fs/ceph/addr.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 #include <linux/ceph/ceph_debug.h>
   3
   4 #include <linux/backing-dev.h>
   5 #include <linux/fs.h>
   6 #include <linux/mm.h>
   7 #include <linux/pagemap.h>
   8 #include <linux/writeback.h>    /* generic_writepages */
   9 #include <linux/slab.h>
  10 #include <linux/pagevec.h>
  11 #include <linux/task_io_accounting_ops.h>
  12 #include <linux/signal.h>
  13 #include <linux/iversion.h>
  14 #include <linux/ktime.h>
  15 #include <linux/netfs.h>
  16
  17 #include "super.h"
  18 #include "mds_client.h"
  19 #include "cache.h"
  20 #include "metric.h"
  21 #include <linux/ceph/osd_client.h>
  22 #include <linux/ceph/striper.h>
  23
  24 /*
  25  * Ceph address space ops.
  26  *
  27  * There are a few funny things going on here.
  28  *
  29  * The page->private field is used to reference a struct
  30  * ceph_snap_context for _every_ dirty page.  This indicates which
  31  * snapshot the page was logically dirtied in, and thus which snap
  32  * context needs to be associated with the osd write during writeback.
  33  *
  34  * Similarly, struct ceph_inode_info maintains a set of counters to
  35  * count dirty pages on the inode.  In the absence of snapshots,
  36  * i_wrbuffer_ref == i_wrbuffer_ref_head == the dirty page count.
  37  *
  38  * When a snapshot is taken (that is, when the client receives
  39  * notification that a snapshot was taken), each inode with caps and
  40  * with dirty pages (dirty pages implies there is a cap) gets a new
  41  * ceph_cap_snap in the i_cap_snaps list (which is sorted in ascending
  42  * order, new snaps go to the tail).  The i_wrbuffer_ref_head count is
  43  * moved to capsnap->dirty. (Unless a sync write is currently in
  44  * progress.  In that case, the capsnap is said to be "pending", new
  45  * writes cannot start, and the capsnap isn't "finalized" until the
  46  * write completes (or fails) and a final size/mtime for the inode for
  47  * that snap can be settled upon.)  i_wrbuffer_ref_head is reset to 0.
  48  *
  49  * On writeback, we must submit writes to the osd IN SNAP ORDER.  So,
  50  * we look for the first capsnap in i_cap_snaps and write out pages in
  51  * that snap context _only_.  Then we move on to the next capsnap,
  52  * eventually reaching the "live" or "head" context (i.e., pages that
  53  * are not yet snapped) and are writing the most recently dirtied
  54  * pages.
  55  *
  56  * Invalidate and so forth must take care to ensure the dirty page
  57  * accounting is preserved.
  58  */
  59
  60 #define CONGESTION_ON_THRESH(congestion_kb) (congestion_kb >> (PAGE_SHIFT-10))
  61 #define CONGESTION_OFF_THRESH(congestion_kb)                            \
  62         (CONGESTION_ON_THRESH(congestion_kb) -                          \
  63          (CONGESTION_ON_THRESH(congestion_kb) >> 2))
  64
  65 static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
  66                                         struct page *page, void **_fsdata);
  67
  68 static inline struct ceph_snap_context *page_snap_context(struct page *page)
  69 {
  70         if (PagePrivate(page))
  71                 return (void *)page->private;
  72         return NULL;
  73 }
  74
  75 /*
  76  * Dirty a page.  Optimistically adjust accounting, on the assumption
  77  * that we won't race with invalidate.  If we do, readjust.
  78  */
  79 static int ceph_set_page_dirty(struct page *page)
  80 {
  81         struct address_space *mapping = page->mapping;
  82         struct inode *inode;
  83         struct ceph_inode_info *ci;
  84         struct ceph_snap_context *snapc;
  85
  86         if (PageDirty(page)) {
  87                 dout("%p set_page_dirty %p idx %lu -- already dirty\n",
  88                      mapping->host, page, page->index);
  89                 BUG_ON(!PagePrivate(page));
  90                 return 0;
  91         }
  92
  93         inode = mapping->host;
  94         ci = ceph_inode(inode);
  95
  96         /* dirty the head */
  97         spin_lock(&ci->i_ceph_lock);
  98         BUG_ON(ci->i_wr_ref == 0); // caller should hold Fw reference
  99         if (__ceph_have_pending_cap_snap(ci)) {
 100                 struct ceph_cap_snap *capsnap =
 101                                 list_last_entry(&ci->i_cap_snaps,
 102                                                 struct ceph_cap_snap,
 103                                                 ci_item);
 104                 snapc = ceph_get_snap_context(capsnap->context);
 105                 capsnap->dirty_pages++;
 106         } else {
 107                 BUG_ON(!ci->i_head_snapc);
 108                 snapc = ceph_get_snap_context(ci->i_head_snapc);
 109                 ++ci->i_wrbuffer_ref_head;
 110         }
 111         if (ci->i_wrbuffer_ref == 0)
 112                 ihold(inode);
 113         ++ci->i_wrbuffer_ref;
 114         dout("%p set_page_dirty %p idx %lu head %d/%d -> %d/%d "
 115              "snapc %p seq %lld (%d snaps)\n",
 116              mapping->host, page, page->index,
 117              ci->i_wrbuffer_ref-1, ci->i_wrbuffer_ref_head-1,
 118              ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
 119              snapc, snapc->seq, snapc->num_snaps);
 120         spin_unlock(&ci->i_ceph_lock);
 121
 122         /*
 123          * Reference snap context in page->private.  Also set
 124          * PagePrivate so that we get invalidatepage callback.
 125          */
 126         BUG_ON(PagePrivate(page));
 127         attach_page_private(page, snapc);
 128
 129         return __set_page_dirty_nobuffers(page);
 130 }
 131
 132 /*
 133  * If we are truncating the full page (i.e. offset == 0), adjust the
 134  * dirty page counters appropriately.  Only called if there is private
 135  * data on the page.
 136  */
 137 static void ceph_invalidatepage(struct page *page, unsigned int offset,
 138                                 unsigned int length)
 139 {
 140         struct inode *inode;
 141         struct ceph_inode_info *ci;
 142         struct ceph_snap_context *snapc;
 143
 144         wait_on_page_fscache(page);
 145
 146         inode = page->mapping->host;
 147         ci = ceph_inode(inode);
 148
 149         if (offset != 0 || length != thp_size(page)) {
 150                 dout("%p invalidatepage %p idx %lu partial dirty page %u~%u\n",
 151                      inode, page, page->index, offset, length);
 152                 return;
 153         }
 154
 155         WARN_ON(!PageLocked(page));
 156         if (!PagePrivate(page))
 157                 return;
 158
 159         dout("%p invalidatepage %p idx %lu full dirty page\n",
 160              inode, page, page->index);
 161
 162         snapc = detach_page_private(page);
 163         ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
 164         ceph_put_snap_context(snapc);
 165 }
 166
 167 static int ceph_releasepage(struct page *page, gfp_t gfp)
 168 {
 169         dout("%p releasepage %p idx %lu (%sdirty)\n", page->mapping->host,
 170              page, page->index, PageDirty(page) ? "" : "not ");
 171
 172         if (PageFsCache(page)) {
 173                 if (!(gfp & __GFP_DIRECT_RECLAIM) || !(gfp & __GFP_FS))
 174                         return 0;
 175                 wait_on_page_fscache(page);
 176         }
 177         return !PagePrivate(page);
 178 }
 179
 180 static void ceph_netfs_expand_readahead(struct netfs_read_request *rreq)
 181 {
 182         struct inode *inode = rreq->mapping->host;
 183         struct ceph_inode_info *ci = ceph_inode(inode);
 184         struct ceph_file_layout *lo = &ci->i_layout;
 185         u32 blockoff;
 186         u64 blockno;
 187
 188         /* Expand the start downward */
 189         blockno = div_u64_rem(rreq->start, lo->stripe_unit, &blockoff);
 190         rreq->start = blockno * lo->stripe_unit;
 191         rreq->len += blockoff;
 192
 193         /* Now, round up the length to the next block */
 194         rreq->len = roundup(rreq->len, lo->stripe_unit);
 195 }
 196
 197 static bool ceph_netfs_clamp_length(struct netfs_read_subrequest *subreq)
 198 {
 199         struct inode *inode = subreq->rreq->mapping->host;
 200         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 201         struct ceph_inode_info *ci = ceph_inode(inode);
 202         u64 objno, objoff;
 203         u32 xlen;
 204
 205         /* Truncate the extent at the end of the current block */
 206         ceph_calc_file_object_mapping(&ci->i_layout, subreq->start, subreq->len,
 207                                       &objno, &objoff, &xlen);
 208         subreq->len = min(xlen, fsc->mount_options->rsize);
 209         return true;
 210 }
 211
 212 static void finish_netfs_read(struct ceph_osd_request *req)
 213 {
 214         struct ceph_fs_client *fsc = ceph_inode_to_client(req->r_inode);
 215         struct ceph_osd_data *osd_data = osd_req_op_extent_osd_data(req, 0);
 216         struct netfs_read_subrequest *subreq = req->r_priv;
 217         int num_pages;
 218         int err = req->r_result;
 219
 220         ceph_update_read_metrics(&fsc->mdsc->metric, req->r_start_latency,
 221                                  req->r_end_latency, osd_data->length, err);
 222
 223         dout("%s: result %d subreq->len=%zu i_size=%lld\n", __func__, req->r_result,
 224              subreq->len, i_size_read(req->r_inode));
 225
 226         /* no object means success but no data */
 227         if (err == -ENOENT)
 228                 err = 0;
 229         else if (err == -EBLOCKLISTED)
 230                 fsc->blocklisted = true;
 231
 232         if (err >= 0 && err < subreq->len)
 233                 __set_bit(NETFS_SREQ_CLEAR_TAIL, &subreq->flags);
 234
 235         netfs_subreq_terminated(subreq, err, true);
 236
 237         num_pages = calc_pages_for(osd_data->alignment, osd_data->length);
 238         ceph_put_page_vector(osd_data->pages, num_pages, false);
 239         iput(req->r_inode);
 240 }
 241
