linux/fs/ceph/file.c
<<
>>
Prefs
   1#include <linux/ceph/ceph_debug.h>
   2
   3#include <linux/module.h>
   4#include <linux/sched.h>
   5#include <linux/slab.h>
   6#include <linux/file.h>
   7#include <linux/mount.h>
   8#include <linux/namei.h>
   9#include <linux/writeback.h>
  10
  11#include "super.h"
  12#include "mds_client.h"
  13
  14/*
  15 * Ceph file operations
  16 *
  17 * Implement basic open/close functionality, and implement
  18 * read/write.
  19 *
  20 * We implement three modes of file I/O:
  21 *  - buffered uses the generic_file_aio_{read,write} helpers
  22 *
  23 *  - synchronous is used when there is multi-client read/write
  24 *    sharing, avoids the page cache, and synchronously waits for an
  25 *    ack from the OSD.
  26 *
  27 *  - direct io takes the variant of the sync path that references
  28 *    user pages directly.
  29 *
  30 * fsync() flushes and waits on dirty pages, but just queues metadata
  31 * for writeback: since the MDS can recover size and mtime there is no
  32 * need to wait for MDS acknowledgement.
  33 */
  34
  35
  36/*
  37 * Prepare an open request.  Preallocate ceph_cap to avoid an
  38 * inopportune ENOMEM later.
  39 */
  40static struct ceph_mds_request *
  41prepare_open_request(struct super_block *sb, int flags, int create_mode)
  42{
  43        struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
  44        struct ceph_mds_client *mdsc = fsc->mdsc;
  45        struct ceph_mds_request *req;
  46        int want_auth = USE_ANY_MDS;
  47        int op = (flags & O_CREAT) ? CEPH_MDS_OP_CREATE : CEPH_MDS_OP_OPEN;
  48
  49        if (flags & (O_WRONLY|O_RDWR|O_CREAT|O_TRUNC))
  50                want_auth = USE_AUTH_MDS;
  51
  52        req = ceph_mdsc_create_request(mdsc, op, want_auth);
  53        if (IS_ERR(req))
  54                goto out;
  55        req->r_fmode = ceph_flags_to_mode(flags);
  56        req->r_args.open.flags = cpu_to_le32(flags);
  57        req->r_args.open.mode = cpu_to_le32(create_mode);
  58out:
  59        return req;
  60}
  61
  62/*
  63 * initialize private struct file data.
  64 * if we fail, clean up by dropping fmode reference on the ceph_inode
  65 */
  66static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
  67{
  68        struct ceph_file_info *cf;
  69        int ret = 0;
  70
  71        switch (inode->i_mode & S_IFMT) {
  72        case S_IFREG:
  73        case S_IFDIR:
  74                dout("init_file %p %p 0%o (regular)\n", inode, file,
  75                     inode->i_mode);
  76                cf = kmem_cache_alloc(ceph_file_cachep, GFP_NOFS | __GFP_ZERO);
  77                if (cf == NULL) {
  78                        ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
  79                        return -ENOMEM;
  80                }
  81                cf->fmode = fmode;
  82                cf->next_offset = 2;
  83                file->private_data = cf;
  84                BUG_ON(inode->i_fop->release != ceph_release);
  85                break;
  86
  87        case S_IFLNK:
  88                dout("init_file %p %p 0%o (symlink)\n", inode, file,
  89                     inode->i_mode);
  90                ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
  91                break;
  92
  93        default:
  94                dout("init_file %p %p 0%o (special)\n", inode, file,
  95                     inode->i_mode);
  96                /*
  97                 * we need to drop the open ref now, since we don't
  98                 * have .release set to ceph_release.
  99                 */
 100                ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
 101                BUG_ON(inode->i_fop->release == ceph_release);
 102
 103                /* call the proper open fop */
 104                ret = inode->i_fop->open(inode, file);
 105        }
 106        return ret;
 107}
 108
 109/*
 110 * If we already have the requisite capabilities, we can satisfy
 111 * the open request locally (no need to request new caps from the
 112 * MDS).  We do, however, need to inform the MDS (asynchronously)
 113 * if our wanted caps set expands.
