len;
1323                 if (!write && pos >= size)
1324                         break;
1325
1326                 if (write && pos > size) {
1327                         if (ceph_inode_set_size(inode, pos))
1328                                 ceph_check_caps(ceph_inode(inode),
1329                                                 CHECK_CAPS_AUTHONLY,
1330                                                 NULL);
1331                 }
1332         }
1333
1334         if (aio_req) {
1335                 LIST_HEAD(osd_reqs);
1336
1337                 if (aio_req->num_reqs == 0) {
1338                         kfree(aio_req);
1339                         return ret;
1340                 }
1341
1342                 ceph_get_cap_refs(ci, write ? CEPH_CAP_FILE_WR :
1343                                               CEPH_CAP_FILE_RD);
1344
1345                 list_splice(&aio_req->osd_reqs, &osd_reqs);
1346                 inode_dio_begin(inode);
1347                 while (!list_empty(&osd_reqs)) {
1348                         req = list_first_entry(&osd_reqs,
1349                                                struct ceph_osd_request,
1350                                                r_private_item);
1351                         list_del_init(&req->r_private_item);
1352                         if (ret >= 0)
1353                                 ret = ceph_osdc_start_request(req->r_osdc,
1354                                                               req, false);
1355                         if (ret < 0) {
1356                                 req->r_result = ret;
1357                                 ceph_aio_complete_req(req);
1358                         }
1359                 }
1360                 return -EIOCBQUEUED;
1361         }
1362
1363         if (ret != -EOLDSNAPC && pos > iocb->ki_pos) {
1364                 ret = pos - iocb->ki_pos;
1365                 iocb->ki_pos = pos;
1366         }
1367         return ret;
1368 }
1369
1370 /*
1371  * Synchronous write, straight from __user pointer or user pages.
1372  *
1373  * If write spans object boundary, just do multiple writes.  (For a
1374  * correct atomic write, we should e.g. take write locks on all
1375  * objects, rollback on failure, etc.)
1376  */
1377 static ssize_t
1378 ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
1379                 struct ceph_snap_context *snapc)
1380 {
1381         struct file *file = iocb->ki_filp;
1382         struct inode *inode = file_inode(file);
1383         struct ceph_inode_info *ci = ceph_inode(inode);
1384         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1385         struct ceph_vino vino;
1386         struct ceph_osd_request *req;
1387         struct page **pages;
1388         u64 len;
1389         int num_pages;
1390         int written = 0;
1391         int flags;
1392         int ret;
1393         bool check_caps = false;
1394         struct timespec64 mtime = current_time(inode);
1395         size_t count = iov_iter_count(from);
1396
1397         if (ceph_snap(file_inode(file)) != CEPH_NOSNAP)
1398                 return -EROFS;
1399
1400         dout("sync_write on file %p %lld~%u snapc %p seq %lld\n",
1401              file, pos, (unsigned)count, snapc, snapc->seq);
1402
1403         ret = filemap_write_and_wait_range(inode->i_mapping,
1404                                            pos, pos + count - 1);
1405         if (ret < 0)
1406                 return ret;
1407
1408         ret = invalidate_inode_pages2_range(inode->i_mapping,
1409                                             pos >> PAGE_SHIFT,
1410                                             (pos + count - 1) >> PAGE_SHIFT);
1411         if (ret < 0)
1412                 dout("invalidate_inode_pages2_range returned %d\n", ret);
1413
1414         flags = /* CEPH_OSD_FLAG_ORDERSNAP | */ CEPH_OSD_FLAG_WRITE;
1415
1416         while ((len = iov_iter_count(from)) > 0) {
1417                 size_t left;
1418                 int n;
1419
1420                 vino = ceph_vino(inode);
1421                 req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1422                                             vino, pos, &len, 0, 1,
1423                                             CEPH_OSD_OP_WRITE, flags, snapc,
1424                                             ci->i_truncate_seq,
1425                                             ci->i_truncate_size,
1426                                             false);
1427                 if (IS_ERR(req)) {
1428                         ret = PTR_ERR(req);
1429                         break;
1430                 }
1431
1432                 /*
1433                  * write from beginning of first page,
1434                  * regardless of io alignment
1435                  */
1436                 num_pages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
1437
1438                 pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
1439                 if (IS_ERR(pages)) {
1440                         ret = PTR_ERR(pages);
1441                         goto out;
1442                 }
1443
1444                 left = len;
1445                 for (n = 0; n < num_pages; n++) {
1446                         size_t plen = min_t(size_t, left, PAGE_SIZE);
1447                         ret = copy_page_from_iter(pages[n], 0, plen, from);
1448                         if (ret != plen) {
1449                                 ret = -EFAULT;
1450                                 break;
1451                         }
1452                         left -= ret;
1453                 }
1454
1455                 if (ret < 0) {
1456                         ceph_release_page_vector(pages, num_pages);
1457                         goto out;
1458                 }
1459
1460                 req->r_inode = inode;
1461
1462                 osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0,
1463                                                 false, true);
1464
1465                 req->r_mtime = mtime;
1466                 ret = ceph_osdc_start_request(&fsc->client->osdc, req, false);
1467                 if (!ret)
1468                         ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
1469
1470                 ceph_update_write_latency(&fsc->mdsc->metric, req->r_start_latency,
1471                                           req->r_end_latency, ret);
1472 out:
1473                 ceph_osdc_put_request(req);
1474                 if (ret != 0) {
1475                         ceph_set_error_write(ci);
1476                         break;
1477                 }
1478
1479                 ceph_clear_error_write(ci);
1480                 pos += len;
1481                 written += len;
1482                 if (pos > i_size_read(inode)) {
1483                         check_caps = ceph_inode_set_size(inode, pos);
1484                         if (check_caps)
1485                                 ceph_check_caps(ceph_inode(inode),
1486                                                 CHECK_CAPS_AUTHONLY,
1487                                                 NULL);
1488                 }
1489
1490         }
1491
1492         if (ret != -EOLDSNAPC && written > 0) {
1493                 ret = written;
1494                 iocb->ki_pos = pos;
1495         }
1496         return ret;
1497 }
1498
1499 /*
1500  * Wrap generic_file_aio_read with checks for cap bits on the inode.
