Blame view
fs/ocfs2/aops.c
40.2 KB
ccd979bdb
|
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 |
/* -*- mode: c; c-basic-offset: 8; -*- * vim: noexpandtab sw=8 ts=8 sts=0: * * Copyright (C) 2002, 2004 Oracle. All rights reserved. * * 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., 59 Temple Place - Suite 330, * Boston, MA 021110-1307, USA. */ #include <linux/fs.h> #include <linux/slab.h> #include <linux/highmem.h> #include <linux/pagemap.h> #include <asm/byteorder.h> |
9517bac6c
|
27 |
#include <linux/swap.h> |
6af67d820
|
28 |
#include <linux/pipe_fs_i.h> |
ccd979bdb
|
29 30 31 32 33 34 35 36 37 38 39 40 41 |
#define MLOG_MASK_PREFIX ML_FILE_IO #include <cluster/masklog.h> #include "ocfs2.h" #include "alloc.h" #include "aops.h" #include "dlmglue.h" #include "extent_map.h" #include "file.h" #include "inode.h" #include "journal.h" |
9517bac6c
|
42 |
#include "suballoc.h" |
ccd979bdb
|
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 68 69 70 71 72 73 74 75 76 77 78 79 80 |
#include "super.h" #include "symlink.h" #include "buffer_head_io.h" static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create) { int err = -EIO; int status; struct ocfs2_dinode *fe = NULL; struct buffer_head *bh = NULL; struct buffer_head *buffer_cache_bh = NULL; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); void *kaddr; mlog_entry("(0x%p, %llu, 0x%p, %d) ", inode, (unsigned long long)iblock, bh_result, create); BUG_ON(ocfs2_inode_is_fast_symlink(inode)); if ((iblock << inode->i_sb->s_blocksize_bits) > PATH_MAX + 1) { mlog(ML_ERROR, "block offset > PATH_MAX: %llu", (unsigned long long)iblock); goto bail; } status = ocfs2_read_block(OCFS2_SB(inode->i_sb), OCFS2_I(inode)->ip_blkno, &bh, OCFS2_BH_CACHED, inode); if (status < 0) { mlog_errno(status); goto bail; } fe = (struct ocfs2_dinode *) bh->b_data; if (!OCFS2_IS_VALID_DINODE(fe)) { |
b0697053f
|
81 82 |
mlog(ML_ERROR, "Invalid dinode #%llu: signature = %.*s ", |
1ca1a111b
|
83 84 |
(unsigned long long)le64_to_cpu(fe->i_blkno), 7, fe->i_signature); |
ccd979bdb
|
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
goto bail; } if ((u64)iblock >= ocfs2_clusters_to_blocks(inode->i_sb, le32_to_cpu(fe->i_clusters))) { mlog(ML_ERROR, "block offset is outside the allocated size: " "%llu ", (unsigned long long)iblock); goto bail; } /* We don't use the page cache to create symlink data, so if * need be, copy it over from the buffer cache. */ if (!buffer_uptodate(bh_result) && ocfs2_inode_is_new(inode)) { u64 blkno = le64_to_cpu(fe->id2.i_list.l_recs[0].e_blkno) + iblock; buffer_cache_bh = sb_getblk(osb->sb, blkno); if (!buffer_cache_bh) { mlog(ML_ERROR, "couldn't getblock for symlink! "); goto bail; } /* we haven't locked out transactions, so a commit * could've happened. Since we've got a reference on * the bh, even if it commits while we're doing the * copy, the data is still good. */ if (buffer_jbd(buffer_cache_bh) && ocfs2_inode_is_new(inode)) { kaddr = kmap_atomic(bh_result->b_page, KM_USER0); if (!kaddr) { mlog(ML_ERROR, "couldn't kmap! "); goto bail; } memcpy(kaddr + (bh_result->b_size * iblock), buffer_cache_bh->b_data, bh_result->b_size); kunmap_atomic(kaddr, KM_USER0); set_buffer_uptodate(bh_result); } brelse(buffer_cache_bh); } map_bh(bh_result, inode->i_sb, le64_to_cpu(fe->id2.i_list.l_recs[0].e_blkno) + iblock); err = 0; bail: if (bh) brelse(bh); mlog_exit(err); return err; } static int ocfs2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create) { int err = 0; |
49cb8d2d4
|
146 |
unsigned int ext_flags; |
ccd979bdb
|
147 |
u64 p_blkno, past_eof; |
25baf2da1
|
148 |
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); |
ccd979bdb
|
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 |
mlog_entry("(0x%p, %llu, 0x%p, %d) ", inode, (unsigned long long)iblock, bh_result, create); if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SYSTEM_FILE) mlog(ML_NOTICE, "get_block on system inode 0x%p (%lu) ", inode, inode->i_ino); if (S_ISLNK(inode->i_mode)) { /* this always does I/O for some reason. */ err = ocfs2_symlink_get_block(inode, iblock, bh_result, create); goto bail; } |
49cb8d2d4
|
164 165 |
err = ocfs2_extent_map_get_blocks(inode, iblock, &p_blkno, NULL, &ext_flags); |
ccd979bdb
|
166 167 |
if (err) { mlog(ML_ERROR, "Error %d from get_blocks(0x%p, %llu, 1, " |
b0697053f
|
168 169 170 |
"%llu, NULL) ", err, inode, (unsigned long long)iblock, (unsigned long long)p_blkno); |
ccd979bdb
|
171 172 |
goto bail; } |
25baf2da1
|
173 174 175 176 177 178 179 180 181 182 |
/* * ocfs2 never allocates in this function - the only time we * need to use BH_New is when we're extending i_size on a file * system which doesn't support holes, in which case BH_New * allows block_prepare_write() to zero. */ mlog_bug_on_msg(create && p_blkno == 0 && ocfs2_sparse_alloc(osb), "ino %lu, iblock %llu ", inode->i_ino, (unsigned long long)iblock); |
49cb8d2d4
|
183 184 |
/* Treat the unwritten extent as a hole for zeroing purposes. */ if (p_blkno && !(ext_flags & OCFS2_EXT_UNWRITTEN)) |
25baf2da1
|
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 |
map_bh(bh_result, inode->i_sb, p_blkno); if (!ocfs2_sparse_alloc(osb)) { if (p_blkno == 0) { err = -EIO; mlog(ML_ERROR, "iblock = %llu p_blkno = %llu blkno=(%llu) ", (unsigned long long)iblock, (unsigned long long)p_blkno, (unsigned long long)OCFS2_I(inode)->ip_blkno); mlog(ML_ERROR, "Size %llu, clusters %u ", (unsigned long long)i_size_read(inode), OCFS2_I(inode)->ip_clusters); dump_stack(); } |
ccd979bdb
|
200 |
|
25baf2da1
|
201 202 203 204 |
past_eof = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode)); mlog(0, "Inode %lu, past_eof = %llu ", inode->i_ino, (unsigned long long)past_eof); |