 114 */
 115int ceph_open(struct inode *inode, struct file *file)
 116{
 117        struct ceph_inode_info *ci = ceph_inode(inode);
 118        struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
 119        struct ceph_mds_client *mdsc = fsc->mdsc;
 120        struct ceph_mds_request *req;
 121        struct ceph_file_info *cf = file->private_data;
 122        struct inode *parent_inode = NULL;
 123        int err;
 124        int flags, fmode, wanted;
 125
 126        if (cf) {
 127                dout("open file %p is already opened\n", file);
 128                return 0;
 129        }
 130
 131        /* filter out O_CREAT|O_EXCL; vfs did that already.  yuck. */
 132        flags = file->f_flags & ~(O_CREAT|O_EXCL);
 133        if (S_ISDIR(inode->i_mode))
 134                flags = O_DIRECTORY;  /* mds likes to know */
 135
 136        dout("open inode %p ino %llx.%llx file %p flags %d (%d)\n", inode,
 137             ceph_vinop(inode), file, flags, file->f_flags);
 138        fmode = ceph_flags_to_mode(flags);
 139        wanted = ceph_caps_for_mode(fmode);
 140
 141        /* snapped files are read-only */
 142        if (ceph_snap(inode) != CEPH_NOSNAP && (file->f_mode & FMODE_WRITE))
 143                return -EROFS;
 144
 145        /* trivially open snapdir */
 146        if (ceph_snap(inode) == CEPH_SNAPDIR) {
 147                spin_lock(&ci->i_ceph_lock);
 148                __ceph_get_fmode(ci, fmode);
 149                spin_unlock(&ci->i_ceph_lock);
 150                return ceph_init_file(inode, file, fmode);
 151        }
 152
 153        /*
 154         * No need to block if we have caps on the auth MDS (for
 155         * write) or any MDS (for read).  Update wanted set
 156         * asynchronously.
 157         */
 158        spin_lock(&ci->i_ceph_lock);
 159        if (__ceph_is_any_real_caps(ci) &&
 160            (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) {
 161                int mds_wanted = __ceph_caps_mds_wanted(ci);
 162                int issued = __ceph_caps_issued(ci, NULL);
 163
 164                dout("open %p fmode %d want %s issued %s using existing\n",
 165                     inode, fmode, ceph_cap_string(wanted),
 166                     ceph_cap_string(issued));
 167                __ceph_get_fmode(ci, fmode);
 168                spin_unlock(&ci->i_ceph_lock);
 169
 170                /* adjust wanted? */
 171                if ((issued & wanted) != wanted &&
 172                    (mds_wanted & wanted) != wanted &&
 173                    ceph_snap(inode) != CEPH_SNAPDIR)
 174                        ceph_check_caps(ci, 0, NULL);
 175
 176                return ceph_init_file(inode, file, fmode);
 177        } else if (ceph_snap(inode) != CEPH_NOSNAP &&
 178                   (ci->i_snap_caps & wanted) == wanted) {
 179                __ceph_get_fmode(ci, fmode);
 180                spin_unlock(&ci->i_ceph_lock);
 181                return ceph_init_file(inode, file, fmode);
 182        }
 183        spin_unlock(&ci->i_ceph_lock);
 184
 185        dout("open fmode %d wants %s\n", fmode, ceph_cap_string(wanted));
 186        req = prepare_open_request(inode->i_sb, flags, 0);
 187        if (IS_ERR(req)) {
 188                err = PTR_ERR(req);
 189                goto out;
 190        }
 191        req->r_inode = inode;
 192        ihold(inode);
 193        req->r_num_caps = 1;
 194        if (flags & (O_CREAT|O_TRUNC))
 195                parent_inode = ceph_get_dentry_parent_inode(file->f_dentry);
 196        err = ceph_mdsc_do_request(mdsc, parent_inode, req);
 197        iput(parent_inode);
 198        if (!err)
 199                err = ceph_init_file(inode, file, req->r_fmode);
 200        ceph_mdsc_put_request(req);
 201        dout("open result=%d on %llx.%llx\n", err, ceph_vinop(inode));
 202out:
 203        return err;
 204}
 205
 206
 207/*
 208 * Do a lookup + open with a single request.  If we get a non-existent
 209 * file or symlink, return 1 so the VFS can retry.