1501  * Atomically grab references, so that those bits are not released
1502  * back to the MDS mid-read.
1503  *
1504  * Hmm, the sync read case isn't actually async... should it be?
1505  */
1506 static ssize_t ceph_read_iter(struct kiocb *iocb, struct iov_iter *to)
1507 {
1508         struct file *filp = iocb->ki_filp;
1509         struct ceph_file_info *fi = filp->private_data;
1510         size_t len = iov_iter_count(to);
1511         struct inode *inode = file_inode(filp);
1512         struct ceph_inode_info *ci = ceph_inode(inode);
1513         struct page *pinned_page = NULL;
1514         bool direct_lock = iocb->ki_flags & IOCB_DIRECT;
1515         ssize_t ret;
1516         int want, got = 0;
1517         int retry_op = 0, read = 0;
1518
1519 again:
1520         dout("aio_read %p %llx.%llx %llu~%u trying to get caps on %p\n",
1521              inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len, inode);
1522
1523         if (direct_lock)
1524                 ceph_start_io_direct(inode);
1525         else
1526                 ceph_start_io_read(inode);
1527
1528         if (fi->fmode & CEPH_FILE_MODE_LAZY)
1529                 want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
1530         else
1531                 want = CEPH_CAP_FILE_CACHE;
1532         ret = ceph_get_caps(filp, CEPH_CAP_FILE_RD, want, -1,
1533                             &got, &pinned_page);
1534         if (ret < 0) {
1535                 if (iocb->ki_flags & IOCB_DIRECT)
1536                         ceph_end_io_direct(inode);
1537                 else
1538                         ceph_end_io_read(inode);
1539                 return ret;
1540         }
1541
1542         if ((got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0 ||
1543             (iocb->ki_flags & IOCB_DIRECT) ||
1544             (fi->flags & CEPH_F_SYNC)) {
1545
1546                 dout("aio_sync_read %p %llx.%llx %llu~%u got cap refs on %s\n",
1547                      inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
1548                      ceph_cap_string(got));
1549
1550                 if (ci->i_inline_version == CEPH_INLINE_NONE) {
1551                         if (!retry_op && (iocb->ki_flags & IOCB_DIRECT)) {
1552                                 ret = ceph_direct_read_write(iocb, to,
1553                                                              NULL, NULL);
1554                                 if (ret >= 0 && ret < len)
1555                                         retry_op = CHECK_EOF;
1556                         } else {
1557                                 ret = ceph_sync_read(iocb, to, &retry_op);
1558                         }
1559                 } else {
1560                         retry_op = READ_INLINE;
1561                 }
1562         } else {
1563                 CEPH_DEFINE_RW_CONTEXT(rw_ctx, got);
1564                 dout("aio_read %p %llx.%llx %llu~%u got cap refs on %s\n",
1565                      inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
1566                      ceph_cap_string(got));
1567                 ceph_add_rw_context(fi, &rw_ctx);
1568                 ret = generic_file_read_iter(iocb, to);
1569                 ceph_del_rw_context(fi, &rw_ctx);
1570         }
1571
1572         dout("aio_read %p %llx.%llx dropping cap refs on %s = %d\n",
1573              inode, ceph_vinop(inode), ceph_cap_string(got), (int)ret);
1574         if (pinned_page) {
1575                 put_page(pinned_page);
1576                 pinned_page = NULL;
1577         }
1578         ceph_put_cap_refs(ci, got);
1579
1580         if (direct_lock)
1581                 ceph_end_io_direct(inode);
1582         else
1583                 ceph_end_io_read(inode);
1584
1585         if (retry_op > HAVE_RETRIED && ret >= 0) {
1586                 int statret;
1587                 struct page *page = NULL;
1588                 loff_t i_size;
1589                 if (retry_op == READ_INLINE) {
1590                         page = __page_cache_alloc(GFP_KERNEL);
1591                         if (!page)
1592                                 return -ENOMEM;
1593                 }
1594
1595                 statret = __ceph_do_getattr(inode, page,
1596                                             CEPH_STAT_CAP_INLINE_DATA, !!page);
1597                 if (statret < 0) {
1598                         if (page)
1599                                 __free_page(page);
1600                         if (statret == -ENODATA) {
1601                                 BUG_ON(retry_op != READ_INLINE);
1602                                 goto again;
1603                         }
1604                         return statret;
1605                 }
1606
1607                 i_size = i_size_read(inode);
1608                 if (retry_op == READ_INLINE) {
1609                         BUG_ON(ret > 0 || read > 0);
1610                         if (iocb->ki_pos < i_size &&
1611                             iocb->ki_pos < PAGE_SIZE) {
1612                                 loff_t end = min_t(loff_t, i_size,
1613                                                    iocb->ki_pos + len);
1614                                 end = min_t(loff_t, end, PAGE_SIZE);
1615                                 if (statret < end)
1616                                         zero_user_segment(page, statret, end);
1617                                 ret = copy_page_to_iter(page,
1618                                                 iocb->ki_pos & ~PAGE_MASK,
1619                                                 end - iocb->ki_pos, to);
1620                                 iocb->ki_pos += ret;
1621                                 read += ret;
1622                         }
1623                         if (iocb->ki_pos < i_size && read < len) {
1624                                 size_t zlen = min_t(size_t, len - read,
1625                                                     i_size - iocb->ki_pos);
1626                                 ret = iov_iter_zero(zlen, to);
1627                                 iocb->ki_pos += ret;
1628                                 read += ret;
1629                         }
1630                         __free_pages(page, 0);
1631                         return read;
1632                 }
1633
1634                 /* hit EOF or hole? */
1635                 if (retry_op == CHECK_EOF && iocb->ki_pos < i_size &&
1636                     ret < len) {
1637                         dout("sync_read hit hole, ppos %lld < size %lld"
1638                              ", reading more\n", iocb->ki_pos, i_size);
1639
1640                         read += ret;
1641                         len -= ret;
1642                         retry_op = HAVE_RETRIED;
1643                         goto again;
1644                 }
1645         }
1646
1647         if (ret >= 0)
1648                 ret += read;
1649
1650         return ret;
1651 }
1652
1653 /*
1654  * Take cap references to avoid releasing caps to MDS mid-write.
