Blame view
fs/nilfs2/mdt.c
14.4 KB
5eb563f5f nilfs2: meta data... |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 |
/* * mdt.c - meta data file for NILFS * * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * * Written by Ryusuke Konishi <ryusuke@osrg.net> */ #include <linux/buffer_head.h> #include <linux/mpage.h> #include <linux/mm.h> #include <linux/writeback.h> #include <linux/backing-dev.h> #include <linux/swap.h> |
5a0e3ad6a include cleanup: ... |
29 |
#include <linux/slab.h> |
5eb563f5f nilfs2: meta data... |
30 |
#include "nilfs.h" |
05d0e94b6 nilfs2: get rid o... |
31 |
#include "btnode.h" |
5eb563f5f nilfs2: meta data... |
32 33 34 35 36 37 |
#include "segment.h" #include "page.h" #include "mdt.h" #define NILFS_MDT_MAX_RA_BLOCKS (16 - 1) |
5eb563f5f nilfs2: meta data... |
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 |
static int nilfs_mdt_insert_new_block(struct inode *inode, unsigned long block, struct buffer_head *bh, void (*init_block)(struct inode *, struct buffer_head *, void *)) { struct nilfs_inode_info *ii = NILFS_I(inode); void *kaddr; int ret; /* Caller exclude read accesses using page lock */ /* set_buffer_new(bh); */ bh->b_blocknr = 0; ret = nilfs_bmap_insert(ii->i_bmap, block, (unsigned long)bh); if (unlikely(ret)) return ret; set_buffer_mapped(bh); kaddr = kmap_atomic(bh->b_page, KM_USER0); memset(kaddr + bh_offset(bh), 0, 1 << inode->i_blkbits); if (init_block) init_block(inode, bh, kaddr); flush_dcache_page(bh->b_page); kunmap_atomic(kaddr, KM_USER0); set_buffer_uptodate(bh); |
5fc7b1417 nilfs2: use mark_... |
68 |
mark_buffer_dirty(bh); |
5eb563f5f nilfs2: meta data... |
69 70 71 72 73 74 75 76 77 78 |
nilfs_mdt_mark_dirty(inode); return 0; } static int nilfs_mdt_create_block(struct inode *inode, unsigned long block, struct buffer_head **out_bh, void (*init_block)(struct inode *, struct buffer_head *, void *)) { |
5eb563f5f nilfs2: meta data... |
79 80 81 82 |
struct super_block *sb = inode->i_sb; struct nilfs_transaction_info ti; struct buffer_head *bh; int err; |
5eb563f5f nilfs2: meta data... |
83 84 85 86 87 88 89 90 |
nilfs_transaction_begin(sb, &ti, 0); err = -ENOMEM; bh = nilfs_grab_buffer(inode, inode->i_mapping, block, 0); if (unlikely(!bh)) goto failed_unlock; err = -EEXIST; |
143511046 nilfs2: always lo... |
91 |
if (buffer_uptodate(bh)) |
5eb563f5f nilfs2: meta data... |
92 |
goto failed_bh; |
143511046 nilfs2: always lo... |
93 |
|
5eb563f5f nilfs2: meta data... |
94 95 96 |
wait_on_buffer(bh); if (buffer_uptodate(bh)) goto failed_bh; |
5eb563f5f nilfs2: meta data... |
97 |
|
090fd5b10 nilfs2: get rid o... |
98 |
bh->b_bdev = sb->s_bdev; |
5eb563f5f nilfs2: meta data... |
99 100 101 102 103 104 105 106 107 108 109 110 |
err = nilfs_mdt_insert_new_block(inode, block, bh, init_block); if (likely(!err)) { get_bh(bh); *out_bh = bh; } failed_bh: unlock_page(bh->b_page); page_cache_release(bh->b_page); brelse(bh); failed_unlock: |
47420c799 nilfs2: avoid dou... |
111 112 113 114 |
if (likely(!err)) err = nilfs_transaction_commit(sb); else nilfs_transaction_abort(sb); |
090fd5b10 nilfs2: get rid o... |
115 |
|
5eb563f5f nilfs2: meta data... |
116 117 118 119 120 121 122 123 |
return err; } static int nilfs_mdt_submit_block(struct inode *inode, unsigned long blkoff, int mode, struct buffer_head **out_bh) { struct buffer_head *bh; |
0f3fe33b3 nilfs2: convert n... |
124 |
__u64 blknum = 0; |
5eb563f5f nilfs2: meta data... |
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 |