 210 */
 211int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
 212                     struct file *file, unsigned flags, umode_t mode,
 213                     int *opened)
 214{
 215        struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
 216        struct ceph_mds_client *mdsc = fsc->mdsc;
 217        struct ceph_mds_request *req;
 218        struct dentry *dn;
 219        int err;
 220
 221        dout("atomic_open %p dentry %p '%.*s' %s flags %d mode 0%o\n",
 222             dir, dentry, dentry->d_name.len, dentry->d_name.name,
 223             d_unhashed(dentry) ? "unhashed" : "hashed", flags, mode);
 224
 225        if (dentry->d_name.len > NAME_MAX)
 226                return -ENAMETOOLONG;
 227
 228        err = ceph_init_dentry(dentry);
 229        if (err < 0)
 230                return err;
 231
 232        /* do the open */
 233        req = prepare_open_request(dir->i_sb, flags, mode);
 234        if (IS_ERR(req))
 235                return PTR_ERR(req);
 236        req->r_dentry = dget(dentry);
 237        req->r_num_caps = 2;
 238        if (flags & O_CREAT) {
 239                req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
 240                req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 241        }
 242        req->r_locked_dir = dir;           /* caller holds dir->i_mutex */
 243        err = ceph_mdsc_do_request(mdsc,
 244                                   (flags & (O_CREAT|O_TRUNC)) ? dir : NULL,
 245                                   req);
 246        err = ceph_handle_snapdir(req, dentry, err);
 247        if (err == 0 && (flags & O_CREAT) && !req->r_reply_info.head->is_dentry)
 248                err = ceph_handle_notrace_create(dir, dentry);
 249
 250        if (d_unhashed(dentry)) {
 251                dn = ceph_finish_lookup(req, dentry, err);
 252                if (IS_ERR(dn))
 253                        err = PTR_ERR(dn);
 254        } else {
 255                /* we were given a hashed negative dentry */
 256                dn = NULL;
 257        }
 258        if (err)
 259                goto out_err;
 260        if (dn || dentry->d_inode == NULL || S_ISLNK(dentry->d_inode->i_mode)) {
 261                /* make vfs retry on splice, ENOENT, or symlink */
 262                dout("atomic_open finish_no_open on dn %p\n", dn);
 263                err = finish_no_open(file, dn);
 264        } else {
 265                dout("atomic_open finish_open on dn %p\n", dn);
 266                err = finish_open(file, dentry, ceph_open, opened);
 267        }
 268
 269out_err:
 270        ceph_mdsc_put_request(req);
 271        dout("atomic_open result=%d\n", err);
 272        return err;
 273}
 274
 275int ceph_release(struct inode *inode, struct file *file)
 276{
 277        struct ceph_inode_info *ci = ceph_inode(inode);
 278        struct ceph_file_info *cf = file->private_data;
 279
 280        dout("release inode %p file %p\n", inode, file);
 281        ceph_put_fmode(ci, cf->fmode);
 282        if (cf->last_readdir)
 283                ceph_mdsc_put_request(cf->last_readdir);
 284        kfree(cf->last_name);
 285        kfree(cf->dir_info);
 286        dput(cf->dentry);
 287        kmem_cache_free(ceph_file_cachep, cf);
 288
 289        /* wake up anyone waiting for caps on this inode */
 290        wake_up_all(&ci->i_cap_wq);
 291        return 0;
 292}
 293
 294/*
 295 * Read a range of bytes striped over one or more objects.  Iterate over
 296 * objects we stripe over.  (That's not atomic, but good enough for now.)
 297 *
 298 * If we get a short result from the OSD, check against i_size; we need to
 299 * only return a short read to the caller if we hit EOF.
 300 */
 301static int striped_read(struct inode *inode,
 302                        u64 off, u64 len,
 303                        struct page **pages, int num_pages,
 304                        int *checkeof, bool o_direct,
 305                        unsigned long buf_align)
 306{
 307        struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 308        struct ceph_inode_info *ci = ceph_inode(inode);
 309        u64 pos, this_len;
 310        int io_align, page_align;
 311        int left, pages_left;
 312        int read;
 313        struct page **page_pos;
 314        int ret;
 315        bool hit_stripe, was_short;
 316
 317        /*
 318         * we may need to do multiple reads.  not atomic, unfortunately.