ccd979bdb
|
205 |
|
25baf2da1
|
206 207 208 |
if (create && (iblock >= past_eof)) set_buffer_new(bh_result); } |
ccd979bdb
|
209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 |
bail: if (err < 0) err = -EIO; mlog_exit(err); return err; } static int ocfs2_readpage(struct file *file, struct page *page) { struct inode *inode = page->mapping->host; loff_t start = (loff_t)page->index << PAGE_CACHE_SHIFT; int ret, unlock = 1; mlog_entry("(0x%p, %lu) ", file, (page ? page->index : 0)); |
4bcec1847
|
226 |
ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page); |
ccd979bdb
|
227 228 229 230 231 232 |
if (ret != 0) { if (ret == AOP_TRUNCATED_PAGE) unlock = 0; mlog_errno(ret); goto out; } |
e9dfc0b2b
|
233 234 235 236 |
if (down_read_trylock(&OCFS2_I(inode)->ip_alloc_sem) == 0) { ret = AOP_TRUNCATED_PAGE; goto out_meta_unlock; } |
ccd979bdb
|
237 238 239 240 241 242 |
/* * i_size might have just been updated as we grabed the meta lock. We * might now be discovering a truncate that hit on another node. * block_read_full_page->get_block freaks out if it is asked to read * beyond the end of a file, so we check here. Callers |
54cb8821d
|
243 |
* (generic_file_read, vm_ops->fault) are clever enough to check i_size |
ccd979bdb
|
244 245 246 247 248 |
* and notice that the page they just read isn't needed. * * XXX sys_readahead() seems to get that wrong? */ if (start >= i_size_read(inode)) { |
5c3c6bb77
|
249 |
zero_user_page(page, 0, PAGE_SIZE, KM_USER0); |
ccd979bdb
|
250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 |
SetPageUptodate(page); ret = 0; goto out_alloc; } ret = ocfs2_data_lock_with_page(inode, 0, page); if (ret != 0) { if (ret == AOP_TRUNCATED_PAGE) unlock = 0; mlog_errno(ret); goto out_alloc; } ret = block_read_full_page(page, ocfs2_get_block); unlock = 0; ocfs2_data_unlock(inode, 0); out_alloc: up_read(&OCFS2_I(inode)->ip_alloc_sem); |
e9dfc0b2b
|
269 |
out_meta_unlock: |
ccd979bdb
|
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 |
ocfs2_meta_unlock(inode, 0); out: if (unlock) unlock_page(page); mlog_exit(ret); return ret; } /* Note: Because we don't support holes, our allocation has * already happened (allocation writes zeros to the file data) * so we don't have to worry about ordered writes in * ocfs2_writepage. * * ->writepage is called during the process of invalidating the page cache * during blocked lock processing. It can't block on any cluster locks * to during block mapping. It's relying on the fact that the block * mapping can't have disappeared under the dirty pages that it is * being asked to write back. */ static int ocfs2_writepage(struct page *page, struct writeback_control *wbc) { int ret; mlog_entry("(0x%p) ", page); ret = block_write_full_page(page, ocfs2_get_block, wbc); mlog_exit(ret); return ret; } |
5069120b7
|
302 303 304 305 306 |
/* * This is called from ocfs2_write_zero_page() which has handled it's * own cluster locking and has ensured allocation exists for those * blocks to be written. */ |
53013cba4
|
307 308 309 310 311 312 313 314 315 316 317 318 319 |
int ocfs2_prepare_write_nolock(struct inode *inode, struct page *page, unsigned from, unsigned to) { int ret; down_read(&OCFS2_I(inode)->ip_alloc_sem); ret = block_prepare_write(page, from, to, ocfs2_get_block); up_read(&OCFS2_I(inode)->ip_alloc_sem); return ret; } |
ccd979bdb
|
320 321 322 323 |
/* Taken from ext3. We don't necessarily need the full blown * functionality yet, but IMHO it's better to cut and paste the whole * thing so we can avoid introducing our own bugs (and easily pick up * their fixes when they happen) --Mark */ |
60b11392f
|
324 325 326 327 328 329 330 |
int walk_page_buffers( handle_t *handle, struct buffer_head *head, unsigned from, unsigned to, int *partial, int (*fn)( handle_t *handle, struct buffer_head *bh)) |
ccd979bdb
|
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 |
{ struct buffer_head *bh; unsigned block_start, block_end; unsigned blocksize = head->b_size; int err, ret = 0; struct buffer_head *next; for ( bh = head, block_start = 0; ret == 0 && (bh != head || !block_start); block_start = block_end, bh = next) { next = bh->b_this_page; block_end = block_start + blocksize; if (block_end <= from || block_start >= to) { if (partial && !buffer_uptodate(bh)) *partial = 1; continue; } err = (*fn)(handle, bh); if (!ret) ret = err; } return ret; } |
1fabe1481
|
355 |
handle_t *ocfs2_start_walk_page_trans(struct inode *inode, |
ccd979bdb
|
356 357 358 359 360 |
struct page *page, unsigned from, unsigned to) { struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); |
1fabe1481
|
361 |
handle_t *handle = NULL; |
ccd979bdb
|
362 |
int ret = 0; |
65eff9ccf
|
363 |
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); |
ccd979bdb
|
364 365 366 367 368 369 370 |
if (!handle) { ret = -ENOMEM; mlog_errno(ret); goto out; } if (ocfs2_should_order_data(inode)) { |
1fabe1481
|
371 |
ret = walk_page_buffers(handle, |
ccd979bdb
|
372 373 374 375 376 377 378 379 380 |
page_buffers(page), from, to, NULL, ocfs2_journal_dirty_data); if (ret < 0) mlog_errno(ret); } out: if (ret) { if (handle) |
02dc1af44
|
381 |
ocfs2_commit_trans(osb, handle); |
ccd979bdb
|
382 383 384 385 |
handle = ERR_PTR(ret); } return handle; } |
ccd979bdb
|
386 387 388 389 390 391 392 393 394 395 396 397 398 399 |
static sector_t ocfs2_bmap(struct address_space *mapping, sector_t block) { sector_t status; u64 p_blkno = 0; int err = 0; struct inode *inode = mapping->host; mlog_entry("(block = %llu) ", (unsigned long long)block); /* We don't need to lock journal system files, since they aren't * accessed concurrently from multiple nodes. */ if (!INODE_JOURNAL(inode)) { |