int ret = -ENOMEM; bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0); if (unlikely(!bh)) goto failed; ret = -EEXIST; /* internal code */ if (buffer_uptodate(bh)) goto out; if (mode == READA) { if (!trylock_buffer(bh)) { ret = -EBUSY; goto failed_bh; } |
1f5abe7e7 nilfs2: replace B... |
140 |
} else /* mode == READ */ |
5eb563f5f nilfs2: meta data... |
141 |
lock_buffer(bh); |
5eb563f5f nilfs2: meta data... |
142 143 144 145 146 |
if (buffer_uptodate(bh)) { unlock_buffer(bh); goto out; } |
143511046 nilfs2: always lo... |
147 148 149 150 151 |
ret = nilfs_bmap_lookup(NILFS_I(inode)->i_bmap, blkoff, &blknum); if (unlikely(ret)) { unlock_buffer(bh); goto failed_bh; |
5eb563f5f nilfs2: meta data... |
152 |
} |
c6e071884 nilfs2: get rid o... |
153 |
map_bh(bh, inode->i_sb, (sector_t)blknum); |
5eb563f5f nilfs2: meta data... |
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 |
bh->b_end_io = end_buffer_read_sync; get_bh(bh); submit_bh(mode, bh); ret = 0; out: get_bh(bh); *out_bh = bh; failed_bh: unlock_page(bh->b_page); page_cache_release(bh->b_page); brelse(bh); failed: return ret; } static int nilfs_mdt_read_block(struct inode *inode, unsigned long block, |
b34a65069 nilfs2: avoid rea... |
172 |
int readahead, struct buffer_head **out_bh) |
5eb563f5f nilfs2: meta data... |
173 174 175 176 177 178 179 180 181 182 183 184 |
{ struct buffer_head *first_bh, *bh; unsigned long blkoff; int i, nr_ra_blocks = NILFS_MDT_MAX_RA_BLOCKS; int err; err = nilfs_mdt_submit_block(inode, block, READ, &first_bh); if (err == -EEXIST) /* internal code */ goto out; if (unlikely(err)) goto failed; |
b34a65069 nilfs2: avoid rea... |
185 186 187 188 189 190 191 192 193 194 195 196 |
if (readahead) { blkoff = block + 1; for (i = 0; i < nr_ra_blocks; i++, blkoff++) { err = nilfs_mdt_submit_block(inode, blkoff, READA, &bh); if (likely(!err || err == -EEXIST)) brelse(bh); else if (err != -EBUSY) break; /* abort readahead if bmap lookup failed */ if (!buffer_locked(first_bh)) goto out_no_wait; } |
5eb563f5f nilfs2: meta data... |
197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 |
} wait_on_buffer(first_bh); out_no_wait: err = -EIO; if (!buffer_uptodate(first_bh)) goto failed_bh; out: *out_bh = first_bh; return 0; failed_bh: brelse(first_bh); failed: return err; } /** * nilfs_mdt_get_block - read or create a buffer on meta data file. * @inode: inode of the meta data file * @blkoff: block offset * @create: create flag * @init_block: initializer used for newly allocated block * @out_bh: output of a pointer to the buffer_head * * nilfs_mdt_get_block() looks up the specified buffer and tries to create * a new buffer if @create is not zero. On success, the returned buffer is * assured to be either existing or formatted using a buffer lock on success. * @out_bh is substituted only when zero is returned. * * Return Value: On success, it returns 0. On error, the following negative * error code is returned. * * %-ENOMEM - Insufficient memory available. * * %-EIO - I/O error * * %-ENOENT - the specified block does not exist (hole block) * |
5eb563f5f nilfs2: meta data... |
237 238 239 240 241 242 243 244 245 246 247 |
* %-EROFS - Read only filesystem (for create mode) */ int nilfs_mdt_get_block(struct inode *inode, unsigned long blkoff, int create, void (*init_block)(struct inode *, struct buffer_head *, void *), struct buffer_head **out_bh) { int ret; /* Should be rewritten with merging nilfs_mdt_read_block() */ retry: |
b34a65069 nilfs2: avoid rea... |
248 |
ret = nilfs_mdt_read_block(inode, blkoff, !create, out_bh); |
5eb563f5f nilfs2: meta data... |
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 |