 242 static void ceph_netfs_issue_op(struct netfs_read_subrequest *subreq)
 243 {
 244         struct netfs_read_request *rreq = subreq->rreq;
 245         struct inode *inode = rreq->mapping->host;
 246         struct ceph_inode_info *ci = ceph_inode(inode);
 247         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 248         struct ceph_osd_request *req;
 249         struct ceph_vino vino = ceph_vino(inode);
 250         struct iov_iter iter;
 251         struct page **pages;
 252         size_t page_off;
 253         int err = 0;
 254         u64 len = subreq->len;
 255
 256         req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, vino, subreq->start, &len,
 257                         0, 1, CEPH_OSD_OP_READ,
 258                         CEPH_OSD_FLAG_READ | fsc->client->osdc.client->options->read_from_replica,
 259                         NULL, ci->i_truncate_seq, ci->i_truncate_size, false);
 260         if (IS_ERR(req)) {
 261                 err = PTR_ERR(req);
 262                 req = NULL;
 263                 goto out;
 264         }
 265
 266         dout("%s: pos=%llu orig_len=%zu len=%llu\n", __func__, subreq->start, subreq->len, len);
 267         iov_iter_xarray(&iter, READ, &rreq->mapping->i_pages, subreq->start, len);
 268         err = iov_iter_get_pages_alloc(&iter, &pages, len, &page_off);
 269         if (err < 0) {
 270                 dout("%s: iov_ter_get_pages_alloc returned %d\n", __func__, err);
 271                 goto out;
 272         }
 273
 274         /* should always give us a page-aligned read */
 275         WARN_ON_ONCE(page_off);
 276         len = err;
 277
 278         osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0, false, false);
 279         req->r_callback = finish_netfs_read;
 280         req->r_priv = subreq;
 281         req->r_inode = inode;
 282         ihold(inode);
 283
 284         err = ceph_osdc_start_request(req->r_osdc, req, false);
 285         if (err)
 286                 iput(inode);
 287 out:
 288         ceph_osdc_put_request(req);
 289         if (err)
 290                 netfs_subreq_terminated(subreq, err, false);
 291         dout("%s: result %d\n", __func__, err);
 292 }
 293
 294 static void ceph_init_rreq(struct netfs_read_request *rreq, struct file *file)
 295 {
 296 }
 297
 298 static void ceph_readahead_cleanup(struct address_space *mapping, void *priv)
 299 {
 300         struct inode *inode = mapping->host;
 301         struct ceph_inode_info *ci = ceph_inode(inode);
 302         int got = (uintptr_t)priv;
 303
 304         if (got)
 305                 ceph_put_cap_refs(ci, got);
 306 }
 307
 308 static const struct netfs_read_request_ops ceph_netfs_read_ops = {
 309         .init_rreq              = ceph_init_rreq,
 310         .is_cache_enabled       = ceph_is_cache_enabled,
 311         .begin_cache_operation  = ceph_begin_cache_operation,
 312         .issue_op               = ceph_netfs_issue_op,
 313         .expand_readahead       = ceph_netfs_expand_readahead,
 314         .clamp_length           = ceph_netfs_clamp_length,
 315         .check_write_begin      = ceph_netfs_check_write_begin,
 316         .cleanup                = ceph_readahead_cleanup,
 317 };
 318
 319 /* read a single page, without unlocking it. */
 320 static int ceph_readpage(struct file *file, struct page *page)
 321 {
 322         struct inode *inode = file_inode(file);
 323         struct ceph_inode_info *ci = ceph_inode(inode);
 324         struct ceph_vino vino = ceph_vino(inode);
 325         u64 off = page_offset(page);
 326         u64 len = thp_size(page);
 327
 328         if (ci->i_inline_version != CEPH_INLINE_NONE) {
 329                 /*
 330                  * Uptodate inline data should have been added
 331                  * into page cache while getting Fcr caps.
 332                  */
 333                 if (off == 0) {
 334                         unlock_page(page);
 335                         return -EINVAL;
 336                 }
 337                 zero_user_segment(page, 0, thp_size(page));
 338                 SetPageUptodate(page);
 339                 unlock_page(page);
 340                 return 0;
 341         }
 342
 343         dout("readpage ino %llx.%llx file %p off %llu len %llu page %p index %lu\n",
 344              vino.ino, vino.snap, file, off, len, page, page->index);
 345
 346         return netfs_readpage(file, page, &ceph_netfs_read_ops, NULL);
 347 }
 348
 349 static void ceph_readahead(struct readahead_control *ractl)
 350 {
 351         struct inode *inode = file_inode(ractl->file);
 352         struct ceph_file_info *fi = ractl->file->private_data;
 353         struct ceph_rw_context *rw_ctx;
 354         int got = 0;
 355         int ret = 0;
 356
 357         if (ceph_inode(inode)->i_inline_version != CEPH_INLINE_NONE)
 358                 return;
 359
 360         rw_ctx = ceph_find_rw_context(fi);
 361         if (!rw_ctx) {
 362                 /*
 363                  * readahead callers do not necessarily hold Fcb caps
 364                  * (e.g. fadvise, madvise).
 365                  */
 366                 int want = CEPH_CAP_FILE_CACHE;
 367
 368                 ret = ceph_try_get_caps(inode, CEPH_CAP_FILE_RD, want, true, &got);
 369                 if (ret < 0)
 370                         dout("start_read %p, error getting cap\n", inode);
 371                 else if (!(got & want))
 372                         dout("start_read %p, no cache cap\n", inode);
 373
 374                 if (ret <= 0)
 375                         return;
 376         }
 377         netfs_readahead(ractl, &ceph_netfs_read_ops, (void *)(uintptr_t)got);
 378 }
 379
 380 struct ceph_writeback_ctl
 381 {
 382         loff_t i_size;
 383         u64 truncate_size;
 384         u32 truncate_seq;
 385         bool size_stable;
 386         bool head_snapc;
 387 };
 388
 389 /*
 390  * Get ref for the oldest snapc for an inode with dirty data... that is, the
 391  * only snap context we are allowed to write back.
 392  */
 393 static struct ceph_snap_context *
 394 get_oldest_context(struct inode *inode, struct ceph_writeback_ctl *ctl,
 395                    struct ceph_snap_context *page_snapc)
 396 {
 397         struct ceph_inode_info *ci = ceph_inode(inode);
 398         struct ceph_snap_context *snapc = NULL;
 399         struct ceph_cap_snap *capsnap = NULL;
 400
 401         spin_lock(&ci->i_ceph_lock);
 402         list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
 403                 dout(" cap_snap %p snapc %p has %d dirty pages\n", capsnap,
 404                      capsnap->context, capsnap->dirty_pages);
 405                 if (!capsnap->dirty_pages)
 406                         continue;
 407
 408                 /* get i_size, truncate_{seq,size} for page_snapc? */
 409                 if (snapc && capsnap->context != page_snapc)
 410                         continue;
 411
 412                 if (ctl) {
 413                         if (capsnap->writing) {
 414                                 ctl->i_size = i_size_read(inode);
 415                                 ctl->size_stable = false;
 416                         } else {
 417                                 ctl->i_size = capsnap->size;
 418                                 ctl->size_stable = true;
 419                         }
 420                         ctl->truncate_size = capsnap->truncate_size;
 421                         ctl->truncate_seq = capsnap->truncate_seq;
 422                         ctl->head_snapc = false;
 423                 }
 424
 425                 if (snapc)
 426                         break;
 427
 428                 snapc = ceph_get_snap_context(capsnap->context);
 429                 if (!page_snapc ||
 430                     page_snapc == snapc ||
 431                     page_snapc->seq > snapc->seq)
 432                         break;
 433         }
 434         if (!snapc && ci->i_wrbuffer_ref_head) {
 435                 snapc = ceph_get_snap_context(ci->i_head_snapc);
 436                 dout(" head snapc %p has %d dirty pages\n",
 437                      snapc, ci->i_wrbuffer_ref_head);
 438                 if (ctl) {
 439                         ctl->i_size = i_size_read(inode);
 440                         ctl->truncate_size = ci->i_truncate_size;
 441                         ctl->truncate_seq = ci->i_truncate_seq;
 442                         ctl->size_stable = false;
 443                         ctl->head_snapc = true;
 444                 }
 445         }
 446         spin_unlock(&ci->i_ceph_lock);
 447         return snapc;
 448 }
 449
 450 static u64 get_writepages_data_length(struct inode *inode,
 451                                       struct page *page, u64 start)
 452 {
 453         struct ceph_inode_info *ci = ceph_inode(inode);
 454         struct ceph_snap_context *snapc = page_snap_context(page);
 455         struct ceph_cap_snap *capsnap = NULL;
 456         u64 end = i_size_read(inode);
 457
 458         if (snapc != ci->i_head_snapc) {
 459                 bool found = false;
 460                 spin_lock(&ci->i_ceph_lock);
 461                 list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
 462                         if (capsnap->context == snapc) {
 463                                 if (!capsnap->writing)
 464                                         end = capsnap->size;
 465                                 found = true;
 466                                 break;
 467                         }
 468                 }
 469                 spin_unlock(&ci->i_ceph_lock);
 470                 WARN_ON(!found);
 471         }
 472         if (end > page_offset(page) + thp_size(page))
 473                 end = page_offset(page) + thp_size(page);
 474         return end > start ? end - start : 0;
 475 }
 476
 477 /*
 478  * Write a single page, but leave the page locked.
 479  *
 480  * If we get a write error, mark the mapping for error, but still adjust the
 481  * dirty page accounting (i.e., page is no longer dirty).