1655  *
1656  * If we are synchronous, and write with an old snap context, the OSD
1657  * may return EOLDSNAPC.  In that case, retry the write.. _after_
1658  * dropping our cap refs and allowing the pending snap to logically
1659  * complete _before_ this write occurs.
1660  *
1661  * If we are near ENOSPC, write synchronously.
1662  */
1663 static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
1664 {
1665         struct file *file = iocb->ki_filp;
1666         struct ceph_file_info *fi = file->private_data;
1667         struct inode *inode = file_inode(file);
1668         struct ceph_inode_info *ci = ceph_inode(inode);
1669         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1670         struct ceph_osd_client *osdc = &fsc->client->osdc;
1671         struct ceph_cap_flush *prealloc_cf;
1672         ssize_t count, written = 0;
1673         int err, want, got;
1674         bool direct_lock = false;
1675         u32 map_flags;
1676         u64 pool_flags;
1677         loff_t pos;
1678         loff_t limit = max(i_size_read(inode), fsc->max_file_size);
1679
1680         if (ceph_snap(inode) != CEPH_NOSNAP)
1681                 return -EROFS;
1682
1683         prealloc_cf = ceph_alloc_cap_flush();
1684         if (!prealloc_cf)
1685                 return -ENOMEM;
1686
1687         if ((iocb->ki_flags & (IOCB_DIRECT | IOCB_APPEND)) == IOCB_DIRECT)
1688                 direct_lock = true;
1689
1690 retry_snap:
1691         if (direct_lock)
1692                 ceph_start_io_direct(inode);
1693         else
1694                 ceph_start_io_write(inode);
1695
1696         /* We can write back this queue in page reclaim */
1697         current->backing_dev_info = inode_to_bdi(inode);
1698
1699         if (iocb->ki_flags & IOCB_APPEND) {
1700                 err = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
1701                 if (err < 0)
1702                         goto out;
1703         }
1704
1705         err = generic_write_checks(iocb, from);
1706         if (err <= 0)
1707                 goto out;
1708
1709         pos = iocb->ki_pos;
1710         if (unlikely(pos >= limit)) {
1711                 err = -EFBIG;
1712                 goto out;
1713         } else {
1714                 iov_iter_truncate(from, limit - pos);
1715         }
1716
1717         count = iov_iter_count(from);
1718         if (ceph_quota_is_max_bytes_exceeded(inode, pos + count)) {
1719                 err = -EDQUOT;
1720                 goto out;
1721         }
1722
1723         err = file_remove_privs(file);
1724         if (err)
1725                 goto out;
1726
1727         err = file_update_time(file);
1728         if (err)
1729                 goto out;
1730
1731         inode_inc_iversion_raw(inode);
1732
1733         if (ci->i_inline_version != CEPH_INLINE_NONE) {
1734                 err = ceph_uninline_data(file, NULL);
1735                 if (err < 0)
1736                         goto out;
1737         }
1738
1739         down_read(&osdc->lock);
1740         map_flags = osdc->osdmap->flags;
1741         pool_flags = ceph_pg_pool_flags(osdc->osdmap, ci->i_layout.pool_id);
1742         up_read(&osdc->lock);
1743         if ((map_flags & CEPH_OSDMAP_FULL) ||
1744             (pool_flags & CEPH_POOL_FLAG_FULL)) {
1745                 err = -ENOSPC;
1746                 goto out;
1747         }
1748
1749         dout("aio_write %p %llx.%llx %llu~%zd getting caps. i_size %llu\n",
1750              inode, ceph_vinop(inode), pos, count, i_size_read(inode));
1751         if (fi->fmode & CEPH_FILE_MODE_LAZY)
1752                 want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
1753         else
1754                 want = CEPH_CAP_FILE_BUFFER;
1755         got = 0;
1756         err = ceph_get_caps(file, CEPH_CAP_FILE_WR, want, pos + count,
1757                             &got, NULL);
1758         if (err < 0)
1759                 goto out;
1760
1761         dout("aio_write %p %llx.%llx %llu~%zd got cap refs on %s\n",
1762              inode, ceph_vinop(inode), pos, count, ceph_cap_string(got));
1763
1764         if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
1765             (iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC) ||
1766             (ci->i_ceph_flags & CEPH_I_ERROR_WRITE)) {
1767                 struct ceph_snap_context *snapc;
1768                 struct iov_iter data;
1769
1770                 spin_lock(&ci->i_ceph_lock);
1771                 if (__ceph_have_pending_cap_snap(ci)) {
1772                         struct ceph_cap_snap *capsnap =
1773                                         list_last_entry(&ci->i_cap_snaps,
1774                                                         struct ceph_cap_snap,
1775                                                         ci_item);
1776                         snapc = ceph_get_snap_context(capsnap->context);
1777                 } else {
1778                         BUG_ON(!ci->i_head_snapc);
1779                         snapc = ceph_get_snap_context(ci->i_head_snapc);
1780                 }
1781                 spin_unlock(&ci->i_ceph_lock);
1782
1783                 /* we might need to revert back to that point */
1784                 data = *from;
1785                 if (iocb->ki_flags & IOCB_DIRECT)
1786                         written = ceph_direct_read_write(iocb, &data, snapc,
1787                                                          &prealloc_cf);
1788                 else
1789                         written = ceph_sync_write(iocb, &data, pos, snapc);
1790                 if (direct_lock)
1791                         ceph_end_io_direct(inode);
1792                 else
1793                         ceph_end_io_write(inode);
1794                 if (written > 0)
1795                         iov_iter_advance(from, written);
1796                 ceph_put_snap_context(snapc);
1797         } else {
1798                 /*
1799                  * No need to acquire the i_truncate_mutex. Because