4bcec1847
|
400 |
err = ocfs2_meta_lock(inode, NULL, 0); |
ccd979bdb
|
401 402 403 404 405 406 407 |
if (err) { if (err != -ENOENT) mlog_errno(err); goto bail; } down_read(&OCFS2_I(inode)->ip_alloc_sem); } |
49cb8d2d4
|
408 |
err = ocfs2_extent_map_get_blocks(inode, block, &p_blkno, NULL, NULL); |
ccd979bdb
|
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 |
if (!INODE_JOURNAL(inode)) { up_read(&OCFS2_I(inode)->ip_alloc_sem); ocfs2_meta_unlock(inode, 0); } if (err) { mlog(ML_ERROR, "get_blocks() failed, block = %llu ", (unsigned long long)block); mlog_errno(err); goto bail; } bail: status = err ? 0 : p_blkno; mlog_exit((int)status); return status; } /* * TODO: Make this into a generic get_blocks function. * * From do_direct_io in direct-io.c: * "So what we do is to permit the ->get_blocks function to populate * bh.b_size with the size of IO which is permitted at this offset and * this i_blkbits." * * This function is called directly from get_more_blocks in direct-io.c. * * called like this: dio->get_blocks(dio->inode, fs_startblk, * fs_count, map_bh, dio->rw == WRITE); */ static int ocfs2_direct_IO_get_blocks(struct inode *inode, sector_t iblock, |
ccd979bdb
|
446 447 448 |
struct buffer_head *bh_result, int create) { int ret; |
4f902c377
|
449 |
u64 p_blkno, inode_blocks, contig_blocks; |
49cb8d2d4
|
450 |
unsigned int ext_flags; |
184d7d20d
|
451 |
unsigned char blocksize_bits = inode->i_sb->s_blocksize_bits; |
1d8fa7a2b
|
452 |
unsigned long max_blocks = bh_result->b_size >> inode->i_blkbits; |
ccd979bdb
|
453 |
|
ccd979bdb
|
454 455 456 |
/* This function won't even be called if the request isn't all * nicely aligned and of the right size, so there's no need * for us to check any of that. */ |
25baf2da1
|
457 |
inode_blocks = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode)); |
564f8a322
|
458 459 460 461 462 |
/* * Any write past EOF is not allowed because we'd be extending. */ if (create && (iblock + max_blocks) > inode_blocks) { |
ccd979bdb
|
463 464 465 |
ret = -EIO; goto bail; } |
ccd979bdb
|
466 467 468 |
/* This figures out the size of the next contiguous block, and * our logical offset */ |
363041a5f
|
469 |
ret = ocfs2_extent_map_get_blocks(inode, iblock, &p_blkno, |
49cb8d2d4
|
470 |
&contig_blocks, &ext_flags); |
ccd979bdb
|
471 472 473 474 475 476 477 |
if (ret) { mlog(ML_ERROR, "get_blocks() failed iblock=%llu ", (unsigned long long)iblock); ret = -EIO; goto bail; } |
25baf2da1
|
478 479 480 481 482 483 484 485 486 487 488 489 490 |
if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)) && !p_blkno) { ocfs2_error(inode->i_sb, "Inode %llu has a hole at block %llu ", (unsigned long long)OCFS2_I(inode)->ip_blkno, (unsigned long long)iblock); ret = -EROFS; goto bail; } /* * get_more_blocks() expects us to describe a hole by clearing * the mapped bit on bh_result(). |
49cb8d2d4
|
491 492 |
* * Consider an unwritten extent as a hole. |
25baf2da1
|
493 |
*/ |
49cb8d2d4
|
494 |
if (p_blkno && !(ext_flags & OCFS2_EXT_UNWRITTEN)) |
25baf2da1
|
495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 |
map_bh(bh_result, inode->i_sb, p_blkno); else { /* * ocfs2_prepare_inode_for_write() should have caught * the case where we'd be filling a hole and triggered * a buffered write instead. */ if (create) { ret = -EIO; mlog_errno(ret); goto bail; } clear_buffer_mapped(bh_result); } |
ccd979bdb
|
510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 |
/* make sure we don't map more than max_blocks blocks here as that's all the kernel will handle at this point. */ if (max_blocks < contig_blocks) contig_blocks = max_blocks; bh_result->b_size = contig_blocks << blocksize_bits; bail: return ret; } /* * ocfs2_dio_end_io is called by the dio core when a dio is finished. We're * particularly interested in the aio/dio case. Like the core uses * i_alloc_sem, we use the rw_lock DLM lock to protect io on one node from * truncation on another. */ static void ocfs2_dio_end_io(struct kiocb *iocb, loff_t offset, ssize_t bytes, void *private) { |
d28c91740
|
531 |
struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode; |
7cdfc3a1c
|
532 |
int level; |
ccd979bdb
|
533 534 535 |
/* this io's submitter should not have unlocked this before we could */ BUG_ON(!ocfs2_iocb_is_rw_locked(iocb)); |
7cdfc3a1c
|
536 |
|
ccd979bdb
|
537 |
ocfs2_iocb_clear_rw_locked(iocb); |
7cdfc3a1c
|
538 539 540 541 542 |
level = ocfs2_iocb_rw_locked_level(iocb); if (!level) up_read(&inode->i_alloc_sem); ocfs2_rw_unlock(inode, level); |
ccd979bdb
|
543 |
} |
03f981cf2
|
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 |
/* * ocfs2_invalidatepage() and ocfs2_releasepage() are shamelessly stolen * from ext3. PageChecked() bits have been removed as OCFS2 does not * do journalled data. */ static void ocfs2_invalidatepage(struct page *page, unsigned long offset) { journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal; journal_invalidatepage(journal, page, offset); } static int ocfs2_releasepage(struct page *page, gfp_t wait) { journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal; if (!page_has_buffers(page)) return 0; return journal_try_to_free_buffers(journal, page, wait); } |
ccd979bdb
|
564 565 566 567 568 569 570 |
static ssize_t ocfs2_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t offset, unsigned long nr_segs) { struct file *file = iocb->ki_filp; |
d28c91740
|
571 |
struct inode *inode = file->f_path.dentry->d_inode->i_mapping->host; |
ccd979bdb
|
572 573 574 |
int ret; mlog_entry_void(); |
53013cba4
|
575 |
|
9517bac6c
|
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 |