 482  */
 483 static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 484 {
 485         struct inode *inode = page->mapping->host;
 486         struct ceph_inode_info *ci = ceph_inode(inode);
 487         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 488         struct ceph_snap_context *snapc, *oldest;
 489         loff_t page_off = page_offset(page);
 490         int err;
 491         loff_t len = thp_size(page);
 492         struct ceph_writeback_ctl ceph_wbc;
 493         struct ceph_osd_client *osdc = &fsc->client->osdc;
 494         struct ceph_osd_request *req;
 495
 496         dout("writepage %p idx %lu\n", page, page->index);
 497
 498         /* verify this is a writeable snap context */
 499         snapc = page_snap_context(page);
 500         if (!snapc) {
 501                 dout("writepage %p page %p not dirty?\n", inode, page);
 502                 return 0;
 503         }
 504         oldest = get_oldest_context(inode, &ceph_wbc, snapc);
 505         if (snapc->seq > oldest->seq) {
 506                 dout("writepage %p page %p snapc %p not writeable - noop\n",
 507                      inode, page, snapc);
 508                 /* we should only noop if called by kswapd */
 509                 WARN_ON(!(current->flags & PF_MEMALLOC));
 510                 ceph_put_snap_context(oldest);
 511                 redirty_page_for_writepage(wbc, page);
 512                 return 0;
 513         }
 514         ceph_put_snap_context(oldest);
 515
 516         /* is this a partial page at end of file? */
 517         if (page_off >= ceph_wbc.i_size) {
 518                 dout("%p page eof %llu\n", page, ceph_wbc.i_size);
 519                 page->mapping->a_ops->invalidatepage(page, 0, thp_size(page));
 520                 return 0;
 521         }
 522
 523         if (ceph_wbc.i_size < page_off + len)
 524                 len = ceph_wbc.i_size - page_off;
 525
 526         dout("writepage %p page %p index %lu on %llu~%llu snapc %p seq %lld\n",
 527              inode, page, page->index, page_off, len, snapc, snapc->seq);
 528
 529         if (atomic_long_inc_return(&fsc->writeback_count) >
 530             CONGESTION_ON_THRESH(fsc->mount_options->congestion_kb))
 531                 set_bdi_congested(inode_to_bdi(inode), BLK_RW_ASYNC);
 532
 533         set_page_writeback(page);
 534         req = ceph_osdc_new_request(osdc, &ci->i_layout, ceph_vino(inode), page_off, &len, 0, 1,
 535                                     CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE, snapc,
 536                                     ceph_wbc.truncate_seq, ceph_wbc.truncate_size,
 537                                     true);
 538         if (IS_ERR(req)) {
 539                 redirty_page_for_writepage(wbc, page);
 540                 end_page_writeback(page);
 541                 return PTR_ERR(req);
 542         }
 543
 544         /* it may be a short write due to an object boundary */
 545         WARN_ON_ONCE(len > thp_size(page));
 546         osd_req_op_extent_osd_data_pages(req, 0, &page, len, 0, false, false);
 547         dout("writepage %llu~%llu (%llu bytes)\n", page_off, len, len);
 548
 549         req->r_mtime = inode->i_mtime;
 550         err = ceph_osdc_start_request(osdc, req, true);
 551         if (!err)
 552                 err = ceph_osdc_wait_request(osdc, req);
 553
 554         ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
 555                                   req->r_end_latency, len, err);
 556
 557         ceph_osdc_put_request(req);
 558         if (err == 0)
 559                 err = len;
 560
 561         if (err < 0) {
 562                 struct writeback_control tmp_wbc;
 563                 if (!wbc)
 564                         wbc = &tmp_wbc;
 565                 if (err == -ERESTARTSYS) {
 566                         /* killed by SIGKILL */
 567                         dout("writepage interrupted page %p\n", page);
 568                         redirty_page_for_writepage(wbc, page);
 569                         end_page_writeback(page);
 570                         return err;
 571                 }
 572                 if (err == -EBLOCKLISTED)
 573                         fsc->blocklisted = true;
 574                 dout("writepage setting page/mapping error %d %p\n",
 575                      err, page);
 576                 mapping_set_error(&inode->i_data, err);
 577                 wbc->pages_skipped++;
 578         } else {
 579                 dout("writepage cleaned page %p\n", page);
 580                 err = 0;  /* vfs expects us to return 0 */
 581         }
 582         oldest = detach_page_private(page);
 583         WARN_ON_ONCE(oldest != snapc);
 584         end_page_writeback(page);
 585         ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
 586         ceph_put_snap_context(snapc);  /* page's reference */
 587
 588         if (atomic_long_dec_return(&fsc->writeback_count) <
 589             CONGESTION_OFF_THRESH(fsc->mount_options->congestion_kb))
 590                 clear_bdi_congested(inode_to_bdi(inode), BLK_RW_ASYNC);
 591
 592         return err;
 593 }
 594
 595 static int ceph_writepage(struct page *page, struct writeback_control *wbc)
 596 {
 597         int err;
 598         struct inode *inode = page->mapping->host;
 599         BUG_ON(!inode);
 600         ihold(inode);
 601         err = writepage_nounlock(page, wbc);
 602         if (err == -ERESTARTSYS) {
 603                 /* direct memory reclaimer was killed by SIGKILL. return 0
 604                  * to prevent caller from setting mapping/page error */
 605                 err = 0;
 606         }
 607         unlock_page(page);
 608         iput(inode);
 609         return err;
 610 }
 611
 612 /*
 613  * async writeback completion handler.
 614  *
 615  * If we get an error, set the mapping error bit, but not the individual
 616  * page error bits.
 617  */
 618 static void writepages_finish(struct ceph_osd_request *req)
 619 {
 620         struct inode *inode = req->r_inode;
 621         struct ceph_inode_info *ci = ceph_inode(inode);
 622         struct ceph_osd_data *osd_data;
 623         struct page *page;
 624         int num_pages, total_pages = 0;
 625         int i, j;
 626         int rc = req->r_result;
 627         struct ceph_snap_context *snapc = req->r_snapc;
 628         struct address_space *mapping = inode->i_mapping;
 629         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 630         unsigned int len = 0;
 631         bool remove_page;
 632
 633         dout("writepages_finish %p rc %d\n", inode, rc);
 634         if (rc < 0) {
 635                 mapping_set_error(mapping, rc);
 636                 ceph_set_error_write(ci);
 637                 if (rc == -EBLOCKLISTED)
 638                         fsc->blocklisted = true;
 639         } else {
 640                 ceph_clear_error_write(ci);
 641         }
 642
 643         /*
 644          * We lost the cache cap, need to truncate the page before
 645          * it is unlocked, otherwise we'd truncate it later in the
 646          * page truncation thread, possibly losing some data that
 647          * raced its way in
 648          */
 649         remove_page = !(ceph_caps_issued(ci) &
 650                         (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO));
 651
 652         /* clean all pages */
 653         for (i = 0; i < req->r_num_ops; i++) {
 654                 if (req->r_ops[i].op != CEPH_OSD_OP_WRITE)
 655                         break;
 656
 657                 osd_data = osd_req_op_extent_osd_data(req, i);
 658                 BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_PAGES);
 659                 len += osd_data->length;
 660                 num_pages = calc_pages_for((u64)osd_data->alignment,
 661                                            (u64)osd_data->length);
 662                 total_pages += num_pages;
 663                 for (j = 0; j < num_pages; j++) {
 664                         page = osd_data->pages[j];
 665                         BUG_ON(!page);
 666                         WARN_ON(!PageUptodate(page));
 667
 668                         if (atomic_long_dec_return(&fsc->writeback_count) <
 669                              CONGESTION_OFF_THRESH(
 670                                         fsc->mount_options->congestion_kb))
 671                                 clear_bdi_congested(inode_to_bdi(inode),
 672                                                     BLK_RW_ASYNC);
 673
 674                         ceph_put_snap_context(detach_page_private(page));
 675                         end_page_writeback(page);
 676                         dout("unlocking %p\n", page);
 677
 678                         if (remove_page)
 679                                 generic_error_remove_page(inode->i_mapping,
 680                                                           page);
 681
 682                         unlock_page(page);
 683                 }
 684                 dout("writepages_finish %p wrote %llu bytes cleaned %d pages\n",
 685                      inode, osd_data->length, rc >= 0 ? num_pages : 0);
 686
 687                 release_pages(osd_data->pages, num_pages);
 688         }
 689
 690         ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
 691                                   req->r_end_latency, len, rc);
 692
 693         ceph_put_wrbuffer_cap_refs(ci, total_pages, snapc);
 694
 695         osd_data = osd_req_op_extent_osd_data(req, 0);
 696         if (osd_data->pages_from_pool)
 697                 mempool_free(osd_data->pages, ceph_wb_pagevec_pool);
 698         else
 699                 kfree(osd_data->pages);
 700         ceph_osdc_put_request(req);
 701 }
 702
 703 /*
 704  * initiate async writeback
 705  */
 706 static int ceph_writepages_start(struct address_space *mapping,
 707                                  struct writeback_control *wbc)
 708 {
 709         struct inode *inode = mapping->host;
 710         struct ceph_inode_info *ci = ceph_inode(inode);
 711         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 712         struct ceph_vino vino = ceph_vino(inode);
 713         pgoff_t index, start_index, end = -1;
 714         struct ceph_snap_context *snapc = NULL, *last_snapc = NULL, *pgsnapc;
 715         struct pagevec pvec;
 716         int rc = 0;
 717         unsigned int wsize = i_blocksize(inode);
 718         struct ceph_osd_request *req = NULL;
 719         struct ceph_writeback_ctl ceph_wbc;
 720         bool should_loop, range_whole = false;
 721         bool done = false;
 722
 723         dout("writepages_start %p (mode=%s)\n", inode,
 724              wbc->sync_mode == WB_SYNC_NONE ? "NONE" :
 725              (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD"));
 726
 727         if (READ_ONCE(fsc->mount_state) >= CEPH_MOUNT_SHUTDOWN) {
 728                 if (ci->i_wrbuffer_ref > 0) {
 729                         pr_warn_ratelimited(
 730                                 "writepage_start %p %lld forced umount\n",
 731                                 inode, ceph_ino(inode));
 732                 }
 733                 mapping_set_error(mapping, -EIO);
 734                 return -EIO; /* we're in a forced umount, don't write! */
 735         }
 736         if (fsc->mount_options->wsize < wsize)
 737                 wsize = fsc->mount_options->wsize;
 738
 739         pagevec_init(&pvec);
 740
 741         start_index = wbc->range_cyclic ? mapping->writeback_index : 0;
 742         index = start_index;
 743
 744 retry:
 745         /* find oldest snap context with dirty data */
 746         snapc = get_oldest_context(inode, &ceph_wbc, NULL);
 747         if (!snapc) {
 748                 /* hmm, why does writepages get called when there
 749                    is no dirty data? */
 750                 dout(" no snap context with dirty data?\n");
 751                 goto out;
 752         }
 753         dout(" oldest snapc is %p seq %lld (%d snaps)\n",
 754              snapc, snapc->seq, snapc->num_snaps);
 755
 756         should_loop = false;
 757         if (ceph_wbc.head_snapc && snapc != last_snapc) {
 758                 /* where to start/end? */
 759                 if (wbc->range_cyclic) {
 760                         index = start_index;
 761                         end = -1;
 762                         if (index > 0)
 763                                 should_loop = true;
 764                         dout(" cyclic, start at %lu\n", index);
 765                 } else {
 766                         index = wbc->range_start >> PAGE_SHIFT;
 767                         end = wbc->range_end >> PAGE_SHIFT;
 768                         if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
 769                                 range_whole = true;
 770                         dout(" not cyclic, %lu to %lu\n", index, end);
 771                 }
 772         } else if (!ceph_wbc.head_snapc) {
 773                 /* Do not respect wbc->range_{start,end}. Dirty pages
 774                  * in that range can be associated with newer snapc.