1800                  * the MDS revokes Fwb caps before sending truncate
1801                  * message to us. We can't get Fwb cap while there
1802                  * are pending vmtruncate. So write and vmtruncate
1803                  * can not run at the same time
1804                  */
1805                 written = generic_perform_write(file, from, pos);
1806                 if (likely(written >= 0))
1807                         iocb->ki_pos = pos + written;
1808                 ceph_end_io_write(inode);
1809         }
1810
1811         if (written >= 0) {
1812                 int dirty;
1813
1814                 spin_lock(&ci->i_ceph_lock);
1815                 ci->i_inline_version = CEPH_INLINE_NONE;
1816                 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
1817                                                &prealloc_cf);
1818                 spin_unlock(&ci->i_ceph_lock);
1819                 if (dirty)
1820                         __mark_inode_dirty(inode, dirty);
1821                 if (ceph_quota_is_max_bytes_approaching(inode, iocb->ki_pos))
1822                         ceph_check_caps(ci, 0, NULL);
1823         }
1824
1825         dout("aio_write %p %llx.%llx %llu~%u  dropping cap refs on %s\n",
1826              inode, ceph_vinop(inode), pos, (unsigned)count,
1827              ceph_cap_string(got));
1828         ceph_put_cap_refs(ci, got);
1829
1830         if (written == -EOLDSNAPC) {
1831                 dout("aio_write %p %llx.%llx %llu~%u" "got EOLDSNAPC, retrying\n",
1832                      inode, ceph_vinop(inode), pos, (unsigned)count);
1833                 goto retry_snap;
1834         }
1835
1836         if (written >= 0) {
1837                 if ((map_flags & CEPH_OSDMAP_NEARFULL) ||
1838                     (pool_flags & CEPH_POOL_FLAG_NEARFULL))
1839                         iocb->ki_flags |= IOCB_DSYNC;
1840                 written = generic_write_sync(iocb, written);
1841         }
1842
1843         goto out_unlocked;
1844 out:
1845         if (direct_lock)
1846                 ceph_end_io_direct(inode);
1847         else
1848                 ceph_end_io_write(inode);
1849 out_unlocked:
1850         ceph_free_cap_flush(prealloc_cf);
1851         current->backing_dev_info = NULL;
1852         return written ? written : err;
1853 }
1854
1855 /*
1856  * llseek.  be sure to verify file size on SEEK_END.
1857  */
1858 static loff_t ceph_llseek(struct file *file, loff_t offset, int whence)
1859 {
1860         struct inode *inode = file->f_mapping->host;
1861         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1862         loff_t i_size;
1863         loff_t ret;
1864
1865         inode_lock(inode);
1866
1867         if (whence == SEEK_END || whence == SEEK_DATA || whence == SEEK_HOLE) {
1868                 ret = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
1869                 if (ret < 0)
1870                         goto out;
1871         }
1872
1873         i_size = i_size_read(inode);
1874         switch (whence) {
1875         case SEEK_END:
1876                 offset += i_size;
1877                 break;
1878         case SEEK_CUR:
1879                 /*
1880                  * Here we special-case the lseek(fd, 0, SEEK_CUR)
1881                  * position-querying operation.  Avoid rewriting the "same"
1882                  * f_pos value back to the file because a concurrent read(),
1883                  * write() or lseek() might have altered it
1884                  */
1885                 if (offset == 0) {
1886                         ret = file->f_pos;
1887                         goto out;
1888                 }
1889                 offset += file->f_pos;
1890                 break;
1891         case SEEK_DATA:
1892                 if (offset < 0 || offset >= i_size) {
1893                         ret = -ENXIO;
1894                         goto out;
1895                 }
1896                 break;
1897         case SEEK_HOLE:
1898                 if (offset < 0 || offset >= i_size) {
1899                         ret = -ENXIO;
1900                         goto out;
1901                 }
1902                 offset = i_size;
1903                 break;
1904         }
1905
1906         ret = vfs_setpos(file, offset, max(i_size, fsc->max_file_size));
1907
1908 out:
1909         inode_unlock(inode);
1910         return ret;
1911 }
1912
1913 static inline void ceph_zero_partial_page(
1914         struct inode *inode, loff_t offset, unsigned size)
1915 {
1916         struct page *page;
1917         pgoff_t index = offset >> PAGE_SHIFT;
1918
1919         page = find_lock_page(inode->i_mapping, index);
1920         if (page) {
1921                 wait_on_page_writeback(page);
1922                 zero_user(page, offset & (PAGE_SIZE - 1), size);
1923                 unlock_page(page);
1924                 put_page(page);
1925         }
1926 }
1927
1928 static void ceph_zero_pagecache_range(struct inode *inode, loff_t offset,
1929                                       loff_t length)
1930 {
1931         loff_t nearly = round_up(offset, PAGE_SIZE);
1932         if (offset < nearly) {
1933                 loff_t size = nearly - offset;
1934                 if (length < size)
1935                         size = length;
1936                 ceph_zero_partial_page(inode, offset, size);
1937                 offset += size;
1938                 length -= size;
1939         }
1940         if (length >= PAGE_SIZE) {
1941                 loff_t size = round_down(length, PAGE_SIZE);
1942                 truncate_pagecache_range(inode, offset, offset + size - 1);
1943                 offset += size;
1944                 length -= size;
1945         }
1946         if (length)
1947                 ceph_zero_partial_page(inode, offset, length);
1948 }
1949
1950 static int ceph_zero_partial_object(struct inode *inode,
1951                                     loff_t offset, loff_t *length)
1952 {
1953         struct ceph_inode_info *ci = ceph_inode(inode);