if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb))) { /* * We get PR data locks even for O_DIRECT. This * allows concurrent O_DIRECT I/O but doesn't let * O_DIRECT with extending and buffered zeroing writes * race. If they did race then the buffered zeroing * could be written back after the O_DIRECT I/O. It's * one thing to tell people not to mix buffered and * O_DIRECT writes, but expecting them to understand * that file extension is also an implicit buffered * write is too much. By getting the PR we force * writeback of the buffered zeroing before * proceeding. */ ret = ocfs2_data_lock(inode, 0); if (ret < 0) { mlog_errno(ret); goto out; } ocfs2_data_unlock(inode, 0); |
53013cba4
|
596 |
} |
53013cba4
|
597 |
|
ccd979bdb
|
598 599 600 601 602 |
ret = blockdev_direct_IO_no_locking(rw, iocb, inode, inode->i_sb->s_bdev, iov, offset, nr_segs, ocfs2_direct_IO_get_blocks, ocfs2_dio_end_io); |
53013cba4
|
603 |
out: |
ccd979bdb
|
604 605 606 |
mlog_exit(ret); return ret; } |
9517bac6c
|
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 |
static void ocfs2_figure_cluster_boundaries(struct ocfs2_super *osb, u32 cpos, unsigned int *start, unsigned int *end) { unsigned int cluster_start = 0, cluster_end = PAGE_CACHE_SIZE; if (unlikely(PAGE_CACHE_SHIFT > osb->s_clustersize_bits)) { unsigned int cpp; cpp = 1 << (PAGE_CACHE_SHIFT - osb->s_clustersize_bits); cluster_start = cpos % cpp; cluster_start = cluster_start << osb->s_clustersize_bits; cluster_end = cluster_start + osb->s_clustersize; } BUG_ON(cluster_start > PAGE_SIZE); BUG_ON(cluster_end > PAGE_SIZE); if (start) *start = cluster_start; if (end) *end = cluster_end; } /* * 'from' and 'to' are the region in the page to avoid zeroing. * * If pagesize > clustersize, this function will avoid zeroing outside * of the cluster boundary. * * from == to == 0 is code for "zero the entire cluster region" */ static void ocfs2_clear_page_regions(struct page *page, struct ocfs2_super *osb, u32 cpos, unsigned from, unsigned to) { void *kaddr; unsigned int cluster_start, cluster_end; ocfs2_figure_cluster_boundaries(osb, cpos, &cluster_start, &cluster_end); kaddr = kmap_atomic(page, KM_USER0); if (from || to) { if (from > cluster_start) memset(kaddr + cluster_start, 0, from - cluster_start); if (to < cluster_end) memset(kaddr + to, 0, cluster_end - to); } else { memset(kaddr + cluster_start, 0, cluster_end - cluster_start); } kunmap_atomic(kaddr, KM_USER0); } /* * Some of this taken from block_prepare_write(). We already have our * mapping by now though, and the entire write will be allocating or * it won't, so not much need to use BH_New. * * This will also skip zeroing, which is handled externally. */ |
60b11392f
|
672 673 674 |
int ocfs2_map_page_blocks(struct page *page, u64 *p_blkno, struct inode *inode, unsigned int from, unsigned int to, int new) |
9517bac6c
|
675 676 677 678 679 680 681 682 683 684 685 686 687 |
{ int ret = 0; struct buffer_head *head, *bh, *wait[2], **wait_bh = wait; unsigned int block_end, block_start; unsigned int bsize = 1 << inode->i_blkbits; if (!page_has_buffers(page)) create_empty_buffers(page, bsize, 0); head = page_buffers(page); for (bh = head, block_start = 0; bh != head || !block_start; bh = bh->b_this_page, block_start += bsize) { block_end = block_start + bsize; |
3a307ffc2
|
688 |
clear_buffer_new(bh); |
9517bac6c
|
689 690 691 692 |
/* * Ignore blocks outside of our i/o range - * they may belong to unallocated clusters. */ |
60b11392f
|
693 |
if (block_start >= to || block_end <= from) { |
9517bac6c
|
694 695 696 697 698 699 700 701 702 |
if (PageUptodate(page)) set_buffer_uptodate(bh); continue; } /* * For an allocating write with cluster size >= page * size, we always write the entire page. */ |
3a307ffc2
|
703 704 |
if (new) set_buffer_new(bh); |
9517bac6c
|
705 706 707 708 709 710 711 712 713 714 |
if (!buffer_mapped(bh)) { map_bh(bh, inode->i_sb, *p_blkno); unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr); } if (PageUptodate(page)) { if (!buffer_uptodate(bh)) set_buffer_uptodate(bh); } else if (!buffer_uptodate(bh) && !buffer_delay(bh) && |
bce997682
|
715 716 |
!buffer_new(bh) && (block_start < from || block_end > to)) { |
9517bac6c
|
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 |
ll_rw_block(READ, 1, &bh); *wait_bh++=bh; } *p_blkno = *p_blkno + 1; } /* * If we issued read requests - let them complete. */ while(wait_bh > wait) { wait_on_buffer(*--wait_bh); if (!buffer_uptodate(*wait_bh)) ret = -EIO; } if (ret == 0 || !new) return ret; /* * If we get -EIO above, zero out any newly allocated blocks * to avoid exposing stale data. */ bh = head; block_start = 0; do { |
9517bac6c
|
743 744 745 746 747 |
block_end = block_start + bsize; if (block_end <= from) goto next_bh; if (block_start >= to) break; |
54c57dc3b
|
748 |
zero_user_page(page, block_start, bh->b_size, KM_USER0); |
9517bac6c
|
749 750 751 752 753 754 755 756 757 758 |
set_buffer_uptodate(bh); mark_buffer_dirty(bh); next_bh: block_start = block_end; bh = bh->b_this_page; } while (bh != head); return ret; } |
3a307ffc2
|
759 760 761 762 763 764 765 |
#if (PAGE_CACHE_SIZE >= OCFS2_MAX_CLUSTERSIZE) #define OCFS2_MAX_CTXT_PAGES 1 #else #define OCFS2_MAX_CTXT_PAGES (OCFS2_MAX_CLUSTERSIZE / PAGE_CACHE_SIZE) #endif #define OCFS2_MAX_CLUSTERS_PER_PAGE (PAGE_CACHE_SIZE / OCFS2_MIN_CLUSTERSIZE) |
9517bac6c
|
766 |
/* |
3a307ffc2
|
767 |
* Describe the state of a single cluster to be written to. |
6af67d820
|
768 |
*/ |
3a307ffc2
|
769 770 771 772 773 774 775 776 |
struct ocfs2_write_cluster_desc { u32 c_cpos; u32 c_phys; /* * Give this a unique field because c_phys eventually gets * filled. */ unsigned c_new; |