 775                  * They are not writeable until we write all dirty pages
 776                  * associated with 'snapc' get written */
 777                 if (index > 0)
 778                         should_loop = true;
 779                 dout(" non-head snapc, range whole\n");
 780         }
 781
 782         ceph_put_snap_context(last_snapc);
 783         last_snapc = snapc;
 784
 785         while (!done && index <= end) {
 786                 int num_ops = 0, op_idx;
 787                 unsigned i, pvec_pages, max_pages, locked_pages = 0;
 788                 struct page **pages = NULL, **data_pages;
 789                 struct page *page;
 790                 pgoff_t strip_unit_end = 0;
 791                 u64 offset = 0, len = 0;
 792                 bool from_pool = false;
 793
 794                 max_pages = wsize >> PAGE_SHIFT;
 795
 796 get_more_pages:
 797                 pvec_pages = pagevec_lookup_range_tag(&pvec, mapping, &index,
 798                                                 end, PAGECACHE_TAG_DIRTY);
 799                 dout("pagevec_lookup_range_tag got %d\n", pvec_pages);
 800                 if (!pvec_pages && !locked_pages)
 801                         break;
 802                 for (i = 0; i < pvec_pages && locked_pages < max_pages; i++) {
 803                         page = pvec.pages[i];
 804                         dout("? %p idx %lu\n", page, page->index);
 805                         if (locked_pages == 0)
 806                                 lock_page(page);  /* first page */
 807                         else if (!trylock_page(page))
 808                                 break;
 809
 810                         /* only dirty pages, or our accounting breaks */
 811                         if (unlikely(!PageDirty(page)) ||
 812                             unlikely(page->mapping != mapping)) {
 813                                 dout("!dirty or !mapping %p\n", page);
 814                                 unlock_page(page);
 815                                 continue;
 816                         }
 817                         /* only if matching snap context */
 818                         pgsnapc = page_snap_context(page);
 819                         if (pgsnapc != snapc) {
 820                                 dout("page snapc %p %lld != oldest %p %lld\n",
 821                                      pgsnapc, pgsnapc->seq, snapc, snapc->seq);
 822                                 if (!should_loop &&
 823                                     !ceph_wbc.head_snapc &&
 824                                     wbc->sync_mode != WB_SYNC_NONE)
 825                                         should_loop = true;
 826                                 unlock_page(page);
 827                                 continue;
 828                         }
 829                         if (page_offset(page) >= ceph_wbc.i_size) {
 830                                 dout("%p page eof %llu\n",
 831                                      page, ceph_wbc.i_size);
 832                                 if ((ceph_wbc.size_stable ||
 833                                     page_offset(page) >= i_size_read(inode)) &&
 834                                     clear_page_dirty_for_io(page))
 835                                         mapping->a_ops->invalidatepage(page,
 836                                                                 0, thp_size(page));
 837                                 unlock_page(page);
 838                                 continue;
 839                         }
 840                         if (strip_unit_end && (page->index > strip_unit_end)) {
 841                                 dout("end of strip unit %p\n", page);
 842                                 unlock_page(page);
 843                                 break;
 844                         }
 845                         if (PageWriteback(page)) {
 846                                 if (wbc->sync_mode == WB_SYNC_NONE) {
 847                                         dout("%p under writeback\n", page);
 848                                         unlock_page(page);
 849                                         continue;
 850                                 }
 851                                 dout("waiting on writeback %p\n", page);
 852                                 wait_on_page_writeback(page);
 853                         }
 854
 855                         if (!clear_page_dirty_for_io(page)) {
 856                                 dout("%p !clear_page_dirty_for_io\n", page);
 857                                 unlock_page(page);
 858                                 continue;
 859                         }
 860
 861                         /*
 862                          * We have something to write.  If this is
 863                          * the first locked page this time through,
 864                          * calculate max possinle write size and
 865                          * allocate a page array
 866                          */
 867                         if (locked_pages == 0) {
 868                                 u64 objnum;
 869                                 u64 objoff;
 870                                 u32 xlen;
 871
 872                                 /* prepare async write request */
 873                                 offset = (u64)page_offset(page);
 874                                 ceph_calc_file_object_mapping(&ci->i_layout,
 875                                                               offset, wsize,
 876                                                               &objnum, &objoff,
 877                                                               &xlen);
 878                                 len = xlen;
 879
 880                                 num_ops = 1;
 881                                 strip_unit_end = page->index +
 882                                         ((len - 1) >> PAGE_SHIFT);
 883
 884                                 BUG_ON(pages);
 885                                 max_pages = calc_pages_for(0, (u64)len);
 886                                 pages = kmalloc_array(max_pages,
 887                                                       sizeof(*pages),
 888                                                       GFP_NOFS);
 889                                 if (!pages) {
 890                                         from_pool = true;
 891                                         pages = mempool_alloc(ceph_wb_pagevec_pool, GFP_NOFS);
 892                                         BUG_ON(!pages);
 893                                 }
 894
 895                                 len = 0;
 896                         } else if (page->index !=
 897                                    (offset + len) >> PAGE_SHIFT) {
 898                                 if (num_ops >= (from_pool ?  CEPH_OSD_SLAB_OPS :
 899                                                              CEPH_OSD_MAX_OPS)) {
 900                                         redirty_page_for_writepage(wbc, page);
 901                                         unlock_page(page);
 902                                         break;
 903                                 }
 904
 905                                 num_ops++;
 906                                 offset = (u64)page_offset(page);
 907                                 len = 0;
 908                         }
 909
 910                         /* note position of first page in pvec */
 911                         dout("%p will write page %p idx %lu\n",
 912                              inode, page, page->index);
 913
 914                         if (atomic_long_inc_return(&fsc->writeback_count) >
 915                             CONGESTION_ON_THRESH(
 916                                     fsc->mount_options->congestion_kb)) {
 917                                 set_bdi_congested(inode_to_bdi(inode),
 918                                                   BLK_RW_ASYNC);
 919                         }
 920
 921
 922                         pages[locked_pages++] = page;
 923                         pvec.pages[i] = NULL;
 924
 925                         len += thp_size(page);
 926                 }
 927
 928                 /* did we get anything? */
 929                 if (!locked_pages)
 930                         goto release_pvec_pages;
 931                 if (i) {
 932                         unsigned j, n = 0;
 933                         /* shift unused page to beginning of pvec */
 934                         for (j = 0; j < pvec_pages; j++) {
 935                                 if (!pvec.pages[j])
 936                                         continue;
 937                                 if (n < j)
 938                                         pvec.pages[n] = pvec.pages[j];
 939                                 n++;
 940                         }
 941                         pvec.nr = n;
 942
 943                         if (pvec_pages && i == pvec_pages &&
 944                             locked_pages < max_pages) {
 945                                 dout("reached end pvec, trying for more\n");
 946                                 pagevec_release(&pvec);
 947                                 goto get_more_pages;
 948                         }
 949                 }
 950
 951 new_request:
 952                 offset = page_offset(pages[0]);
 953                 len = wsize;
 954
 955                 req = ceph_osdc_new_request(&fsc->client->osdc,
 956                                         &ci->i_layout, vino,
 957                                         offset, &len, 0, num_ops,
 958                                         CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
 959                                         snapc, ceph_wbc.truncate_seq,
 960                                         ceph_wbc.truncate_size, false);
 961                 if (IS_ERR(req)) {
 962                         req = ceph_osdc_new_request(&fsc->client->osdc,
 963                                                 &ci->i_layout, vino,
 964                                                 offset, &len, 0,
 965                                                 min(num_ops,
 966                                                     CEPH_OSD_SLAB_OPS),
 967                                                 CEPH_OSD_OP_WRITE,
 968                                                 CEPH_OSD_FLAG_WRITE,
 969                                                 snapc, ceph_wbc.truncate_seq,
 970                                                 ceph_wbc.truncate_size, true);
 971                         BUG_ON(IS_ERR(req));
 972                 }
 973                 BUG_ON(len < page_offset(pages[locked_pages - 1]) +
 974                              thp_size(page) - offset);
 975
 976                 req->r_callback = writepages_finish;
 977                 req->r_inode = inode;
 978
 979                 /* Format the osd request message and submit the write */
 980                 len = 0;
 981                 data_pages = pages;
 982                 op_idx = 0;
 983                 for (i = 0; i < locked_pages; i++) {
 984                         u64 cur_offset = page_offset(pages[i]);