1954         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1955         struct ceph_osd_request *req;
1956         int ret = 0;
1957         loff_t zero = 0;
1958         int op;
1959
1960         if (!length) {
1961                 op = offset ? CEPH_OSD_OP_DELETE : CEPH_OSD_OP_TRUNCATE;
1962                 length = &zero;
1963         } else {
1964                 op = CEPH_OSD_OP_ZERO;
1965         }
1966
1967         req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1968                                         ceph_vino(inode),
1969                                         offset, length,
1970                                         0, 1, op,
1971                                         CEPH_OSD_FLAG_WRITE,
1972                                         NULL, 0, 0, false);
1973         if (IS_ERR(req)) {
1974                 ret = PTR_ERR(req);
1975                 goto out;
1976         }
1977
1978         req->r_mtime = inode->i_mtime;
1979         ret = ceph_osdc_start_request(&fsc->client->osdc, req, false);
1980         if (!ret) {
1981                 ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
1982                 if (ret == -ENOENT)
1983                         ret = 0;
1984         }
1985         ceph_osdc_put_request(req);
1986
1987 out:
1988         return ret;
1989 }
1990
1991 static int ceph_zero_objects(struct inode *inode, loff_t offset, loff_t length)
1992 {
1993         int ret = 0;
1994         struct ceph_inode_info *ci = ceph_inode(inode);
1995         s32 stripe_unit = ci->i_layout.stripe_unit;
1996         s32 stripe_count = ci->i_layout.stripe_count;
1997         s32 object_size = ci->i_layout.object_size;
1998         u64 object_set_size = object_size * stripe_count;
1999         u64 nearly, t;
2000
2001         /* round offset up to next period boundary */
2002         nearly = offset + object_set_size - 1;
2003         t = nearly;
2004         nearly -= do_div(t, object_set_size);
2005
2006         while (length && offset < nearly) {
2007                 loff_t size = length;
2008                 ret = ceph_zero_partial_object(inode, offset, &size);
2009                 if (ret < 0)
2010                         return ret;
2011                 offset += size;
2012                 length -= size;
2013         }
2014         while (length >= object_set_size) {
2015                 int i;
2016                 loff_t pos = offset;
2017                 for (i = 0; i < stripe_count; ++i) {
2018                         ret = ceph_zero_partial_object(inode, pos, NULL);
2019                         if (ret < 0)
2020                                 return ret;
2021                         pos += stripe_unit;
2022                 }
2023                 offset += object_set_size;
2024                 length -= object_set_size;
2025         }
2026         while (length) {
2027                 loff_t size = length;
2028                 ret = ceph_zero_partial_object(inode, offset, &size);
2029                 if (ret < 0)
2030                         return ret;
2031                 offset += size;
2032                 length -= size;
2033         }
2034         return ret;
2035 }
2036
2037 static long ceph_fallocate(struct file *file, int mode,
2038                                 loff_t offset, loff_t length)
2039 {
2040         struct ceph_file_info *fi = file->private_data;
2041         struct inode *inode = file_inode(file);
2042         struct ceph_inode_info *ci = ceph_inode(inode);
2043         struct ceph_cap_flush *prealloc_cf;
2044         int want, got = 0;
2045         int dirty;
2046         int ret = 0;
2047         loff_t endoff = 0;
2048         loff_t size;
2049
2050         if (mode != (FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
2051                 return -EOPNOTSUPP;
2052
2053         if (!S_ISREG(inode->i_mode))
2054                 return -EOPNOTSUPP;
2055
2056         prealloc_cf = ceph_alloc_cap_flush();
2057         if (!prealloc_cf)
2058                 return -ENOMEM;
2059
2060         inode_lock(inode);
2061
2062         if (ceph_snap(inode) != CEPH_NOSNAP) {
2063                 ret = -EROFS;
2064                 goto unlock;
2065         }
2066
2067         if (ci->i_inline_version != CEPH_INLINE_NONE) {
2068                 ret = ceph_uninline_data(file, NULL);
2069                 if (ret < 0)
2070                         goto unlock;
2071         }
2072
2073         size = i_size_read(inode);
2074
2075         /* Are we punching a hole beyond EOF? */
2076         if (offset >= size)
2077                 goto unlock;
2078         if ((offset + length) > size)
2079                 length = size - offset;
2080
2081         if (fi->fmode & CEPH_FILE_MODE_LAZY)
2082                 want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
2083         else
2084                 want = CEPH_CAP_FILE_BUFFER;
2085
2086         ret = ceph_get_caps(file, CEPH_CAP_FILE_WR, want, endoff, &got, NULL);
2087         if (ret < 0)
2088                 goto unlock;
2089
2090         ceph_zero_pagecache_range(inode, offset, length);
2091         ret = ceph_zero_objects(inode, offset, length);
2092
2093         if (!ret) {
2094                 spin_lock(&ci->i_ceph_lock);
2095                 ci->i_inline_version = CEPH_INLINE_NONE;
2096                 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
2097                                                &prealloc_cf);
2098                 spin_unlock(&ci->i_ceph_lock);
2099                 if (dirty)
2100                         __mark_inode_dirty(inode, dirty);
2101         }
2102
2103         ceph_put_cap_refs(ci, got);
2104 unlock:
2105         inode_unlock(inode);
2106         ceph_free_cap_flush(prealloc_cf);
2107         return ret;
2108 }
2109
2110 /*
2111  * This function tries to get FILE_WR capabilities for dst_ci and FILE_RD for
2112  * src_ci.  Two attempts are made to obtain both caps, and an error is return if
2113  * this fails; zero is returned on success.