 319         */
 320        pos = off;
 321        left = len;
 322        page_pos = pages;
 323        pages_left = num_pages;
 324        read = 0;
 325        io_align = off & ~PAGE_MASK;
 326
 327more:
 328        if (o_direct)
 329                page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
 330        else
 331                page_align = pos & ~PAGE_MASK;
 332        this_len = left;
 333        ret = ceph_osdc_readpages(&fsc->client->osdc, ceph_vino(inode),
 334                                  &ci->i_layout, pos, &this_len,
 335                                  ci->i_truncate_seq,
 336                                  ci->i_truncate_size,
 337                                  page_pos, pages_left, page_align);
 338        if (ret == -ENOENT)
 339                ret = 0;
 340        hit_stripe = this_len < left;
 341        was_short = ret >= 0 && ret < this_len;
 342        dout("striped_read %llu~%u (read %u) got %d%s%s\n", pos, left, read,
 343             ret, hit_stripe ? " HITSTRIPE" : "", was_short ? " SHORT" : "");
 344
 345        if (ret > 0) {
 346                int didpages = (page_align + ret) >> PAGE_CACHE_SHIFT;
 347
 348                if (read < pos - off) {
 349                        dout(" zero gap %llu to %llu\n", off + read, pos);
 350                        ceph_zero_page_vector_range(page_align + read,
 351                                                    pos - off - read, pages);
 352                }
 353                pos += ret;
 354                read = pos - off;
 355                left -= ret;
 356                page_pos += didpages;
 357                pages_left -= didpages;
 358
 359                /* hit stripe? */
 360                if (left && hit_stripe)
 361                        goto more;
 362        }
 363
 364        if (was_short) {
 365                /* did we bounce off eof? */
 366                if (pos + left > inode->i_size)
 367                        *checkeof = 1;
 368
 369                /* zero trailing bytes (inside i_size) */
 370                if (left > 0 && pos < inode->i_size) {
 371                        if (pos + left > inode->i_size)
 372                                left = inode->i_size - pos;
 373
 374                        dout("zero tail %d\n", left);
 375                        ceph_zero_page_vector_range(page_align + read, left,
 376                                                    pages);
 377                        read += left;
 378                }
 379        }
 380
 381        if (ret >= 0)
 382                ret = read;
 383        dout("striped_read returns %d\n", ret);
 384        return ret;
 385}
 386
 387/*
 388 * Completely synchronous read and write methods.  Direct from __user
 389 * buffer to osd, or directly to user pages (if O_DIRECT).
 390 *
 391 * If the read spans object boundary, just do multiple reads.
 392 */
 393static ssize_t ceph_sync_read(struct file *file, char __user *data,
 394                              unsigned len, loff_t *poff, int *checkeof)
 395{
 396        struct inode *inode = file->f_dentry->d_inode;
 397        struct page **pages;
 398        u64 off = *poff;
 399        int num_pages, ret;
 400
 401        dout("sync_read on file %p %llu~%u %s\n", file, off, len,
 402             (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
 403
 404        if (file->f_flags & O_DIRECT) {
 405                num_pages = calc_pages_for((unsigned long)data, len);
 406                pages = ceph_get_direct_page_vector(data, num_pages, true);
 407        } else {
 408                num_pages = calc_pages_for(off, len);
 409                pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
 410        }
 411        if (IS_ERR(pages))
 412                return PTR_ERR(pages);
 413
 414        /*
 415         * flush any page cache pages in this range.  this
 416         * will make concurrent normal and sync io slow,
 417         * but it will at least behave sensibly when they are
 418         * in sequence.
 419         */
 420        ret = filemap_write_and_wait(inode->i_mapping);
 421        if (ret < 0)
 422                goto done;
 423
 424        ret = striped_read(inode, off, len, pages, num_pages, checkeof,
 425                           file->f_flags & O_DIRECT,
 426                           (unsigned long)data & ~PAGE_MASK);
 427
 428        if (ret >= 0 && (file->f_flags & O_DIRECT) == 0)
 429                ret = ceph_copy_page_vector_to_user(pages, data, off, ret);
 430        if (ret >= 0)
 431                *poff = off + ret;
 432
 433done:
 434        if (file->f_flags & O_DIRECT)
 435                ceph_put_page_vector(pages, num_pages, true);
 436        else
 437                ceph_release_page_vector(pages, num_pages);
 438        dout("sync_read result %d\n", ret);
 439        return ret;
 440}
 441
 442/*
 443 * Write commit callback, called if we requested both an ACK and
 444 * ONDISK commit reply from the OSD.