 985                         if (offset + len != cur_offset) {
 986                                 if (op_idx + 1 == req->r_num_ops)
 987                                         break;
 988                                 osd_req_op_extent_dup_last(req, op_idx,
 989                                                            cur_offset - offset);
 990                                 dout("writepages got pages at %llu~%llu\n",
 991                                      offset, len);
 992                                 osd_req_op_extent_osd_data_pages(req, op_idx,
 993                                                         data_pages, len, 0,
 994                                                         from_pool, false);
 995                                 osd_req_op_extent_update(req, op_idx, len);
 996
 997                                 len = 0;
 998                                 offset = cur_offset;
 999                                 data_pages = pages + i;
1000                                 op_idx++;
1001                         }
1002
1003                         set_page_writeback(pages[i]);
1004                         len += thp_size(page);
1005                 }
1006
1007                 if (ceph_wbc.size_stable) {
1008                         len = min(len, ceph_wbc.i_size - offset);
1009                 } else if (i == locked_pages) {
1010                         /* writepages_finish() clears writeback pages
1011                          * according to the data length, so make sure
1012                          * data length covers all locked pages */
1013                         u64 min_len = len + 1 - thp_size(page);
1014                         len = get_writepages_data_length(inode, pages[i - 1],
1015                                                          offset);
1016                         len = max(len, min_len);
1017                 }
1018                 dout("writepages got pages at %llu~%llu\n", offset, len);
1019
1020                 osd_req_op_extent_osd_data_pages(req, op_idx, data_pages, len,
1021                                                  0, from_pool, false);
1022                 osd_req_op_extent_update(req, op_idx, len);
1023
1024                 BUG_ON(op_idx + 1 != req->r_num_ops);
1025
1026                 from_pool = false;
1027                 if (i < locked_pages) {
1028                         BUG_ON(num_ops <= req->r_num_ops);
1029                         num_ops -= req->r_num_ops;
1030                         locked_pages -= i;
1031
1032                         /* allocate new pages array for next request */
1033                         data_pages = pages;
1034                         pages = kmalloc_array(locked_pages, sizeof(*pages),
1035                                               GFP_NOFS);
1036                         if (!pages) {
1037                                 from_pool = true;
1038                                 pages = mempool_alloc(ceph_wb_pagevec_pool, GFP_NOFS);
1039                                 BUG_ON(!pages);
1040                         }
1041                         memcpy(pages, data_pages + i,
1042                                locked_pages * sizeof(*pages));
1043                         memset(data_pages + i, 0,
1044                                locked_pages * sizeof(*pages));
1045                 } else {
1046                         BUG_ON(num_ops != req->r_num_ops);
1047                         index = pages[i - 1]->index + 1;
1048                         /* request message now owns the pages array */
1049                         pages = NULL;
1050                 }
1051
1052                 req->r_mtime = inode->i_mtime;
1053                 rc = ceph_osdc_start_request(&fsc->client->osdc, req, true);
1054                 BUG_ON(rc);
1055                 req = NULL;
1056
1057                 wbc->nr_to_write -= i;
1058                 if (pages)
1059                         goto new_request;
1060
1061                 /*
1062                  * We stop writing back only if we are not doing
1063                  * integrity sync. In case of integrity sync we have to
1064                  * keep going until we have written all the pages
1065                  * we tagged for writeback prior to entering this loop.
1066                  */
1067                 if (wbc->nr_to_write <= 0 && wbc->sync_mode == WB_SYNC_NONE)
1068                         done = true;
1069
1070 release_pvec_pages:
1071                 dout("pagevec_release on %d pages (%p)\n", (int)pvec.nr,
1072                      pvec.nr ? pvec.pages[0] : NULL);
1073                 pagevec_release(&pvec);
1074         }
1075
1076         if (should_loop && !done) {
1077                 /* more to do; loop back to beginning of file */
1078                 dout("writepages looping back to beginning of file\n");
1079                 end = start_index - 1; /* OK even when start_index == 0 */
1080
1081                 /* to write dirty pages associated with next snapc,
1082                  * we need to wait until current writes complete */
1083                 if (wbc->sync_mode != WB_SYNC_NONE &&
1084                     start_index == 0 && /* all dirty pages were checked */
1085                     !ceph_wbc.head_snapc) {
1086                         struct page *page;
1087                         unsigned i, nr;
1088                         index = 0;
1089                         while ((index <= end) &&
1090                                (nr = pagevec_lookup_tag(&pvec, mapping, &index,
1091                                                 PAGECACHE_TAG_WRITEBACK))) {
1092                                 for (i = 0; i < nr; i++) {
1093                                         page = pvec.pages[i];
1094                                         if (page_snap_context(page) != snapc)
1095                                                 continue;
1096                                         wait_on_page_writeback(page);
1097                                 }
1098                                 pagevec_release(&pvec);
1099                                 cond_resched();
1100                         }
1101                 }
1102
1103                 start_index = 0;
1104                 index = 0;
1105                 goto retry;
1106         }
1107
1108         if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
1109                 mapping->writeback_index = index;
1110
1111 out:
1112         ceph_osdc_put_request(req);
1113         ceph_put_snap_context(last_snapc);
1114         dout("writepages dend - startone, rc = %d\n", rc);
1115         return rc;
1116 }
1117
1118
1119
1120 /*
1121  * See if a given @snapc is either writeable, or already written.
1122  */
1123 static int context_is_writeable_or_written(struct inode *inode,
1124                                            struct ceph_snap_context *snapc)
1125 {
1126         struct ceph_snap_context *oldest = get_oldest_context(inode, NULL, NULL);
1127         int ret = !oldest || snapc->seq <= oldest->seq;
1128
1129         ceph_put_snap_context(oldest);
1130         return ret;
1131 }
1132
1133 /**
1134  * ceph_find_incompatible - find an incompatible context and return it
1135  * @page: page being dirtied
1136  *
1137  * We are only allowed to write into/dirty a page if the page is
1138  * clean, or already dirty within the same snap context. Returns a
1139  * conflicting context if there is one, NULL if there isn't, or a
1140  * negative error code on other errors.
1141  *
1142  * Must be called with page lock held.
1143  */
1144 static struct ceph_snap_context *
1145 ceph_find_incompatible(struct page *page)
1146 {
1147         struct inode *inode = page->mapping->host;
1148         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1149         struct ceph_inode_info *ci = ceph_inode(inode);
1150
1151         if (READ_ONCE(fsc->mount_state) >= CEPH_MOUNT_SHUTDOWN) {
1152                 dout(" page %p forced umount\n", page);
1153                 return ERR_PTR(-EIO);
1154         }
1155
1156         for (;;) {
1157                 struct ceph_snap_context *snapc, *oldest;
1158
1159                 wait_on_page_writeback(page);
1160
1161                 snapc = page_snap_context(page);
1162                 if (!snapc || snapc == ci->i_head_snapc)
1163                         break;
1164
1165                 /*
1166                  * this page is already dirty in another (older) snap
1167                  * context!  is it writeable now?
1168                  */
1169                 oldest = get_oldest_context(inode, NULL, NULL);
1170                 if (snapc->seq > oldest->seq) {
1171                         /* not writeable -- return it for the caller to deal with */
1172                         ceph_put_snap_context(oldest);
1173                         dout(" page %p snapc %p not current or oldest\n", page, snapc);
1174                         return ceph_get_snap_context(snapc);
1175                 }
1176                 ceph_put_snap_context(oldest);
1177
1178                 /* yay, writeable, do it now (without dropping page lock) */
1179                 dout(" page %p snapc %p not current, but oldest\n", page, snapc);
1180                 if (clear_page_dirty_for_io(page)) {
1181                         int r = writepage_nounlock(page, NULL);
1182                         if (r < 0)
1183                                 return ERR_PTR(r);
1184                 }
1185         }
1186         return NULL;
1187 }
1188
1189 static int ceph_netfs_check_write_begin(struct file *file, loff_t pos, unsigned int len,
1190                                         struct page *page, void **_fsdata)
1191 {
1192         struct inode *inode = file_inode(file);
1193         struct ceph_inode_info *ci = ceph_inode(inode);
1194         struct ceph_snap_context *snapc;
1195
1196         snapc = ceph_find_incompatible(page);
1197         if (snapc) {
1198                 int r;
1199
1200                 unlock_page(page);
1201                 put_page(page);
1202                 if (IS_ERR(snapc))
1203                         return PTR_ERR(snapc);
1204
1205                 ceph_queue_writeback(inode);
1206                 r = wait_event_killable(ci->i_cap_wq,
1207                                         context_is_writeable_or_written(inode, snapc));
1208                 ceph_put_snap_context(snapc);
1209                 return r == 0 ? -EAGAIN : r;
1210         }
1211         return 0;
1212 }
1213
1214 /*
1215  * We are only allowed to write into/dirty the page if the page is
1216  * clean, or already dirty within the same snap context.
1217  */
1218 static int ceph_write_begin(struct file *file, struct address_space *mapping,
1219                             loff_t pos, unsigned len, unsigned flags,
1220                             struct page **pagep, void **fsdata)
1221 {
1222         struct inode *inode = file_inode(file);
1223         struct ceph_inode_info *ci = ceph_inode(inode);
1224         struct page *page = NULL;
1225         pgoff_t index = pos >> PAGE_SHIFT;
1226         int r;
1227
1228         /*
1229          * Uninlining should have already been done and everything updated, EXCEPT
1230          * for inline_version sent to the MDS.