b27b7cbcf
|
777 |
unsigned c_unwritten; |
3a307ffc2
|
778 |
}; |
6af67d820
|
779 |
|
b27b7cbcf
|
780 781 782 783 |
static inline int ocfs2_should_zero_cluster(struct ocfs2_write_cluster_desc *d) { return d->c_new || d->c_unwritten; } |
3a307ffc2
|
784 785 786 787 |
struct ocfs2_write_ctxt { /* Logical cluster position / len of write */ u32 w_cpos; u32 w_clen; |
6af67d820
|
788 |
|
3a307ffc2
|
789 |
struct ocfs2_write_cluster_desc w_desc[OCFS2_MAX_CLUSTERS_PER_PAGE]; |
6af67d820
|
790 |
|
3a307ffc2
|
791 792 793 794 795 796 797 |
/* * This is true if page_size > cluster_size. * * It triggers a set of special cases during write which might * have to deal with allocating writes to partial pages. */ unsigned int w_large_pages; |
6af67d820
|
798 |
|
3a307ffc2
|
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 |
/* * Pages involved in this write. * * w_target_page is the page being written to by the user. * * w_pages is an array of pages which always contains * w_target_page, and in the case of an allocating write with * page_size < cluster size, it will contain zero'd and mapped * pages adjacent to w_target_page which need to be written * out in so that future reads from that region will get * zero's. */ struct page *w_pages[OCFS2_MAX_CTXT_PAGES]; unsigned int w_num_pages; struct page *w_target_page; |
eeb47d123
|
814 |
|
3a307ffc2
|
815 816 817 818 819 820 821 822 823 824 825 826 827 828 |
/* * ocfs2_write_end() uses this to know what the real range to * write in the target should be. */ unsigned int w_target_from; unsigned int w_target_to; /* * We could use journal_current_handle() but this is cleaner, * IMHO -Mark */ handle_t *w_handle; struct buffer_head *w_di_bh; |
b27b7cbcf
|
829 830 |
struct ocfs2_cached_dealloc_ctxt w_dealloc; |
3a307ffc2
|
831 832 833 834 835 836 837 838 839 840 841 842 843 |
}; static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc) { int i; for(i = 0; i < wc->w_num_pages; i++) { if (wc->w_pages[i] == NULL) continue; unlock_page(wc->w_pages[i]); mark_page_accessed(wc->w_pages[i]); page_cache_release(wc->w_pages[i]); |
6af67d820
|
844 |
} |
3a307ffc2
|
845 846 847 848 849 850 |
brelse(wc->w_di_bh); kfree(wc); } static int ocfs2_alloc_write_ctxt(struct ocfs2_write_ctxt **wcp, struct ocfs2_super *osb, loff_t pos, |
607d44aa3
|
851 |
unsigned len, struct buffer_head *di_bh) |
3a307ffc2
|
852 853 854 855 856 857 |
{ struct ocfs2_write_ctxt *wc; wc = kzalloc(sizeof(struct ocfs2_write_ctxt), GFP_NOFS); if (!wc) return -ENOMEM; |
6af67d820
|
858 |
|
3a307ffc2
|
859 860 |
wc->w_cpos = pos >> osb->s_clustersize_bits; wc->w_clen = ocfs2_clusters_for_bytes(osb->sb, len); |
607d44aa3
|
861 862 |
get_bh(di_bh); wc->w_di_bh = di_bh; |
6af67d820
|
863 |
|
3a307ffc2
|
864 865 866 867 |
if (unlikely(PAGE_CACHE_SHIFT > osb->s_clustersize_bits)) wc->w_large_pages = 1; else wc->w_large_pages = 0; |
b27b7cbcf
|
868 |
ocfs2_init_dealloc_ctxt(&wc->w_dealloc); |
3a307ffc2
|
869 |
*wcp = wc; |
6af67d820
|
870 |
|
3a307ffc2
|
871 |
return 0; |
6af67d820
|
872 873 874 |
} /* |
3a307ffc2
|
875 876 877 |
* If a page has any new buffers, zero them out here, and mark them uptodate * and dirty so they'll be written out (in order to prevent uninitialised * block data from leaking). And clear the new bit. |
9517bac6c
|
878 |
*/ |
3a307ffc2
|
879 |
static void ocfs2_zero_new_buffers(struct page *page, unsigned from, unsigned to) |
9517bac6c
|
880 |
{ |
3a307ffc2
|
881 882 |
unsigned int block_start, block_end; struct buffer_head *head, *bh; |
9517bac6c
|
883 |
|
3a307ffc2
|
884 885 886 |
BUG_ON(!PageLocked(page)); if (!page_has_buffers(page)) return; |
9517bac6c
|
887 |
|
3a307ffc2
|
888 889 890 891 892 893 894 895 896 |
bh = head = page_buffers(page); block_start = 0; do { block_end = block_start + bh->b_size; if (buffer_new(bh)) { if (block_end > from && block_start < to) { if (!PageUptodate(page)) { unsigned start, end; |
3a307ffc2
|
897 898 899 |
start = max(from, block_start); end = min(to, block_end); |
54c57dc3b
|
900 |
zero_user_page(page, start, end - start, KM_USER0); |
3a307ffc2
|
901 902 903 904 905 906 907 |
set_buffer_uptodate(bh); } clear_buffer_new(bh); mark_buffer_dirty(bh); } } |
9517bac6c
|
908 |
|
3a307ffc2
|
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 |
block_start = block_end; bh = bh->b_this_page; } while (bh != head); } /* * Only called when we have a failure during allocating write to write * zero's to the newly allocated region. */ static void ocfs2_write_failure(struct inode *inode, struct ocfs2_write_ctxt *wc, loff_t user_pos, unsigned user_len) { int i; unsigned from, to; struct page *tmppage; ocfs2_zero_new_buffers(wc->w_target_page, user_pos, user_len); |
9517bac6c
|
927 |
|
9517bac6c
|
928 |
if (wc->w_large_pages) { |
3a307ffc2
|
929 930 |
from = wc->w_target_from; to = wc->w_target_to; |
9517bac6c
|
931 |
} else { |
3a307ffc2
|
932 933 |
from = 0; to = PAGE_CACHE_SIZE; |
9517bac6c
|
934 |
} |
3a307ffc2
|
935 936 |
for(i = 0; i < wc->w_num_pages; i++) { tmppage = wc->w_pages[i]; |
9517bac6c
|
937 |
|
3a307ffc2
|
938 939 940 941 |
if (ocfs2_should_order_data(inode)) walk_page_buffers(wc->w_handle, page_buffers(tmppage), from, to, NULL, ocfs2_journal_dirty_data); |
eeb47d123
|
942 |
|
3a307ffc2
|
943 |
block_commit_write(tmppage, from, to); |
9517bac6c
|
944 |
} |
9517bac6c
|
945 |
} |
3a307ffc2
|
946 947 948 949 950 |
static int ocfs2_prepare_page_for_write(struct inode *inode, u64 *p_blkno, struct ocfs2_write_ctxt *wc, struct page *page, u32 cpos, loff_t user_pos, unsigned user_len, int new) |
9517bac6c
|
951 |
{ |
3a307ffc2
|
952 953 |