 445 */
 446static void sync_write_commit(struct ceph_osd_request *req,
 447                              struct ceph_msg *msg)
 448{
 449        struct ceph_inode_info *ci = ceph_inode(req->r_inode);
 450
 451        dout("sync_write_commit %p tid %llu\n", req, req->r_tid);
 452        spin_lock(&ci->i_unsafe_lock);
 453        list_del_init(&req->r_unsafe_item);
 454        spin_unlock(&ci->i_unsafe_lock);
 455        ceph_put_cap_refs(ci, CEPH_CAP_FILE_WR);
 456}
 457
 458/*
 459 * Synchronous write, straight from __user pointer or user pages (if
 460 * O_DIRECT).
 461 *
 462 * If write spans object boundary, just do multiple writes.  (For a
 463 * correct atomic write, we should e.g. take write locks on all
 464 * objects, rollback on failure, etc.)
 465 */
 466static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 467                               size_t left, loff_t *offset)
 468{
 469        struct inode *inode = file->f_dentry->d_inode;
 470        struct ceph_inode_info *ci = ceph_inode(inode);
 471        struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 472        struct ceph_osd_request *req;
 473        struct page **pages;
 474        int num_pages;
 475        long long unsigned pos;
 476        u64 len;
 477        int written = 0;
 478        int flags;
 479        int do_sync = 0;
 480        int check_caps = 0;
 481        int page_align, io_align;
 482        unsigned long buf_align;
 483        int ret;
 484        struct timespec mtime = CURRENT_TIME;
 485
 486        if (ceph_snap(file->f_dentry->d_inode) != CEPH_NOSNAP)
 487                return -EROFS;
 488
 489        dout("sync_write on file %p %lld~%u %s\n", file, *offset,
 490             (unsigned)left, (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
 491
 492        if (file->f_flags & O_APPEND)
 493                pos = i_size_read(inode);
 494        else
 495                pos = *offset;
 496
 497        ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + left);
 498        if (ret < 0)
 499                return ret;
 500
 501        ret = invalidate_inode_pages2_range(inode->i_mapping,
 502                                            pos >> PAGE_CACHE_SHIFT,
 503                                            (pos + left) >> PAGE_CACHE_SHIFT);
 504        if (ret < 0)
 505                dout("invalidate_inode_pages2_range returned %d\n", ret);
 506
 507        flags = CEPH_OSD_FLAG_ORDERSNAP |
 508                CEPH_OSD_FLAG_ONDISK |
 509                CEPH_OSD_FLAG_WRITE;
 510        if ((file->f_flags & (O_SYNC|O_DIRECT)) == 0)
 511                flags |= CEPH_OSD_FLAG_ACK;
 512        else
 513                do_sync = 1;
 514
 515        /*
 516         * we may need to do multiple writes here if we span an object
 517         * boundary.  this isn't atomic, unfortunately.  :(
 518         */
 519more:
 520        io_align = pos & ~PAGE_MASK;
 521        buf_align = (unsigned long)data & ~PAGE_MASK;
 522        len = left;
 523        if (file->f_flags & O_DIRECT) {
 524                /* write from beginning of first page, regardless of
 525                   io alignment */
 526                page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
 527                num_pages = calc_pages_for((unsigned long)data, len);
 528        } else {
 529                page_align = pos & ~PAGE_MASK;
 530                num_pages = calc_pages_for(pos, len);
 531        }
 532        req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 533                                    ceph_vino(inode), pos, &len,
 534                                    CEPH_OSD_OP_WRITE, flags,
 535                                    ci->i_snap_realm->cached_context,
 536                                    do_sync,
 537                                    ci->i_truncate_seq, ci->i_truncate_size,
 538                                    &mtime, false, 2, page_align);
 539        if (IS_ERR(req))
 540                return PTR_ERR(req);
 541
 542        if (file->f_flags & O_DIRECT) {
 543                pages = ceph_get_direct_page_vector(data, num_pages, false);
 544                if (IS_ERR(pages)) {
 545                        ret = PTR_ERR(pages);
 546                        goto out;
 547                }
 548
 549                /*
 550                 * throw out any page cache pages in this range. this
 551                 * may block.