1231          */
1232         if (ci->i_inline_version != CEPH_INLINE_NONE) {
1233                 page = grab_cache_page_write_begin(mapping, index, flags);
1234                 if (!page)
1235                         return -ENOMEM;
1236
1237                 /*
1238                  * The inline_version on a new inode is set to 1. If that's the
1239                  * case, then the page is brand new and isn't yet Uptodate.
1240                  */
1241                 r = 0;
1242                 if (index == 0 && ci->i_inline_version != 1) {
1243                         if (!PageUptodate(page)) {
1244                                 WARN_ONCE(1, "ceph: write_begin called on still-inlined inode (inline_version %llu)!\n",
1245                                           ci->i_inline_version);
1246                                 r = -EINVAL;
1247                         }
1248                         goto out;
1249                 }
1250                 zero_user_segment(page, 0, thp_size(page));
1251                 SetPageUptodate(page);
1252                 goto out;
1253         }
1254
1255         r = netfs_write_begin(file, inode->i_mapping, pos, len, 0, &page, NULL,
1256                               &ceph_netfs_read_ops, NULL);
1257 out:
1258         if (r == 0)
1259                 wait_on_page_fscache(page);
1260         if (r < 0) {
1261                 if (page)
1262                         put_page(page);
1263         } else {
1264                 WARN_ON_ONCE(!PageLocked(page));
1265                 *pagep = page;
1266         }
1267         return r;
1268 }
1269
1270 /*
1271  * we don't do anything in here that simple_write_end doesn't do
1272  * except adjust dirty page accounting
1273  */
1274 static int ceph_write_end(struct file *file, struct address_space *mapping,
1275                           loff_t pos, unsigned len, unsigned copied,
1276                           struct page *page, void *fsdata)
1277 {
1278         struct inode *inode = file_inode(file);
1279         bool check_cap = false;
1280
1281         dout("write_end file %p inode %p page %p %d~%d (%d)\n", file,
1282              inode, page, (int)pos, (int)copied, (int)len);
1283
1284         if (!PageUptodate(page)) {
1285                 /* just return that nothing was copied on a short copy */
1286                 if (copied < len) {
1287                         copied = 0;
1288                         goto out;
1289                 }
1290                 SetPageUptodate(page);
1291         }
1292
1293         /* did file size increase? */
1294         if (pos+copied > i_size_read(inode))
1295                 check_cap = ceph_inode_set_size(inode, pos+copied);
1296
1297         set_page_dirty(page);
1298
1299 out:
1300         unlock_page(page);
1301         put_page(page);
1302
1303         if (check_cap)
1304                 ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY, NULL);
1305
1306         return copied;
1307 }
1308
1309 /*
1310  * we set .direct_IO to indicate direct io is supported, but since we
1311  * intercept O_DIRECT reads and writes early, this function should
1312  * never get called.
1313  */
1314 static ssize_t ceph_direct_io(struct kiocb *iocb, struct iov_iter *iter)
1315 {
1316         WARN_ON(1);
1317         return -EINVAL;
1318 }
1319
1320 const struct address_space_operations ceph_aops = {
1321         .readpage = ceph_readpage,
1322         .readahead = ceph_readahead,
1323         .writepage = ceph_writepage,
1324         .writepages = ceph_writepages_start,
1325         .write_begin = ceph_write_begin,
1326         .write_end = ceph_write_end,
1327         .set_page_dirty = ceph_set_page_dirty,
1328         .invalidatepage = ceph_invalidatepage,
1329         .releasepage = ceph_releasepage,
1330         .direct_IO = ceph_direct_io,
1331 };
1332
1333 static void ceph_block_sigs(sigset_t *oldset)
1334 {
1335         sigset_t mask;
1336         siginitsetinv(&mask, sigmask(SIGKILL));
1337         sigprocmask(SIG_BLOCK, &mask, oldset);
1338 }
1339
1340 static void ceph_restore_sigs(sigset_t *oldset)
1341 {
1342         sigprocmask(SIG_SETMASK, oldset, NULL);
1343 }
1344
1345 /*
1346  * vm ops
1347  */
1348 static vm_fault_t ceph_filemap_fault(struct vm_fault *vmf)
1349 {
1350         struct vm_area_struct *vma = vmf->vma;
1351         struct inode *inode = file_inode(vma->vm_file);
1352         struct ceph_inode_info *ci = ceph_inode(inode);
1353         struct ceph_file_info *fi = vma->vm_file->private_data;
1354         loff_t off = (loff_t)vmf->pgoff << PAGE_SHIFT;
1355         int want, got, err;
1356         sigset_t oldset;
1357         vm_fault_t ret = VM_FAULT_SIGBUS;
1358
1359         ceph_block_sigs(&oldset);
1360
1361         dout("filemap_fault %p %llx.%llx %llu trying to get caps\n",
1362              inode, ceph_vinop(inode), off);
1363         if (fi->fmode & CEPH_FILE_MODE_LAZY)
1364                 want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
1365         else
1366                 want = CEPH_CAP_FILE_CACHE;
1367
1368         got = 0;
1369         err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_RD, want, -1, &got);
1370         if (err < 0)
1371                 goto out_restore;
1372
1373         dout("filemap_fault %p %llu got cap refs on %s\n",
1374              inode, off, ceph_cap_string(got));
1375
1376         if ((got & (CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO)) ||
1377             ci->i_inline_version == CEPH_INLINE_NONE) {
1378                 CEPH_DEFINE_RW_CONTEXT(rw_ctx, got);
1379                 ceph_add_rw_context(fi, &rw_ctx);
1380                 ret = filemap_fault(vmf);
1381                 ceph_del_rw_context(fi, &rw_ctx);
1382                 dout("filemap_fault %p %llu drop cap refs %s ret %x\n",
1383                      inode, off, ceph_cap_string(got), ret);
1384         } else
1385                 err = -EAGAIN;
1386
1387         ceph_put_cap_refs(ci, got);
1388
1389         if (err != -EAGAIN)
1390                 goto out_restore;
1391
1392         /* read inline data */
1393         if (off >= PAGE_SIZE) {
1394                 /* does not support inline data > PAGE_SIZE */
1395                 ret = VM_FAULT_SIGBUS;
1396         } else {
1397                 struct address_space *mapping = inode->i_mapping;
1398                 struct page *page;
1399
1400                 filemap_invalidate_lock_shared(mapping);
1401                 page = find_or_create_page(mapping, 0,
1402                                 mapping_gfp_constraint(mapping, ~__GFP_FS));
1403                 if (!page) {
1404                         ret = VM_FAULT_OOM;
1405                         goto out_inline;
1406                 }
1407                 err = __ceph_do_getattr(inode, page,
1408                                          CEPH_STAT_CAP_INLINE_DATA, true);
1409                 if (err < 0 || off >= i_size_read(inode)) {
1410                         unlock_page(page);
1411                         put_page(page);
1412                         ret = vmf_error(err);
1413                         goto out_inline;
1414                 }
1415                 if (err < PAGE_SIZE)
1416                         zero_user_segment(page, err, PAGE_SIZE);
1417                 else
1418                         flush_dcache_page(page);
1419                 SetPageUptodate(page);
1420                 vmf->page = page;
1421                 ret = VM_FAULT_MAJOR | VM_FAULT_LOCKED;
1422 out_inline:
1423                 filemap_invalidate_unlock_shared(mapping);
1424                 dout("filemap_fault %p %llu read inline data ret %x\n",
1425                      inode, off, ret);
1426         }
1427 out_restore:
1428         ceph_restore_sigs(&oldset);
1429         if (err < 0)
1430                 ret = vmf_error(err);
1431
1432         return ret;
1433 }
1434
1435 static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf)
1436 {
1437         struct vm_area_struct *vma = vmf->vma;
1438         struct inode *inode = file_inode(vma->vm_file);
1439         struct ceph_inode_info *ci = ceph_inode(inode);
1440         struct ceph_file_info *fi = vma->vm_file->private_data;
1441         struct ceph_cap_flush *prealloc_cf;
1442         struct page *page = vmf->page;
1443         loff_t off = page_offset(page);
1444         loff_t size = i_size_read(inode);
1445         size_t len;
1446         int want, got, err;
1447         sigset_t oldset;
1448         vm_fault_t ret = VM_FAULT_SIGBUS;
1449
1450         prealloc_cf = ceph_alloc_cap_flush();
1451         if (!prealloc_cf)
1452                 return VM_FAULT_OOM;
1453
1454         sb_start_pagefault(inode->i_sb);
1455         ceph_block_sigs(&oldset);
1456
1457         if (ci->i_inline_version != CEPH_INLINE_NONE) {
1458                 struct page *locked_page = NULL;
1459                 if (off == 0) {
1460                         lock_page(page);
1461                         locked_page = page;
1462                 }
1463                 err = ceph_uninline_data(vma->vm_file, locked_page);
1464                 if (locked_page)
1465                         unlock_page(locked_page);
1466                 if (err < 0)
1467                         goto out_free;
1468         }
1469
1470         if (off + thp_size(page) <= size)
1471                 len = thp_size(page);
1472         else
1473                 len = offset_in_thp(page, size);
1474
1475         dout("page_mkwrite %p %llx.%llx %llu~%zd getting caps i_size %llu\n",
1476              inode, ceph_vinop(inode), off, len, size);
1477         if (fi->fmode & CEPH_FILE_MODE_LAZY)
1478                 want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
1479         else
1480                 want = CEPH_CAP_FILE_BUFFER;
1481
1482         got = 0;
1483         err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_WR, want, off + len, &got);
1484         if (err < 0)
1485                 goto out_free;
1486
1487         dout("page_mkwrite %p %llu~%zd got cap refs on %s\n",
1488              inode, off, len, ceph_cap_string(got));
1489
1490         /* Update time before taking page lock */
1491         file_update_time(vma->vm_file);
1492         inode_inc_iversion_raw(inode);
1493