int ret; unsigned int map_from = 0, map_to = 0; |
9517bac6c
|
954 |
unsigned int cluster_start, cluster_end; |
3a307ffc2
|
955 |
unsigned int user_data_from = 0, user_data_to = 0; |
9517bac6c
|
956 |
|
3a307ffc2
|
957 |
ocfs2_figure_cluster_boundaries(OCFS2_SB(inode->i_sb), cpos, |
9517bac6c
|
958 |
&cluster_start, &cluster_end); |
3a307ffc2
|
959 960 961 962 963 964 965 966 967 968 969 970 |
if (page == wc->w_target_page) { map_from = user_pos & (PAGE_CACHE_SIZE - 1); map_to = map_from + user_len; if (new) ret = ocfs2_map_page_blocks(page, p_blkno, inode, cluster_start, cluster_end, new); else ret = ocfs2_map_page_blocks(page, p_blkno, inode, map_from, map_to, new); if (ret) { |
9517bac6c
|
971 972 973 |
mlog_errno(ret); goto out; } |
3a307ffc2
|
974 975 |
user_data_from = map_from; user_data_to = map_to; |
9517bac6c
|
976 |
if (new) { |
3a307ffc2
|
977 978 |
map_from = cluster_start; map_to = cluster_end; |
9517bac6c
|
979 |
} |
3a307ffc2
|
980 981 982 |
wc->w_target_from = map_from; wc->w_target_to = map_to; |
9517bac6c
|
983 984 985 986 987 988 989 |
} else { /* * If we haven't allocated the new page yet, we * shouldn't be writing it out without copying user * data. This is likely a math error from the caller. */ BUG_ON(!new); |
3a307ffc2
|
990 991 |
map_from = cluster_start; map_to = cluster_end; |
9517bac6c
|
992 993 |
ret = ocfs2_map_page_blocks(page, p_blkno, inode, |
3a307ffc2
|
994 |
cluster_start, cluster_end, new); |
9517bac6c
|
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 |
if (ret) { mlog_errno(ret); goto out; } } /* * Parts of newly allocated pages need to be zero'd. * * Above, we have also rewritten 'to' and 'from' - as far as * the rest of the function is concerned, the entire cluster * range inside of a page needs to be written. * * We can skip this if the page is up to date - it's already * been zero'd from being read in as a hole. */ if (new && !PageUptodate(page)) ocfs2_clear_page_regions(page, OCFS2_SB(inode->i_sb), |
3a307ffc2
|
1013 |
cpos, user_data_from, user_data_to); |
9517bac6c
|
1014 1015 |
flush_dcache_page(page); |
9517bac6c
|
1016 |
out: |
3a307ffc2
|
1017 |
return ret; |
9517bac6c
|
1018 1019 1020 |
} /* |
3a307ffc2
|
1021 |
* This function will only grab one clusters worth of pages. |
9517bac6c
|
1022 |
*/ |
3a307ffc2
|
1023 1024 |
static int ocfs2_grab_pages_for_write(struct address_space *mapping, struct ocfs2_write_ctxt *wc, |
7307de805
|
1025 1026 |
u32 cpos, loff_t user_pos, int new, struct page *mmap_page) |
9517bac6c
|
1027 |
{ |
3a307ffc2
|
1028 1029 |
int ret = 0, i; unsigned long start, target_index, index; |
9517bac6c
|
1030 |
struct inode *inode = mapping->host; |
9517bac6c
|
1031 |
|
3a307ffc2
|
1032 |
target_index = user_pos >> PAGE_CACHE_SHIFT; |
9517bac6c
|
1033 1034 1035 |
/* * Figure out how many pages we'll be manipulating here. For |
60b11392f
|
1036 1037 |
* non allocating write, we just change the one * page. Otherwise, we'll need a whole clusters worth. |
9517bac6c
|
1038 |
*/ |
9517bac6c
|
1039 |
if (new) { |
3a307ffc2
|
1040 1041 |
wc->w_num_pages = ocfs2_pages_per_cluster(inode->i_sb); start = ocfs2_align_clusters_to_page_index(inode->i_sb, cpos); |
9517bac6c
|
1042 |
} else { |
3a307ffc2
|
1043 1044 |
wc->w_num_pages = 1; start = target_index; |
9517bac6c
|
1045 |
} |
3a307ffc2
|
1046 |
for(i = 0; i < wc->w_num_pages; i++) { |
9517bac6c
|
1047 |
index = start + i; |
7307de805
|
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 |
if (index == target_index && mmap_page) { /* * ocfs2_pagemkwrite() is a little different * and wants us to directly use the page * passed in. */ lock_page(mmap_page); if (mmap_page->mapping != mapping) { unlock_page(mmap_page); /* * Sanity check - the locking in * ocfs2_pagemkwrite() should ensure * that this code doesn't trigger. */ ret = -EINVAL; mlog_errno(ret); goto out; } page_cache_get(mmap_page); wc->w_pages[i] = mmap_page; } else { wc->w_pages[i] = find_or_create_page(mapping, index, GFP_NOFS); if (!wc->w_pages[i]) { ret = -ENOMEM; mlog_errno(ret); goto out; } |
9517bac6c
|
1078 |
} |
3a307ffc2
|
1079 1080 1081 |
if (index == target_index) wc->w_target_page = wc->w_pages[i]; |
9517bac6c
|
1082 |
} |
3a307ffc2
|
1083 1084 1085 1086 1087 1088 1089 1090 |
out: return ret; } /* * Prepare a single cluster for write one cluster into the file. */ static int ocfs2_write_cluster(struct address_space *mapping, |
b27b7cbcf
|
1091 1092 |
u32 phys, unsigned int unwritten, struct ocfs2_alloc_context *data_ac, |
3a307ffc2
|
1093 1094 1095 1096 |
struct ocfs2_alloc_context *meta_ac, struct ocfs2_write_ctxt *wc, u32 cpos, loff_t user_pos, unsigned user_len) { |
b27b7cbcf
|
1097 |
int ret, i, new, should_zero = 0; |
3a307ffc2
|
1098 1099 1100 1101 |
u64 v_blkno, p_blkno; struct inode *inode = mapping->host; new = phys == 0 ? 1 : 0; |
b27b7cbcf
|
1102 1103 |
if (new || unwritten) should_zero = 1; |
9517bac6c
|
1104 1105 |
if (new) { |
3a307ffc2
|
1106 |
u32 tmp_pos; |
9517bac6c
|
1107 1108 1109 1110 |
/* * This is safe to call with the page locks - it won't take * any additional semaphores or cluster locks. */ |
3a307ffc2
|
1111 |
tmp_pos = cpos; |
9517bac6c
|
1112 |
ret = ocfs2_do_extend_allocation(OCFS2_SB(inode->i_sb), inode, |
2ae99a603
|
1113 |
&tmp_pos, 1, 0, wc->w_di_bh, |
3a307ffc2
|
1114 1115 |
wc->w_handle, data_ac, meta_ac, NULL); |
9517bac6c
|
1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 |
/* * This shouldn't happen because we must have already * calculated the correct meta data allocation required. The * internal tree allocation code should know how to increase * transaction credits itself. * * If need be, we could handle -EAGAIN for a * RESTART_TRANS here. */ mlog_bug_on_msg(ret == -EAGAIN, "Inode %llu: EAGAIN return during allocation. ", (unsigned long long)OCFS2_I(inode)->ip_blkno); if (ret < 0) { mlog_errno(ret); goto out; } |