 552                 */
 553                truncate_inode_pages_range(inode->i_mapping, pos,
 554                                           (pos+len) | (PAGE_CACHE_SIZE-1));
 555        } else {
 556                pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
 557                if (IS_ERR(pages)) {
 558                        ret = PTR_ERR(pages);
 559                        goto out;
 560                }
 561                ret = ceph_copy_user_to_page_vector(pages, data, pos, len);
 562                if (ret < 0) {
 563                        ceph_release_page_vector(pages, num_pages);
 564                        goto out;
 565                }
 566
 567                if ((file->f_flags & O_SYNC) == 0) {
 568                        /* get a second commit callback */
 569                        req->r_safe_callback = sync_write_commit;
 570                        req->r_own_pages = 1;
 571                }
 572        }
 573        req->r_pages = pages;
 574        req->r_num_pages = num_pages;
 575        req->r_inode = inode;
 576
 577        ret = ceph_osdc_start_request(&fsc->client->osdc, req, false);
 578        if (!ret) {
 579                if (req->r_safe_callback) {
 580                        /*
 581                         * Add to inode unsafe list only after we
 582                         * start_request so that a tid has been assigned.
 583                         */
 584                        spin_lock(&ci->i_unsafe_lock);
 585                        list_add_tail(&req->r_unsafe_item,
 586                                      &ci->i_unsafe_writes);
 587                        spin_unlock(&ci->i_unsafe_lock);
 588                        ceph_get_cap_refs(ci, CEPH_CAP_FILE_WR);
 589                }
 590                
 591                ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
 592                if (ret < 0 && req->r_safe_callback) {
 593                        spin_lock(&ci->i_unsafe_lock);
 594                        list_del_init(&req->r_unsafe_item);
 595                        spin_unlock(&ci->i_unsafe_lock);
 596                        ceph_put_cap_refs(ci, CEPH_CAP_FILE_WR);
 597                }
 598        }
 599
 600        if (file->f_flags & O_DIRECT)
 601                ceph_put_page_vector(pages, num_pages, false);
 602        else if (file->f_flags & O_SYNC)
 603                ceph_release_page_vector(pages, num_pages);
 604
 605out:
 606        ceph_osdc_put_request(req);
 607        if (ret == 0) {
 608                pos += len;
 609                written += len;
 610                left -= len;
 611                data += written;
 612                if (left)
 613                        goto more;
 614
 615                ret = written;
 616                *offset = pos;
 617                if (pos > i_size_read(inode))
 618                        check_caps = ceph_inode_set_size(inode, pos);
 619                if (check_caps)
 620                        ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY,
 621                                        NULL);
 622        }
 623        return ret;
 624}
 625
 626/*
 627 * Wrap generic_file_aio_read with checks for cap bits on the inode.
 628 * Atomically grab references, so that those bits are not released
 629 * back to the MDS mid-read.
 630 *
 631 * Hmm, the sync read case isn't actually async... should it be?
 632 */
 633static ssize_t ceph_aio_read(struct kiocb *iocb, const struct iovec *iov,
 634                             unsigned long nr_segs, loff_t pos)
 635{
 636        struct file *filp = iocb->ki_filp;
 637        struct ceph_file_info *fi = filp->private_data;
 638        loff_t *ppos = &iocb->ki_pos;
 639        size_t len = iov->iov_len;
 640        struct inode *inode = filp->f_dentry->d_inode;
 641        struct ceph_inode_info *ci = ceph_inode(inode);
 642        void __user *base = iov->iov_base;
 643        ssize_t ret;
 644        int want, got = 0;
 645        int checkeof = 0, read = 0;
 646
 647        dout("aio_read %p %llx.%llx %llu~%u trying to get caps on %p\n",
 648             inode, ceph_vinop(inode), pos, (unsigned)len, inode);
 649again:
 650        __ceph_do_pending_vmtruncate(inode);
 651        if (fi->fmode & CEPH_FILE_MODE_LAZY)
 652                want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
 653        else
 654                want = CEPH_CAP_FILE_CACHE;
 655        ret = ceph_get_caps(ci, CEPH_CAP_FILE_RD, want, &got, -1);