1494         do {
1495                 struct ceph_snap_context *snapc;
1496
1497                 lock_page(page);
1498
1499                 if (page_mkwrite_check_truncate(page, inode) < 0) {
1500                         unlock_page(page);
1501                         ret = VM_FAULT_NOPAGE;
1502                         break;
1503                 }
1504
1505                 snapc = ceph_find_incompatible(page);
1506                 if (!snapc) {
1507                         /* success.  we'll keep the page locked. */
1508                         set_page_dirty(page);
1509                         ret = VM_FAULT_LOCKED;
1510                         break;
1511                 }
1512
1513                 unlock_page(page);
1514
1515                 if (IS_ERR(snapc)) {
1516                         ret = VM_FAULT_SIGBUS;
1517                         break;
1518                 }
1519
1520                 ceph_queue_writeback(inode);
1521                 err = wait_event_killable(ci->i_cap_wq,
1522                                 context_is_writeable_or_written(inode, snapc));
1523                 ceph_put_snap_context(snapc);
1524         } while (err == 0);
1525
1526         if (ret == VM_FAULT_LOCKED ||
1527             ci->i_inline_version != CEPH_INLINE_NONE) {
1528                 int dirty;
1529                 spin_lock(&ci->i_ceph_lock);
1530                 ci->i_inline_version = CEPH_INLINE_NONE;
1531                 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
1532                                                &prealloc_cf);
1533                 spin_unlock(&ci->i_ceph_lock);
1534                 if (dirty)
1535                         __mark_inode_dirty(inode, dirty);
1536         }
1537
1538         dout("page_mkwrite %p %llu~%zd dropping cap refs on %s ret %x\n",
1539              inode, off, len, ceph_cap_string(got), ret);
1540         ceph_put_cap_refs_async(ci, got);
1541 out_free:
1542         ceph_restore_sigs(&oldset);
1543         sb_end_pagefault(inode->i_sb);
1544         ceph_free_cap_flush(prealloc_cf);
1545         if (err < 0)
1546                 ret = vmf_error(err);
1547         return ret;
1548 }
1549
1550 void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
1551                            char *data, size_t len)
1552 {
1553         struct address_space *mapping = inode->i_mapping;
1554         struct page *page;
1555
1556         if (locked_page) {
1557                 page = locked_page;
1558         } else {
1559                 if (i_size_read(inode) == 0)
1560                         return;
1561                 page = find_or_create_page(mapping, 0,
1562                                            mapping_gfp_constraint(mapping,
1563                                            ~__GFP_FS));
1564                 if (!page)
1565                         return;
1566                 if (PageUptodate(page)) {
1567                         unlock_page(page);
1568                         put_page(page);
1569                         return;
1570                 }
1571         }
1572
1573         dout("fill_inline_data %p %llx.%llx len %zu locked_page %p\n",
1574              inode, ceph_vinop(inode), len, locked_page);
1575
1576         if (len > 0) {
1577                 void *kaddr = kmap_atomic(page);
1578                 memcpy(kaddr, data, len);
1579                 kunmap_atomic(kaddr);
1580         }
1581
1582         if (page != locked_page) {
1583                 if (len < PAGE_SIZE)
1584                         zero_user_segment(page, len, PAGE_SIZE);
1585                 else
1586                         flush_dcache_page(page);
1587
1588                 SetPageUptodate(page);
1589                 unlock_page(page);
1590                 put_page(page);
1591         }
1592 }
1593
1594 int ceph_uninline_data(struct file *filp, struct page *locked_page)
1595 {
1596         struct inode *inode = file_inode(filp);
1597         struct ceph_inode_info *ci = ceph_inode(inode);
1598         struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1599         struct ceph_osd_request *req;
1600         struct page *page = NULL;
1601         u64 len, inline_version;
1602         int err = 0;
1603         bool from_pagecache = false;
1604
1605         spin_lock(&ci->i_ceph_lock);
1606         inline_version = ci->i_inline_version;
1607         spin_unlock(&ci->i_ceph_lock);
1608
1609         dout("uninline_data %p %llx.%llx inline_version %llu\n",
1610              inode, ceph_vinop(inode), inline_version);
1611
1612         if (inline_version == 1 || /* initial version, no data */
1613             inline_version == CEPH_INLINE_NONE)
1614                 goto out;
1615
1616         if (locked_page) {
1617                 page = locked_page;
1618                 WARN_ON(!PageUptodate(page));
1619         } else if (ceph_caps_issued(ci) &
1620                    (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) {
1621                 page = find_get_page(inode->i_mapping, 0);
1622                 if (page) {
1623                         if (PageUptodate(page)) {
1624                                 from_pagecache = true;
1625                                 lock_page(page);
1626                         } else {
1627                                 put_page(page);
1628                                 page = NULL;
1629                         }
1630                 }
1631         }
1632
1633         if (page) {
1634                 len = i_size_read(inode);
1635                 if (len > PAGE_SIZE)
1636                         len = PAGE_SIZE;
1637         } else {
1638                 page = __page_cache_alloc(GFP_NOFS);
1639                 if (!page) {
1640                         err = -ENOMEM;
1641                         goto out;
1642                 }
1643                 err = __ceph_do_getattr(inode, page,
1644                                         CEPH_STAT_CAP_INLINE_DATA, true);
1645                 if (err < 0) {
1646                         /* no inline data */
1647                         if (err == -ENODATA)
1648                                 err = 0;
1649                         goto out;
1650                 }
1651                 len = err;
1652         }
1653
1654         req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1655                                     ceph_vino(inode), 0, &len, 0, 1,
1656                                     CEPH_OSD_OP_CREATE, CEPH_OSD_FLAG_WRITE,
1657                                     NULL, 0, 0, false);
1658         if (IS_ERR(req)) {
1659                 err = PTR_ERR(req);
1660                 goto out;
1661         }
1662
1663         req->r_mtime = inode->i_mtime;
1664         err = ceph_osdc_start_request(&fsc->client->osdc, req, false);
1665         if (!err)
1666                 err = ceph_osdc_wait_request(&fsc->client->osdc, req);
1667         ceph_osdc_put_request(req);
1668         if (err < 0)
1669                 goto out;
1670
1671         req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1672                                     ceph_vino(inode), 0, &len, 1, 3,
1673                                     CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
1674                                     NULL, ci->i_truncate_seq,
1675                                     ci->i_truncate_size, false);
1676         if (IS_ERR(req)) {
1677                 err = PTR_ERR(req);
1678                 goto out;
1679         }
1680
1681         osd_req_op_extent_osd_data_pages(req, 1, &page, len, 0, false, false);
1682
1683         {
1684                 __le64 xattr_buf = cpu_to_le64(inline_version);
1685                 err = osd_req_op_xattr_init(req, 0, CEPH_OSD_OP_CMPXATTR,
1686                                             "inline_version", &xattr_buf,
1687                                             sizeof(xattr_buf),
1688                                             CEPH_OSD_CMPXATTR_OP_GT,
1689                                             CEPH_OSD_CMPXATTR_MODE_U64);
1690                 if (err)
1691                         goto out_put;
1692         }
1693
1694         {
1695                 char xattr_buf[32];
1696                 int xattr_len = snprintf(xattr_buf, sizeof(xattr_buf),
1697                                          "%llu", inline_version);
1698                 err = osd_req_op_xattr_init(req, 2, CEPH_OSD_OP_SETXATTR,
1699                                             "inline_version",
1700                                             xattr_buf, xattr_len, 0, 0);
1701                 if (err)
1702                         goto out_put;
1703         }
1704
1705         req->r_mtime = inode->i_mtime;
1706         err = ceph_osdc_start_request(&fsc->client->osdc, req, false);
1707         if (!err)
1708                 err = ceph_osdc_wait_request(&fsc->client->osdc, req);
1709
1710         ceph_update_write_metrics(&fsc->mdsc->metric, req->r_start_latency,
1711                                   req->r_end_latency, len, err);
1712
1713 out_put:
1714         ceph_osdc_put_request(req);
1715         if (err == -ECANCELED)
1716                 err = 0;
1717 out:
1718         if (page && page != locked_page) {
1719                 if (from_pagecache) {
1720                         unlock_page(page);
1721                         put_page(page);
1722                 } else
1723                         __free_pages(page, 0);
1724         }
1725
1726         dout("uninline_data %p %llx.%llx inline_version %llu = %d\n",
1727              inode, ceph_vinop(inode), inline_version, err);
1728         return err;
1729 }
1730
1731 static const struct vm_operations_struct ceph_vmops = {
1732         .fault          = ceph_filemap_fault,
1733         .page_mkwrite   = ceph_page_mkwrite,
1734 };
1735
1736 int ceph_mmap(struct file *file, struct vm_area_struct *vma)
1737 {
1738         struct address_space *mapping = file->f_mapping;
1739
1740         if (!mapping->a_ops->readpage)
1741                 return -ENOEXEC;
1742         file_accessed(file);
1743         vma->vm_ops = &ceph_vmops;
1744         return 0;
1745 }
1746
1747 enum {
1748         POOL_READ       = 1,
1749         POOL_WRITE      = 2,
1750 };
1751