b27b7cbcf
|
1133 1134 1135 1136 1137 1138 1139 1140 1141 |
} else if (unwritten) { ret = ocfs2_mark_extent_written(inode, wc->w_di_bh, wc->w_handle, cpos, 1, phys, meta_ac, &wc->w_dealloc); if (ret < 0) { mlog_errno(ret); goto out; } } |
3a307ffc2
|
1142 |
|
b27b7cbcf
|
1143 |
if (should_zero) |
3a307ffc2
|
1144 |
v_blkno = ocfs2_clusters_to_blocks(inode->i_sb, cpos); |
b27b7cbcf
|
1145 |
else |
3a307ffc2
|
1146 |
v_blkno = user_pos >> inode->i_sb->s_blocksize_bits; |
9517bac6c
|
1147 |
|
3a307ffc2
|
1148 1149 1150 1151 |
/* * The only reason this should fail is due to an inability to * find the extent added. */ |
49cb8d2d4
|
1152 1153 |
ret = ocfs2_extent_map_get_blocks(inode, v_blkno, &p_blkno, NULL, NULL); |
9517bac6c
|
1154 |
if (ret < 0) { |
3a307ffc2
|
1155 1156 1157 1158 |
ocfs2_error(inode->i_sb, "Corrupting extend for inode %llu, " "at logical block %llu", (unsigned long long)OCFS2_I(inode)->ip_blkno, (unsigned long long)v_blkno); |
9517bac6c
|
1159 1160 1161 1162 |
goto out; } BUG_ON(p_blkno == 0); |
3a307ffc2
|
1163 1164 |
for(i = 0; i < wc->w_num_pages; i++) { int tmpret; |
9517bac6c
|
1165 |
|
3a307ffc2
|
1166 1167 |
tmpret = ocfs2_prepare_page_for_write(inode, &p_blkno, wc, wc->w_pages[i], cpos, |
b27b7cbcf
|
1168 1169 |
user_pos, user_len, should_zero); |
3a307ffc2
|
1170 1171 1172 1173 1174 |
if (tmpret) { mlog_errno(tmpret); if (ret == 0) tmpret = ret; } |
9517bac6c
|
1175 |
} |
3a307ffc2
|
1176 1177 1178 1179 1180 |
/* * We only have cleanup to do in case of allocating write. */ if (ret && new) ocfs2_write_failure(inode, wc, user_pos, user_len); |
9517bac6c
|
1181 |
out: |
9517bac6c
|
1182 |
|
3a307ffc2
|
1183 |
return ret; |
9517bac6c
|
1184 |
} |
0d172baa5
|
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 |
static int ocfs2_write_cluster_by_desc(struct address_space *mapping, struct ocfs2_alloc_context *data_ac, struct ocfs2_alloc_context *meta_ac, struct ocfs2_write_ctxt *wc, loff_t pos, unsigned len) { int ret, i; struct ocfs2_write_cluster_desc *desc; for (i = 0; i < wc->w_clen; i++) { desc = &wc->w_desc[i]; |
b27b7cbcf
|
1196 1197 1198 |
ret = ocfs2_write_cluster(mapping, desc->c_phys, desc->c_unwritten, data_ac, meta_ac, wc, desc->c_cpos, pos, len); |
0d172baa5
|
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 |
if (ret) { mlog_errno(ret); goto out; } } ret = 0; out: return ret; } |
3a307ffc2
|
1209 1210 1211 1212 1213 1214 1215 1216 |
/* * ocfs2_write_end() wants to know which parts of the target page it * should complete the write on. It's easiest to compute them ahead of * time when a more complete view of the write is available. */ static void ocfs2_set_target_boundaries(struct ocfs2_super *osb, struct ocfs2_write_ctxt *wc, loff_t pos, unsigned len, int alloc) |
9517bac6c
|
1217 |
{ |
3a307ffc2
|
1218 |
struct ocfs2_write_cluster_desc *desc; |
9517bac6c
|
1219 |
|
3a307ffc2
|
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 |
wc->w_target_from = pos & (PAGE_CACHE_SIZE - 1); wc->w_target_to = wc->w_target_from + len; if (alloc == 0) return; /* * Allocating write - we may have different boundaries based * on page size and cluster size. * * NOTE: We can no longer compute one value from the other as * the actual write length and user provided length may be * different. */ |
9517bac6c
|
1234 |
|
3a307ffc2
|
1235 1236 1237 |
if (wc->w_large_pages) { /* * We only care about the 1st and last cluster within |
b27b7cbcf
|
1238 |
* our range and whether they should be zero'd or not. Either |
3a307ffc2
|
1239 1240 1241 1242 |
* value may be extended out to the start/end of a * newly allocated cluster. */ desc = &wc->w_desc[0]; |
b27b7cbcf
|
1243 |
if (ocfs2_should_zero_cluster(desc)) |
3a307ffc2
|
1244 1245 1246 1247 1248 1249 |
ocfs2_figure_cluster_boundaries(osb, desc->c_cpos, &wc->w_target_from, NULL); desc = &wc->w_desc[wc->w_clen - 1]; |
b27b7cbcf
|
1250 |
if (ocfs2_should_zero_cluster(desc)) |
3a307ffc2
|
1251 1252 1253 1254 1255 1256 1257 1258 |
ocfs2_figure_cluster_boundaries(osb, desc->c_cpos, NULL, &wc->w_target_to); } else { wc->w_target_from = 0; wc->w_target_to = PAGE_CACHE_SIZE; } |
9517bac6c
|
1259 |
} |
0d172baa5
|
1260 1261 1262 |
/* * Populate each single-cluster write descriptor in the write context * with information about the i/o to be done. |
b27b7cbcf
|
1263 1264 1265 1266 |
* * Returns the number of clusters that will have to be allocated, as * well as a worst case estimate of the number of extent records that * would have to be created during a write to an unwritten region. |
0d172baa5
|
1267 1268 1269 |
*/ static int ocfs2_populate_write_desc(struct inode *inode, struct ocfs2_write_ctxt *wc, |
b27b7cbcf
|
1270 1271 |
unsigned int *clusters_to_alloc, unsigned int *extents_to_split) |
9517bac6c
|
1272 |
{ |
0d172baa5
|
1273 |
int ret; |
3a307ffc2
|
1274 |
struct ocfs2_write_cluster_desc *desc; |
0d172baa5
|
1275 |
unsigned int num_clusters = 0; |
b27b7cbcf
|
1276 |
unsigned int ext_flags = 0; |
0d172baa5
|
1277 1278 |
u32 phys = 0; int i; |
9517bac6c
|
1279 |
|
b27b7cbcf
|
1280 1281 |
*clusters_to_alloc = 0; *extents_to_split = 0; |
3a307ffc2
|
1282 1283 1284 1285 1286 |
for (i = 0; i < wc->w_clen; i++) { desc = &wc->w_desc[i]; desc->c_cpos = wc->w_cpos + i; if (num_clusters == 0) { |
b27b7cbcf
|
1287 1288 1289 |
/* * Need to look up the next extent record. */ |
3a307ffc2
|
1290 |
ret = ocfs2_get_clusters(inode, desc->c_cpos, &phys, |
b27b7cbcf
|
1291 |
&num_clusters, &ext_flags); |