 656        if (ret < 0)
 657                goto out;
 658        dout("aio_read %p %llx.%llx %llu~%u got cap refs on %s\n",
 659             inode, ceph_vinop(inode), pos, (unsigned)len,
 660             ceph_cap_string(got));
 661
 662        if ((got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0 ||
 663            (iocb->ki_filp->f_flags & O_DIRECT) ||
 664            (inode->i_sb->s_flags & MS_SYNCHRONOUS) ||
 665            (fi->flags & CEPH_F_SYNC))
 666                /* hmm, this isn't really async... */
 667                ret = ceph_sync_read(filp, base, len, ppos, &checkeof);
 668        else
 669                ret = generic_file_aio_read(iocb, iov, nr_segs, pos);
 670
 671out:
 672        dout("aio_read %p %llx.%llx dropping cap refs on %s = %d\n",
 673             inode, ceph_vinop(inode), ceph_cap_string(got), (int)ret);
 674        ceph_put_cap_refs(ci, got);
 675
 676        if (checkeof && ret >= 0) {
 677                int statret = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE);
 678
 679                /* hit EOF or hole? */
 680                if (statret == 0 && *ppos < inode->i_size) {
 681                        dout("aio_read sync_read hit hole, ppos %lld < size %lld, reading more\n", *ppos, inode->i_size);
 682                        read += ret;
 683                        base += ret;
 684                        len -= ret;
 685                        checkeof = 0;
 686                        goto again;
 687                }
 688        }
 689        if (ret >= 0)
 690                ret += read;
 691
 692        return ret;
 693}
 694
 695/*
 696 * Take cap references to avoid releasing caps to MDS mid-write.
 697 *
 698 * If we are synchronous, and write with an old snap context, the OSD
 699 * may return EOLDSNAPC.  In that case, retry the write.. _after_
 700 * dropping our cap refs and allowing the pending snap to logically
 701 * complete _before_ this write occurs.
 702 *
 703 * If we are near ENOSPC, write synchronously.
 704 */
 705static ssize_t ceph_aio_write(struct kiocb *iocb, const struct iovec *iov,
 706                       unsigned long nr_segs, loff_t pos)
 707{
 708        struct file *file = iocb->ki_filp;
 709        struct ceph_file_info *fi = file->private_data;
 710        struct inode *inode = file->f_dentry->d_inode;
 711        struct ceph_inode_info *ci = ceph_inode(inode);
 712        struct ceph_osd_client *osdc =
 713                &ceph_sb_to_client(inode->i_sb)->client->osdc;
 714        loff_t endoff = pos + iov->iov_len;
 715        int want, got = 0;
 716        int ret, err;
 717
 718        if (ceph_snap(inode) != CEPH_NOSNAP)
 719                return -EROFS;
 720
 721retry_snap:
 722        if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL))
 723                return -ENOSPC;
 724        __ceph_do_pending_vmtruncate(inode);
 725        dout("aio_write %p %llx.%llx %llu~%u getting caps. i_size %llu\n",
 726             inode, ceph_vinop(inode), pos, (unsigned)iov->iov_len,
 727             inode->i_size);
 728        if (fi->fmode & CEPH_FILE_MODE_LAZY)
 729                want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
 730        else
 731                want = CEPH_CAP_FILE_BUFFER;
 732        ret = ceph_get_caps(ci, CEPH_CAP_FILE_WR, want, &got, endoff);
 733        if (ret < 0)
 734                goto out_put;
 735
 736        dout("aio_write %p %llx.%llx %llu~%u  got cap refs on %s\n",
 737             inode, ceph_vinop(inode), pos, (unsigned)iov->iov_len,
 738             ceph_cap_string(got));
 739
 740        if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
 741            (iocb->ki_filp->f_flags & O_DIRECT) ||
 742            (inode->i_sb->s_flags & MS_SYNCHRONOUS) ||
 743            (fi->flags & CEPH_F_SYNC)) {
 744                ret = ceph_sync_write(file, iov->iov_base, iov->iov_len,
 745                        &iocb->ki_pos);
 746        } else {
 747                /*
 748                 * buffered write; drop Fw early to avoid slow
 749                 * revocation if we get stuck on balance_dirty_pages
 750                 */
 751                int dirty;
 752
 753                spin_lock(&ci->i_ceph_lock);
 754                dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
 755                spin_unlock(&ci->i_ceph_lock);
 756                ceph_put_cap_refs(ci, got);
 757
 758                ret = generic_file_aio_write(iocb, iov, nr_segs, pos);