if (!create || ret != -ENOENT) return ret; ret = nilfs_mdt_create_block(inode, blkoff, out_bh, init_block); if (unlikely(ret == -EEXIST)) { /* create = 0; */ /* limit read-create loop retries */ goto retry; } return ret; } /** * nilfs_mdt_delete_block - make a hole on the meta data file. * @inode: inode of the meta data file * @block: block offset * * Return Value: On success, zero is returned. * On error, one of the following negative error code is returned. * * %-ENOMEM - Insufficient memory available. * * %-EIO - I/O error |
5eb563f5f nilfs2: meta data... |
271 272 273 274 275 276 277 |
*/ int nilfs_mdt_delete_block(struct inode *inode, unsigned long block) { struct nilfs_inode_info *ii = NILFS_I(inode); int err; err = nilfs_bmap_delete(ii->i_bmap, block); |
843382370 nilfs2: ensure to... |
278 |
if (!err || err == -ENOENT) { |
5eb563f5f nilfs2: meta data... |
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 |
nilfs_mdt_mark_dirty(inode); nilfs_mdt_forget_block(inode, block); } return err; } /** * nilfs_mdt_forget_block - discard dirty state and try to remove the page * @inode: inode of the meta data file * @block: block offset * * nilfs_mdt_forget_block() clears a dirty flag of the specified buffer, and * tries to release the page including the buffer from a page cache. * * Return Value: On success, 0 is returned. On error, one of the following * negative error code is returned. * * %-EBUSY - page has an active buffer. * * %-ENOENT - page cache has no page addressed by the offset. */ int nilfs_mdt_forget_block(struct inode *inode, unsigned long block) { pgoff_t index = (pgoff_t)block >> (PAGE_CACHE_SHIFT - inode->i_blkbits); struct page *page; unsigned long first_block; int ret = 0; int still_dirty; page = find_lock_page(inode->i_mapping, index); if (!page) return -ENOENT; wait_on_page_writeback(page); first_block = (unsigned long)index << (PAGE_CACHE_SHIFT - inode->i_blkbits); if (page_has_buffers(page)) { struct buffer_head *bh; bh = nilfs_page_get_nth_block(page, block - first_block); nilfs_forget_buffer(bh); } still_dirty = PageDirty(page); unlock_page(page); page_cache_release(page); if (still_dirty || invalidate_inode_pages2_range(inode->i_mapping, index, index) != 0) ret = -EBUSY; return ret; } /** * nilfs_mdt_mark_block_dirty - mark a block on the meta data file dirty. * @inode: inode of the meta data file * @block: block offset * * Return Value: On success, it returns 0. On error, the following negative * error code is returned. * * %-ENOMEM - Insufficient memory available. * * %-EIO - I/O error * * %-ENOENT - the specified block does not exist (hole block) |
5eb563f5f nilfs2: meta data... |
346 347 348 349 350 |
*/ int nilfs_mdt_mark_block_dirty(struct inode *inode, unsigned long block) { struct buffer_head *bh; int err; |
b34a65069 nilfs2: avoid rea... |
351 |
err = nilfs_mdt_read_block(inode, block, 0, &bh); |
5eb563f5f nilfs2: meta data... |
352 353 |
if (unlikely(err)) return err; |
5fc7b1417 nilfs2: use mark_... |
354 |
mark_buffer_dirty(bh); |
5eb563f5f nilfs2: meta data... |
355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 |
nilfs_mdt_mark_dirty(inode); brelse(bh); return 0; } int nilfs_mdt_fetch_dirty(struct inode *inode) { struct nilfs_inode_info *ii = NILFS_I(inode); if (nilfs_bmap_test_and_clear_dirty(ii->i_bmap)) { set_bit(NILFS_I_DIRTY, &ii->i_state); return 1; } return test_bit(NILFS_I_DIRTY, &ii->i_state); } static int nilfs_mdt_write_page(struct page *page, struct writeback_control *wbc) { |
ebdfed4dc nilfs2: add routi... |
374 375 |
struct inode *inode; struct super_block *sb; |