2114  */
2115 static int get_rd_wr_caps(struct file *src_filp, int *src_got,
2116                           struct file *dst_filp,
2117                           loff_t dst_endoff, int *dst_got)
2118 {
2119         int ret = 0;
2120         bool retrying = false;
2121
2122 retry_caps:
2123         ret = ceph_get_caps(dst_filp, CEPH_CAP_FILE_WR, CEPH_CAP_FILE_BUFFER,
2124                             dst_endoff, dst_got, NULL);
2125         if (ret < 0)
2126                 return ret;
2127
2128         /*
2129          * Since we're already holding the FILE_WR capability for the dst file,
2130          * we would risk a deadlock by using ceph_get_caps.  Thus, we'll do some
2131          * retry dance instead to try to get both capabilities.
2132          */
2133         ret = ceph_try_get_caps(file_inode(src_filp),
2134                                 CEPH_CAP_FILE_RD, CEPH_CAP_FILE_SHARED,
2135                                 false, src_got);
2136         if (ret <= 0) {
2137                 /* Start by dropping dst_ci caps and getting src_ci caps */
2138                 ceph_put_cap_refs(ceph_inode(file_inode(dst_filp)), *dst_got);
2139                 if (retrying) {
2140                         if (!ret)
2141                                 /* ceph_try_get_caps masks EAGAIN */
2142                                 ret = -EAGAIN;
2143                         return ret;
2144                 }
2145                 ret = ceph_get_caps(src_filp, CEPH_CAP_FILE_RD,
2146                                     CEPH_CAP_FILE_SHARED, -1, src_got, NULL);
2147                 if (ret < 0)
2148                         return ret;
2149                 /*... drop src_ci caps too, and retry */
2150                 ceph_put_cap_refs(ceph_inode(file_inode(src_filp)), *src_got);
2151                 retrying = true;
2152                 goto retry_caps;
2153         }
2154         return ret;
2155 }
2156
2157 static void put_rd_wr_caps(struct ceph_inode_info *src_ci, int src_got,
2158                            struct ceph_inode_info *dst_ci, int dst_got)
2159 {
2160         ceph_put_cap_refs(src_ci, src_got);
2161         ceph_put_cap_refs(dst_ci, dst_got);
2162 }
2163
2164 /*
2165  * This function does several size-related checks, returning an error if:
2166  *  - source file is smaller than off+len
2167  *  - destination file size is not OK (inode_newsize_ok())
2168  *  - max bytes quotas is exceeded
2169  */
2170 static int is_file_size_ok(struct inode *src_inode, struct inode *dst_inode,
2171                            loff_t src_off, loff_t dst_off, size_t len)
2172 {
2173         loff_t size, endoff;
2174
2175         size = i_size_read(src_inode);
2176         /*
2177          * Don't copy beyond source file EOF.  Instead of simply setting length
2178          * to (size - src_off), just drop to VFS default implementation, as the
2179          * local i_size may be stale due to other clients writing to the source
2180          * inode.
2181          */
2182         if (src_off + len > size) {
2183                 dout("Copy beyond EOF (%llu + %zu > %llu)\n",
2184                      src_off, len, size);
2185                 return -EOPNOTSUPP;
2186         }
2187         size = i_size_read(dst_inode);
2188
2189         endoff = dst_off + len;
2190         if (inode_newsize_ok(dst_inode, endoff))
2191                 return -EOPNOTSUPP;
2192
2193         if (ceph_quota_is_max_bytes_exceeded(dst_inode, endoff))
2194                 return -EDQUOT;
2195
2196         return 0;
2197 }
2198
2199 static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off,
2200                                     struct ceph_inode_info *dst_ci, u64 *dst_off,
2201                                     struct ceph_fs_client *fsc,
2202                                     size_t len, unsigned int flags)
2203 {
2204         struct ceph_object_locator src_oloc, dst_oloc;
2205         struct ceph_object_id src_oid, dst_oid;
2206         size_t bytes = 0;
2207         u64 src_objnum, src_objoff, dst_objnum, dst_objoff;
2208         u32 src_objlen, dst_objlen;
2209         u32 object_size = src_ci->i_layout.object_size;
2210         int ret;
2211
2212         src_oloc.pool = src_ci->i_layout.pool_id;
2213         src_oloc.pool_ns = ceph_try_get_string(src_ci->i_layout.pool_ns);
2214         dst_oloc.pool = dst_ci->i_layout.pool_id;
2215         dst_oloc.pool_ns = ceph_try_get_string(dst_ci->i_layout.pool_ns);
2216
2217         while (len >= object_size) {
2218                 ceph_calc_file_object_mapping(&src_ci->i_layout, *src_off,
2219                                               object_size, &src_objnum,
2220                                               &src_objoff, &src_objlen);
2221                 ceph_calc_file_object_mapping(&dst_ci->i_layout, *dst_off,
2222                                               object_size, &dst_objnum,
2223                                               &dst_objoff, &dst_objlen);
2224                 ceph_oid_init(&src_oid);
2225                 ceph_oid_printf(&src_oid, "%llx.%08llx",
2226                                 src_ci->i_vino.ino, src_objnum);
2227                 ceph_oid_init(&dst_oid);
2228                 ceph_oid_printf(&dst_oid, "%llx.%08llx",
2229                                 dst_ci->i_vino.ino, dst_objnum);
2230                 /* Do an object remote copy */
2231                 ret = ceph_osdc_copy_from(&fsc->client->osdc,