 759                if ((ret >= 0 || ret == -EIOCBQUEUED) &&
 760                    ((file->f_flags & O_SYNC) || IS_SYNC(file->f_mapping->host)
 761                     || ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_NEARFULL))) {
 762                        err = vfs_fsync_range(file, pos, pos + ret - 1, 1);
 763                        if (err < 0)
 764                                ret = err;
 765                }
 766
 767                if (dirty)
 768                        __mark_inode_dirty(inode, dirty);
 769                goto out;
 770        }
 771
 772        if (ret >= 0) {
 773                int dirty;
 774                spin_lock(&ci->i_ceph_lock);
 775                dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
 776                spin_unlock(&ci->i_ceph_lock);
 777                if (dirty)
 778                        __mark_inode_dirty(inode, dirty);
 779        }
 780
 781out_put:
 782        dout("aio_write %p %llx.%llx %llu~%u  dropping cap refs on %s\n",
 783             inode, ceph_vinop(inode), pos, (unsigned)iov->iov_len,
 784             ceph_cap_string(got));
 785        ceph_put_cap_refs(ci, got);
 786
 787out:
 788        if (ret == -EOLDSNAPC) {
 789                dout("aio_write %p %llx.%llx %llu~%u got EOLDSNAPC, retrying\n",
 790                     inode, ceph_vinop(inode), pos, (unsigned)iov->iov_len);
 791                goto retry_snap;
 792        }
 793
 794        return ret;
 795}
 796
 797/*
 798 * llseek.  be sure to verify file size on SEEK_END.
 799 */
 800static loff_t ceph_llseek(struct file *file, loff_t offset, int origin)
 801{
 802        struct inode *inode = file->f_mapping->host;
 803        int ret;
 804
 805        mutex_lock(&inode->i_mutex);
 806        __ceph_do_pending_vmtruncate(inode);
 807
 808        if (origin == SEEK_END || origin == SEEK_DATA || origin == SEEK_HOLE) {
 809                ret = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE);
 810                if (ret < 0) {
 811                        offset = ret;
 812                        goto out;
 813                }
 814        }
 815
 816        switch (origin) {
 817        case SEEK_END:
 818                offset += inode->i_size;
 819                break;
 820        case SEEK_CUR:
 821                /*
 822                 * Here we special-case the lseek(fd, 0, SEEK_CUR)
 823                 * position-querying operation.  Avoid rewriting the "same"
 824                 * f_pos value back to the file because a concurrent read(),
 825                 * write() or lseek() might have altered it
 826                 */
 827                if (offset == 0) {
 828                        offset = file->f_pos;
 829                        goto out;
 830                }
 831                offset += file->f_pos;
 832                break;
 833        case SEEK_DATA:
 834                if (offset >= inode->i_size) {
 835                        ret = -ENXIO;
 836                        goto out;
 837                }
 838                break;
 839        case SEEK_HOLE:
 840                if (offset >= inode->i_size) {
 841                        ret = -ENXIO;
 842                        goto out;
 843                }
 844                offset = inode->i_size;
 845                break;
 846        }
 847
 848        if (offset < 0 || offset > inode->i_sb->s_maxbytes) {
 849                offset = -EINVAL;
 850                goto out;
 851        }
 852
 853        /* Special lock needed here? */
 854        if (offset != file->f_pos) {
 855                file->f_pos = offset;
 856                file->f_version = 0;
 857        }
 858
 859out:
 860        mutex_unlock(&inode->i_mutex);
 861        return offset;
 862}
 863
 864const struct file_operations ceph_file_fops = {
 865        .open = ceph_open,
 866        .release = ceph_release,
 867        .llseek = ceph_llseek,
 868        .read = do_sync_read,
 869        .write = do_sync_write,
 870        .aio_read = ceph_aio_read,
 871        .aio_write = ceph_aio_write,
 872        .mmap = ceph_mmap,
 873        .fsync = ceph_fsync,
 874        .lock = ceph_lock,
 875        .flock = ceph_flock,
 876        .splice_read = generic_file_splice_read,
 877        .splice_write = generic_file_splice_write,
 878        .unlocked_ioctl = ceph_ioctl,
 879        .compat_ioctl   = ceph_ioctl,
 880};
 881
 882
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.