5eb563f5f nilfs2: meta data... |
376 377 378 379 |
int err = 0; redirty_page_for_writepage(wbc, page); unlock_page(page); |
ebdfed4dc nilfs2: add routi... |
380 381 382 383 384 |
inode = page->mapping->host; if (!inode) return 0; sb = inode->i_sb; |
5eb563f5f nilfs2: meta data... |
385 386 387 388 389 |
if (wbc->sync_mode == WB_SYNC_ALL) err = nilfs_construct_segment(sb); else if (wbc->for_reclaim) nilfs_flush_segment(sb, inode->i_ino); |
5eb563f5f nilfs2: meta data... |
390 391 |
return err; } |
7f09410bb const: mark remai... |
392 |
static const struct address_space_operations def_mdt_aops = { |
5eb563f5f nilfs2: meta data... |
393 394 |
.writepage = nilfs_mdt_write_page, }; |
6e1d5dcc2 const: mark remai... |
395 |
static const struct inode_operations def_mdt_iops; |
828c09509 const: constify r... |
396 |
static const struct file_operations def_mdt_fops; |
5eb563f5f nilfs2: meta data... |
397 |
|
5e19a995f nilfs2: separate ... |
398 |
|
f1e89c86f nilfs2: use iget ... |
399 |
int nilfs_mdt_init(struct inode *inode, gfp_t gfp_mask, size_t objsz) |
5e19a995f nilfs2: separate ... |
400 401 402 403 404 405 |
{ struct nilfs_mdt_info *mi; mi = kzalloc(max(sizeof(*mi), objsz), GFP_NOFS); if (!mi) return -ENOMEM; |
5e19a995f nilfs2: separate ... |
406 407 408 409 410 |
init_rwsem(&mi->mi_sem); inode->i_private = mi; inode->i_mode = S_IFREG; mapping_set_gfp_mask(inode->i_mapping, gfp_mask); |
f1e89c86f nilfs2: use iget ... |
411 412 413 414 415 |
inode->i_mapping->backing_dev_info = inode->i_sb->s_bdi; inode->i_op = &def_mdt_iops; inode->i_fop = &def_mdt_fops; inode->i_mapping->a_ops = &def_mdt_aops; |
5e19a995f nilfs2: separate ... |
416 417 418 |
return 0; } |
5eb563f5f nilfs2: meta data... |
419 420 421 422 423 424 425 426 427 |
void nilfs_mdt_set_entry_size(struct inode *inode, unsigned entry_size, unsigned header_size) { struct nilfs_mdt_info *mi = NILFS_MDT(inode); mi->mi_entry_size = entry_size; mi->mi_entries_per_block = (1 << inode->i_blkbits) / entry_size; mi->mi_first_entry_offset = DIV_ROUND_UP(header_size, entry_size); } |
ebdfed4dc nilfs2: add routi... |
428 429 430 431 432 433 434 435 436 |
/** * nilfs_mdt_setup_shadow_map - setup shadow map and bind it to metadata file * @inode: inode of the metadata file * @shadow: shadow mapping */ int nilfs_mdt_setup_shadow_map(struct inode *inode, struct nilfs_shadow_map *shadow) { struct nilfs_mdt_info *mi = NILFS_MDT(inode); |
026a7d63d nilfs2: get rid o... |
437 |
struct backing_dev_info *bdi = inode->i_sb->s_bdi; |
ebdfed4dc nilfs2: add routi... |
438 439 |
INIT_LIST_HEAD(&shadow->frozen_buffers); |
2aa15890f mm: prevent concu... |
440 |
address_space_init_once(&shadow->frozen_data); |
aa405b1f4 nilfs2: always se... |
441 |
nilfs_mapping_init(&shadow->frozen_data, inode, bdi); |
2aa15890f mm: prevent concu... |
442 |
address_space_init_once(&shadow->frozen_btnodes); |
aa405b1f4 nilfs2: always se... |
443 |
nilfs_mapping_init(&shadow->frozen_btnodes, inode, bdi); |
ebdfed4dc nilfs2: add routi... |
444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 |
mi->mi_shadow = shadow; return 0; } /** * nilfs_mdt_save_to_shadow_map - copy bmap and dirty pages to shadow map * @inode: inode of the metadata file */ int nilfs_mdt_save_to_shadow_map(struct inode *inode) { struct nilfs_mdt_info *mi = NILFS_MDT(inode); struct nilfs_inode_info *ii = NILFS_I(inode); struct nilfs_shadow_map *shadow = mi->mi_shadow; int ret; ret = nilfs_copy_dirty_pages(&shadow->frozen_data, inode->i_mapping); if (ret) goto out; ret = nilfs_copy_dirty_pages(&shadow->frozen_btnodes, &ii->i_btnode_cache); if (ret) goto out; nilfs_bmap_save(ii->i_bmap, &shadow->bmap_store); out: return ret; } |