2232                                           src_ci->i_vino.snap, 0,
2233                                           &src_oid, &src_oloc,
2234                                           CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
2235                                           CEPH_OSD_OP_FLAG_FADVISE_NOCACHE,
2236                                           &dst_oid, &dst_oloc,
2237                                           CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
2238                                           CEPH_OSD_OP_FLAG_FADVISE_DONTNEED,
2239                                           dst_ci->i_truncate_seq,
2240                                           dst_ci->i_truncate_size,
2241                                           CEPH_OSD_COPY_FROM_FLAG_TRUNCATE_SEQ);
2242                 if (ret) {
2243                         if (ret == -EOPNOTSUPP) {
2244                                 fsc->have_copy_from2 = false;
2245                                 pr_notice("OSDs don't support copy-from2; disabling copy offload\n");
2246                         }
2247                         dout("ceph_osdc_copy_from returned %d\n", ret);
2248                         if (!bytes)
2249                                 bytes = ret;
2250                         goto out;
2251                 }
2252                 len -= object_size;
2253                 bytes += object_size;
2254                 *src_off += object_size;
2255                 *dst_off += object_size;
2256         }
2257
2258 out:
2259         ceph_oloc_destroy(&src_oloc);
2260         ceph_oloc_destroy(&dst_oloc);
2261         return bytes;
2262 }
2263
2264 static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
2265                                       struct file *dst_file, loff_t dst_off,
2266                                       size_t len, unsigned int flags)
2267 {
2268         struct inode *src_inode = file_inode(src_file);
2269         struct inode *dst_inode = file_inode(dst_file);
2270         struct ceph_inode_info *src_ci = ceph_inode(src_inode);
2271         struct ceph_inode_info *dst_ci = ceph_inode(dst_inode);
2272         struct ceph_cap_flush *prealloc_cf;
2273         struct ceph_fs_client *src_fsc = ceph_inode_to_client(src_inode);
2274         loff_t size;
2275         ssize_t ret = -EIO, bytes;
2276         u64 src_objnum, dst_objnum, src_objoff, dst_objoff;
2277         u32 src_objlen, dst_objlen;
2278         int src_got = 0, dst_got = 0, err, dirty;
2279
2280         if (src_inode->i_sb != dst_inode->i_sb) {
2281                 struct ceph_fs_client *dst_fsc = ceph_inode_to_client(dst_inode);
2282
2283                 if (ceph_fsid_compare(&src_fsc->client->fsid,
2284                                       &dst_fsc->client->fsid)) {
2285                         dout("Copying files across clusters: src: %pU dst: %pU\n",
2286                              &src_fsc->client->fsid, &dst_fsc->client->fsid);
2287                         return -EXDEV;
2288                 }
2289         }
2290         if (ceph_snap(dst_inode) != CEPH_NOSNAP)
2291                 return -EROFS;
2292
2293         /*
2294          * Some of the checks below will return -EOPNOTSUPP, which will force a
2295          * fallback to the default VFS copy_file_range implementation.  This is
2296          * desirable in several cases (for ex, the 'len' is smaller than the
2297          * size of the objects, or in cases where that would be more
2298          * efficient).
2299          */
2300
2301         if (ceph_test_mount_opt(src_fsc, NOCOPYFROM))
2302                 return -EOPNOTSUPP;
2303
2304         if (!src_fsc->have_copy_from2)
2305                 return -EOPNOTSUPP;
2306
2307         /*
2308          * Striped file layouts require that we copy partial objects, but the
2309          * OSD copy-from operation only supports full-object copies.  Limit
2310          * this to non-striped file layouts for now.
2311          */
2312         if ((src_ci->i_layout.stripe_unit != dst_ci->i_layout.stripe_unit) ||
2313             (src_ci->i_layout.stripe_count != 1) ||
2314             (dst_ci->i_layout.stripe_count != 1) ||
2315             (src_ci->i_layout.object_size != dst_ci->i_layout.object_size)) {
2316                 dout("Invalid src/dst files layout\n");
2317                 return -EOPNOTSUPP;
2318         }
2319
2320         if (len < src_ci->i_layout.object_size)
2321                 return -EOPNOTSUPP; /* no remote copy will be done */
2322
2323         prealloc_cf = ceph_alloc_cap_flush();
2324         if (!prealloc_cf)
2325                 return -ENOMEM;
2326
2327         /* Start by sync'ing the source and destination files */
2328         ret = file_write_and_wait_range(src_file, src_off, (src_off + len));
2329         if (ret < 0) {
2330                 dout("failed to write src file (%zd)\n", ret);
2331                 goto out;
2332         }
2333         ret = file_write_and_wait_range(dst_file, dst_off, (dst_off + len));
2334         if (ret < 0) {
2335                 dout("failed to write dst file (%zd)\n", ret);
2336                 goto out;
2337         }
2338
2339         /*
2340          * We need FILE_WR caps for dst_ci and FILE_RD for src_ci as other
2341          * clients may have dirty data in their caches.  And OSDs know nothing
2342          * about caps, so they can't safely do the remote object copies.