1752 static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
1753                                 s64 pool, struct ceph_string *pool_ns)
1754 {
1755         struct ceph_fs_client *fsc = ceph_inode_to_client(&ci->vfs_inode);
1756         struct ceph_mds_client *mdsc = fsc->mdsc;
1757         struct ceph_osd_request *rd_req = NULL, *wr_req = NULL;
1758         struct rb_node **p, *parent;
1759         struct ceph_pool_perm *perm;
1760         struct page **pages;
1761         size_t pool_ns_len;
1762         int err = 0, err2 = 0, have = 0;
1763
1764         down_read(&mdsc->pool_perm_rwsem);
1765         p = &mdsc->pool_perm_tree.rb_node;
1766         while (*p) {
1767                 perm = rb_entry(*p, struct ceph_pool_perm, node);
1768                 if (pool < perm->pool)
1769                         p = &(*p)->rb_left;
1770                 else if (pool > perm->pool)
1771                         p = &(*p)->rb_right;
1772                 else {
1773                         int ret = ceph_compare_string(pool_ns,
1774                                                 perm->pool_ns,
1775                                                 perm->pool_ns_len);
1776                         if (ret < 0)
1777                                 p = &(*p)->rb_left;
1778                         else if (ret > 0)
1779                                 p = &(*p)->rb_right;
1780                         else {
1781                                 have = perm->perm;
1782                                 break;
1783                         }
1784                 }
1785         }
1786         up_read(&mdsc->pool_perm_rwsem);
1787         if (*p)
1788                 goto out;
1789
1790         if (pool_ns)
1791                 dout("__ceph_pool_perm_get pool %lld ns %.*s no perm cached\n",
1792                      pool, (int)pool_ns->len, pool_ns->str);
1793         else
1794                 dout("__ceph_pool_perm_get pool %lld no perm cached\n", pool);
1795
1796         down_write(&mdsc->pool_perm_rwsem);
1797         p = &mdsc->pool_perm_tree.rb_node;
1798         parent = NULL;
1799         while (*p) {
1800                 parent = *p;
1801                 perm = rb_entry(parent, struct ceph_pool_perm, node);
1802                 if (pool < perm->pool)
1803                         p = &(*p)->rb_left;
1804                 else if (pool > perm->pool)
1805                         p = &(*p)->rb_right;
1806                 else {
1807                         int ret = ceph_compare_string(pool_ns,
1808                                                 perm->pool_ns,
1809                                                 perm->pool_ns_len);
1810                         if (ret < 0)
1811                                 p = &(*p)->rb_left;
1812                         else if (ret > 0)
1813                                 p = &(*p)->rb_right;
1814                         else {
1815                                 have = perm->perm;
1816                                 break;
1817                         }
1818                 }
1819         }
1820         if (*p) {
1821                 up_write(&mdsc->pool_perm_rwsem);
1822                 goto out;
1823         }
1824
1825         rd_req = ceph_osdc_alloc_request(&fsc->client->osdc, NULL,
1826                                          1, false, GFP_NOFS);
1827         if (!rd_req) {
1828                 err = -ENOMEM;
1829                 goto out_unlock;
1830         }
1831
1832         rd_req->r_flags = CEPH_OSD_FLAG_READ;
1833         osd_req_op_init(rd_req, 0, CEPH_OSD_OP_STAT, 0);
1834         rd_req->r_base_oloc.pool = pool;
1835         if (pool_ns)
1836                 rd_req->r_base_oloc.pool_ns = ceph_get_string(pool_ns);
1837         ceph_oid_printf(&rd_req->r_base_oid, "%llx.00000000", ci->i_vino.ino);
1838
1839         err = ceph_osdc_alloc_messages(rd_req, GFP_NOFS);
1840         if (err)
1841                 goto out_unlock;
1842
1843         wr_req = ceph_osdc_alloc_request(&fsc->client->osdc, NULL,
1844                                          1, false, GFP_NOFS);
1845         if (!wr_req) {
1846                 err = -ENOMEM;
1847                 goto out_unlock;
1848         }
1849
1850         wr_req->r_flags = CEPH_OSD_FLAG_WRITE;
1851         osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL);
1852         ceph_oloc_copy(&wr_req->r_base_oloc, &rd_req->r_base_oloc);
1853         ceph_oid_copy(&wr_req->r_base_oid, &rd_req->r_base_oid);
1854
1855         err = ceph_osdc_alloc_messages(wr_req, GFP_NOFS);
1856         if (err)
1857                 goto out_unlock;
1858
1859         /* one page should be large enough for STAT data */
1860         pages = ceph_alloc_page_vector(1, GFP_KERNEL);
1861         if (IS_ERR(pages)) {
1862                 err = PTR_ERR(pages);
1863                 goto out_unlock;
1864         }
1865
1866         osd_req_op_raw_data_in_pages(rd_req, 0, pages, PAGE_SIZE,
1867                                      0, false, true);
1868         err = ceph_osdc_start_request(&fsc->client->osdc, rd_req, false);
1869
1870         wr_req->r_mtime = ci->vfs_inode.i_mtime;
1871         err2 = ceph_osdc_start_request(&fsc->client->osdc, wr_req, false);
1872
1873         if (!err)
1874                 err = ceph_osdc_wait_request(&fsc->client->osdc, rd_req);
1875         if (!err2)
1876                 err2 = ceph_osdc_wait_request(&fsc->client->osdc, wr_req);
1877
1878         if (err >= 0 || err == -ENOENT)
1879                 have |= POOL_READ;
1880         else if (err != -EPERM) {
1881                 if (err == -EBLOCKLISTED)
1882                         fsc->blocklisted = true;
1883                 goto out_unlock;
1884         }
1885
1886         if (err2 == 0 || err2 == -EEXIST)
1887                 have |= POOL_WRITE;
1888         else if (err2 != -EPERM) {
1889                 if (err2 == -EBLOCKLISTED)
1890                         fsc->blocklisted = true;
1891                 err = err2;
1892                 goto out_unlock;
1893         }
1894
1895         pool_ns_len = pool_ns ? pool_ns->len : 0;
1896         perm = kmalloc(sizeof(*perm) + pool_ns_len + 1, GFP_NOFS);
1897         if (!perm) {
1898                 err = -ENOMEM;
1899                 goto out_unlock;
1900         }
1901
1902         perm->pool = pool;
1903         perm->perm = have;
1904         perm->pool_ns_len = pool_ns_len;
1905         if (pool_ns_len > 0)
1906                 memcpy(perm->pool_ns, pool_ns->str, pool_ns_len);
1907         perm->pool_ns[pool_ns_len] = 0;
1908
1909         rb_link_node(&perm->node, parent, p);
1910         rb_insert_color(&perm->node, &mdsc->pool_perm_tree);
1911         err = 0;
1912 out_unlock:
1913         up_write(&mdsc->pool_perm_rwsem);
1914
1915         ceph_osdc_put_request(rd_req);
1916         ceph_osdc_put_request(wr_req);
1917 out:
1918         if (!err)
1919                 err = have;
1920         if (pool_ns)
1921                 dout("__ceph_pool_perm_get pool %lld ns %.*s result = %d\n",
1922                      pool, (int)pool_ns->len, pool_ns->str, err);
1923         else
1924                 dout("__ceph_pool_perm_get pool %lld result = %d\n", pool, err);
1925         return err;
1926 }
1927
1928 int ceph_pool_perm_check(struct inode *inode, int need)
1929 {
1930         struct ceph_inode_info *ci = ceph_inode(inode);
1931         struct ceph_string *pool_ns;
1932         s64 pool;
1933         int ret, flags;
1934
1935         /* Only need to do this for regular files */
1936         if (!S_ISREG(inode->i_mode))
1937                 return 0;
1938
1939         if (ci->i_vino.snap != CEPH_NOSNAP) {
1940                 /*
1941                  * Pool permission check needs to write to the first object.
1942                  * But for snapshot, head of the first object may have alread
1943                  * been deleted. Skip check to avoid creating orphan object.
1944                  */
1945                 return 0;
1946         }
1947
1948         if (ceph_test_mount_opt(ceph_inode_to_client(inode),
1949                                 NOPOOLPERM))
1950                 return 0;
1951
1952         spin_lock(&ci->i_ceph_lock);
1953         flags = ci->i_ceph_flags;
1954         pool = ci->i_layout.pool_id;
1955         spin_unlock(&ci->i_ceph_lock);
1956 check:
1957         if (flags & CEPH_I_POOL_PERM) {
1958                 if ((need & CEPH_CAP_FILE_RD) && !(flags & CEPH_I_POOL_RD)) {
1959                         dout("ceph_pool_perm_check pool %lld no read perm\n",
1960                              pool);
1961                         return -EPERM;
1962                 }
1963                 if ((need & CEPH_CAP_FILE_WR) && !(flags & CEPH_I_POOL_WR)) {
1964                         dout("ceph_pool_perm_check pool %lld no write perm\n",
1965                              pool);
1966                         return -EPERM;
1967                 }
1968                 return 0;
1969         }
1970
1971         pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
1972         ret = __ceph_pool_perm_get(ci, pool, pool_ns);
1973         ceph_put_string(pool_ns);
1974         if (ret < 0)
1975                 return ret;
1976
1977         flags = CEPH_I_POOL_PERM;
1978         if (ret & POOL_READ)
1979                 flags |= CEPH_I_POOL_RD;
1980         if (ret & POOL_WRITE)
1981                 flags |= CEPH_I_POOL_WR;
1982
1983         spin_lock(&ci->i_ceph_lock);
1984         if (pool == ci->i_layout.pool_id &&
1985             pool_ns == rcu_dereference_raw(ci->i_layout.pool_ns)) {
1986                 ci->i_ceph_flags |= flags;
1987         } else {
1988                 pool = ci->i_layout.pool_id;
1989                 flags = ci->i_ceph_flags;
1990         }
1991         spin_unlock(&ci->i_ceph_lock);
1992         goto check;
1993 }
1994
1995 void ceph_pool_perm_destroy(struct ceph_mds_client *mdsc)
1996 {
1997         struct ceph_pool_perm *perm;
1998         struct rb_node *n;
1999
2000         while (!RB_EMPTY_ROOT(&mdsc->pool_perm_tree)) {
2001                 n = rb_first(&mdsc->pool_perm_tree);
2002                 perm = rb_entry(n, struct ceph_pool_perm, node);
2003                 rb_erase(n, &mdsc->pool_perm_tree);
2004                 kfree(perm);
2005         }
2006 }