b1f6a4f29 nilfs2: add routi... |
472 473 474 475 476 477 |
int nilfs_mdt_freeze_buffer(struct inode *inode, struct buffer_head *bh) { struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow; struct buffer_head *bh_frozen; struct page *page; int blkbits = inode->i_blkbits; |
b1f6a4f29 nilfs2: add routi... |
478 479 480 |
page = grab_cache_page(&shadow->frozen_data, bh->b_page->index); if (!page) |
a7a8447ed nilfs2: simplify ... |
481 |
return -ENOMEM; |
b1f6a4f29 nilfs2: add routi... |
482 483 484 485 486 |
if (!page_has_buffers(page)) create_empty_buffers(page, 1 << blkbits, 0); bh_frozen = nilfs_page_get_nth_block(page, bh_offset(bh) >> blkbits); |
a7a8447ed nilfs2: simplify ... |
487 488 489 490 491 492 493 494 495 |
if (!buffer_uptodate(bh_frozen)) nilfs_copy_buffer(bh_frozen, bh); if (list_empty(&bh_frozen->b_assoc_buffers)) { list_add_tail(&bh_frozen->b_assoc_buffers, &shadow->frozen_buffers); set_buffer_nilfs_redirected(bh); } else { brelse(bh_frozen); /* already frozen */ |
b1f6a4f29 nilfs2: add routi... |
496 |
} |
a7a8447ed nilfs2: simplify ... |
497 |
|
b1f6a4f29 nilfs2: add routi... |
498 499 |
unlock_page(page); page_cache_release(page); |
a7a8447ed nilfs2: simplify ... |
500 |
return 0; |
b1f6a4f29 nilfs2: add routi... |
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 |
} struct buffer_head * nilfs_mdt_get_frozen_buffer(struct inode *inode, struct buffer_head *bh) { struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow; struct buffer_head *bh_frozen = NULL; struct page *page; int n; page = find_lock_page(&shadow->frozen_data, bh->b_page->index); if (page) { if (page_has_buffers(page)) { n = bh_offset(bh) >> inode->i_blkbits; bh_frozen = nilfs_page_get_nth_block(page, n); } unlock_page(page); page_cache_release(page); } return bh_frozen; } static void nilfs_release_frozen_buffers(struct nilfs_shadow_map *shadow) { struct list_head *head = &shadow->frozen_buffers; struct buffer_head *bh; while (!list_empty(head)) { bh = list_first_entry(head, struct buffer_head, b_assoc_buffers); list_del_init(&bh->b_assoc_buffers); brelse(bh); /* drop ref-count to make it releasable */ } } |
ebdfed4dc nilfs2: add routi... |
535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 |
/** * nilfs_mdt_restore_from_shadow_map - restore dirty pages and bmap state * @inode: inode of the metadata file */ void nilfs_mdt_restore_from_shadow_map(struct inode *inode) { struct nilfs_mdt_info *mi = NILFS_MDT(inode); struct nilfs_inode_info *ii = NILFS_I(inode); struct nilfs_shadow_map *shadow = mi->mi_shadow; down_write(&mi->mi_sem); if (mi->mi_palloc_cache) nilfs_palloc_clear_cache(inode); nilfs_clear_dirty_pages(inode->i_mapping); nilfs_copy_back_pages(inode->i_mapping, &shadow->frozen_data); nilfs_clear_dirty_pages(&ii->i_btnode_cache); nilfs_copy_back_pages(&ii->i_btnode_cache, &shadow->frozen_btnodes); nilfs_bmap_restore(ii->i_bmap, &shadow->bmap_store); up_write(&mi->mi_sem); } /** * nilfs_mdt_clear_shadow_map - truncate pages in shadow map caches * @inode: inode of the metadata file */ void nilfs_mdt_clear_shadow_map(struct inode *inode) { struct nilfs_mdt_info *mi = NILFS_MDT(inode); struct nilfs_shadow_map *shadow = mi->mi_shadow; down_write(&mi->mi_sem); |
b1f6a4f29 nilfs2: add routi... |
571 |
nilfs_release_frozen_buffers(shadow); |
ebdfed4dc nilfs2: add routi... |
572 573 574 575 |
truncate_inode_pages(&shadow->frozen_data, 0); truncate_inode_pages(&shadow->frozen_btnodes, 0); up_write(&mi->mi_sem); } |