2343          */
2344         err = get_rd_wr_caps(src_file, &src_got,
2345                              dst_file, (dst_off + len), &dst_got);
2346         if (err < 0) {
2347                 dout("get_rd_wr_caps returned %d\n", err);
2348                 ret = -EOPNOTSUPP;
2349                 goto out;
2350         }
2351
2352         ret = is_file_size_ok(src_inode, dst_inode, src_off, dst_off, len);
2353         if (ret < 0)
2354                 goto out_caps;
2355
2356         /* Drop dst file cached pages */
2357         ret = invalidate_inode_pages2_range(dst_inode->i_mapping,
2358                                             dst_off >> PAGE_SHIFT,
2359                                             (dst_off + len) >> PAGE_SHIFT);
2360         if (ret < 0) {
2361                 dout("Failed to invalidate inode pages (%zd)\n", ret);
2362                 ret = 0; /* XXX */
2363         }
2364         ceph_calc_file_object_mapping(&src_ci->i_layout, src_off,
2365                                       src_ci->i_layout.object_size,
2366                                       &src_objnum, &src_objoff, &src_objlen);
2367         ceph_calc_file_object_mapping(&dst_ci->i_layout, dst_off,
2368                                       dst_ci->i_layout.object_size,
2369                                       &dst_objnum, &dst_objoff, &dst_objlen);
2370         /* object-level offsets need to the same */
2371         if (src_objoff != dst_objoff) {
2372                 ret = -EOPNOTSUPP;
2373                 goto out_caps;
2374         }
2375
2376         /*
2377          * Do a manual copy if the object offset isn't object aligned.
2378          * 'src_objlen' contains the bytes left until the end of the object,
2379          * starting at the src_off
2380          */
2381         if (src_objoff) {
2382                 dout("Initial partial copy of %u bytes\n", src_objlen);
2383
2384                 /*
2385                  * we need to temporarily drop all caps as we'll be calling
2386                  * {read,write}_iter, which will get caps again.
2387                  */
2388                 put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
2389                 ret = do_splice_direct(src_file, &src_off, dst_file,
2390                                        &dst_off, src_objlen, flags);
2391                 /* Abort on short copies or on error */
2392                 if (ret < src_objlen) {
2393                         dout("Failed partial copy (%zd)\n", ret);
2394                         goto out;
2395                 }
2396                 len -= ret;
2397                 err = get_rd_wr_caps(src_file, &src_got,
2398                                      dst_file, (dst_off + len), &dst_got);
2399                 if (err < 0)
2400                         goto out;
2401                 err = is_file_size_ok(src_inode, dst_inode,
2402                                       src_off, dst_off, len);
2403                 if (err < 0)
2404                         goto out_caps;
2405         }
2406
2407         size = i_size_read(dst_inode);
2408         bytes = ceph_do_objects_copy(src_ci, &src_off, dst_ci, &dst_off,
2409                                      src_fsc, len, flags);
2410         if (bytes <= 0) {
2411                 if (!ret)
2412                         ret = bytes;
2413                 goto out_caps;
2414         }
2415         dout("Copied %zu bytes out of %zu\n", bytes, len);
2416         len -= bytes;
2417         ret += bytes;
2418
2419         file_update_time(dst_file);
2420         inode_inc_iversion_raw(dst_inode);
2421
2422         if (dst_off > size) {
2423                 /* Let the MDS know about dst file size change */
2424                 if (ceph_inode_set_size(dst_inode, dst_off) ||
2425                     ceph_quota_is_max_bytes_approaching(dst_inode, dst_off))
2426                         ceph_check_caps(dst_ci, CHECK_CAPS_AUTHONLY, NULL);
2427         }
2428         /* Mark Fw dirty */
2429         spin_lock(&dst_ci->i_ceph_lock);
2430         dst_ci->i_inline_version = CEPH_INLINE_NONE;
2431         dirty = __ceph_mark_dirty_caps(dst_ci, CEPH_CAP_FILE_WR, &prealloc_cf);
2432         spin_unlock(&dst_ci->i_ceph_lock);
2433         if (dirty)
2434                 __mark_inode_dirty(dst_inode, dirty);
2435
2436 out_caps:
2437         put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
2438
2439         /*
2440          * Do the final manual copy if we still have some bytes left, unless
2441          * there were errors in remote object copies (len >= object_size).
2442          */
2443         if (len && (len < src_ci->i_layout.object_size)) {
2444                 dout("Final partial copy of %zu bytes\n", len);
2445                 bytes = do_splice_direct(src_file, &src_off, dst_file,
2446                                          &dst_off, len, flags);
2447                 if (bytes > 0)
2448                         ret += bytes;
2449                 else
2450                         dout("Failed partial copy (%zd)\n", bytes);
2451         }
2452
2453 out:
2454         ceph_free_cap_flush(prealloc_cf);
2455
2456         return ret;
2457 }
2458
2459 static ssize_t ceph_copy_file_range(struct file *src_file, loff_t src_off,
2460                                     struct file *dst_file, loff_t dst_off,
2461                                     size_t len, unsigned int flags)
2462 {
2463         ssize_t ret;
2464
2465         ret = __ceph_copy_file_range(src_file, src_off, dst_file, dst_off,
2466                                      len, flags);
2467
2468         if (ret == -EOPNOTSUPP || ret == -EXDEV)
2469                 ret = generic_copy_file_range(src_file, src_off, dst_file,
2470                                               dst_off, len, flags);
2471         return ret;
2472 }
2473
2474 const struct file_operations ceph_file_fops = {
2475         .open = ceph_open,
2476         .release = ceph_release,
2477         .llseek = ceph_llseek,
2478         .read_iter = ceph_read_iter,
2479         .write_iter = ceph_write_iter,
2480         .mmap = ceph_mmap,
2481         .fsync = ceph_fsync,
2482         .lock = ceph_lock,
2483         .setlease = simple_nosetlease,
2484         .flock = ceph_flock,
2485         .splice_read = generic_file_splice_read,
2486         .splice_write = iter_file_splice_write,
2487         .unlocked_ioctl = ceph_ioctl,
2488         .compat_ioctl = compat_ptr_ioctl,
2489         .fallocate      = ceph_fallocate,
2490         .copy_file_range = ceph_copy_file_range,
2491 };