3a307ffc2
|
1292 1293 |
if (ret) { mlog_errno(ret); |
607d44aa3
|
1294 |
goto out; |
3a307ffc2
|
1295 |
} |
b27b7cbcf
|
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 |
/* * Assume worst case - that we're writing in * the middle of the extent. * * We can assume that the write proceeds from * left to right, in which case the extent * insert code is smart enough to coalesce the * next splits into the previous records created. */ if (ext_flags & OCFS2_EXT_UNWRITTEN) *extents_to_split = *extents_to_split + 2; |
3a307ffc2
|
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 |
} else if (phys) { /* * Only increment phys if it doesn't describe * a hole. */ phys++; } desc->c_phys = phys; if (phys == 0) { desc->c_new = 1; |
0d172baa5
|
1319 |
*clusters_to_alloc = *clusters_to_alloc + 1; |
3a307ffc2
|
1320 |
} |
b27b7cbcf
|
1321 1322 |
if (ext_flags & OCFS2_EXT_UNWRITTEN) desc->c_unwritten = 1; |
3a307ffc2
|
1323 1324 |
num_clusters--; |
9517bac6c
|
1325 |
} |
0d172baa5
|
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 |
ret = 0; out: return ret; } int ocfs2_write_begin_nolock(struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata, struct buffer_head *di_bh, struct page *mmap_page) { int ret, credits = OCFS2_INODE_UPDATE_CREDITS; |
b27b7cbcf
|
1337 |
unsigned int clusters_to_alloc, extents_to_split; |
0d172baa5
|
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 |
struct ocfs2_write_ctxt *wc; struct inode *inode = mapping->host; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_dinode *di; struct ocfs2_alloc_context *data_ac = NULL; struct ocfs2_alloc_context *meta_ac = NULL; handle_t *handle; ret = ocfs2_alloc_write_ctxt(&wc, osb, pos, len, di_bh); if (ret) { mlog_errno(ret); return ret; } |
b27b7cbcf
|
1351 1352 |
ret = ocfs2_populate_write_desc(inode, wc, &clusters_to_alloc, &extents_to_split); |
0d172baa5
|
1353 1354 1355 1356 1357 1358 |
if (ret) { mlog_errno(ret); goto out; } di = (struct ocfs2_dinode *)wc->w_di_bh->b_data; |
3a307ffc2
|
1359 1360 1361 1362 1363 1364 |
/* * We set w_target_from, w_target_to here so that * ocfs2_write_end() knows which range in the target page to * write out. An allocation requires that we write the entire * cluster range. */ |
b27b7cbcf
|
1365 |
if (clusters_to_alloc || extents_to_split) { |
3a307ffc2
|
1366 1367 |
/* * XXX: We are stretching the limits of |
b27b7cbcf
|
1368 |
* ocfs2_lock_allocators(). It greatly over-estimates |
3a307ffc2
|
1369 1370 1371 |
* the work to be done. */ ret = ocfs2_lock_allocators(inode, di, clusters_to_alloc, |
b27b7cbcf
|
1372 |
extents_to_split, &data_ac, &meta_ac); |
9517bac6c
|
1373 1374 |
if (ret) { mlog_errno(ret); |
607d44aa3
|
1375 |
goto out; |
9517bac6c
|
1376 |
} |
3a307ffc2
|
1377 1378 |
credits = ocfs2_calc_extend_credits(inode->i_sb, di, clusters_to_alloc); |
9517bac6c
|
1379 |
} |
b27b7cbcf
|
1380 1381 |
ocfs2_set_target_boundaries(osb, wc, pos, len, clusters_to_alloc + extents_to_split); |
3a307ffc2
|
1382 |
|
9517bac6c
|
1383 1384 1385 1386 |
handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { ret = PTR_ERR(handle); mlog_errno(ret); |
607d44aa3
|
1387 |
goto out; |
9517bac6c
|
1388 |
} |
3a307ffc2
|
1389 1390 1391 1392 1393 1394 1395 1396 1397 |
wc->w_handle = handle; /* * We don't want this to fail in ocfs2_write_end(), so do it * here. */ ret = ocfs2_journal_access(handle, inode, wc->w_di_bh, OCFS2_JOURNAL_ACCESS_WRITE); if (ret) { |
9517bac6c
|
1398 1399 1400 |
mlog_errno(ret); goto out_commit; } |
3a307ffc2
|
1401 1402 1403 1404 1405 1406 |
/* * Fill our page array first. That way we've grabbed enough so * that we can zero and flush if we error after adding the * extent. */ ret = ocfs2_grab_pages_for_write(mapping, wc, wc->w_cpos, pos, |
b27b7cbcf
|
1407 1408 |
clusters_to_alloc + extents_to_split, mmap_page); |
9517bac6c
|
1409 1410 1411 1412 |
if (ret) { mlog_errno(ret); goto out_commit; } |
0d172baa5
|
1413 1414 1415 1416 1417 |
ret = ocfs2_write_cluster_by_desc(mapping, data_ac, meta_ac, wc, pos, len); if (ret) { mlog_errno(ret); goto out_commit; |
9517bac6c
|
1418 |
} |
9517bac6c
|
1419 |
|
3a307ffc2
|
1420 1421 1422 1423 |
if (data_ac) ocfs2_free_alloc_context(data_ac); if (meta_ac) ocfs2_free_alloc_context(meta_ac); |
9517bac6c
|
1424 |
|
3a307ffc2
|
1425 1426 1427 |
*pagep = wc->w_target_page; *fsdata = wc; return 0; |
9517bac6c
|
1428 1429 |
out_commit: ocfs2_commit_trans(osb, handle); |
9517bac6c
|
1430 |
out: |
3a307ffc2
|
1431 |
ocfs2_free_write_ctxt(wc); |
9517bac6c
|
1432 1433 1434 1435 |
if (data_ac) ocfs2_free_alloc_context(data_ac); if (meta_ac) ocfs2_free_alloc_context(meta_ac); |
3a307ffc2
|
1436 1437 |
return ret; } |
607d44aa3
|
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 |
int ocfs2_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) { int ret; struct buffer_head *di_bh = NULL; struct inode *inode = mapping->host; ret = ocfs2_meta_lock(inode, &di_bh, 1); if (ret) { mlog_errno(ret); return ret; } /* * Take alloc sem here to prevent concurrent lookups. That way * the mapping, zeroing and tree manipulation within * ocfs2_write() will be safe against ->readpage(). This * should also serve to lock out allocation from a shared * writeable region. */ down_write(&OCFS2_I(inode)->ip_alloc_sem); ret = ocfs2_data_lock(inode, 1); if (ret) { mlog_errno(ret); goto out_fail; } ret = ocfs2_write_begin_nolock(mapping, pos, len, flags, pagep, |
7307de805
|
1468 |
fsdata, di_bh, NULL); |
607d44aa3
|
1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 |
if (ret) { mlog_errno(ret); goto out_fail_data; } brelse(di_bh); return 0; out_fail_data: ocfs2_data_unlock(inode, 1); out_fail: up_write(&OCFS2_I(inode)->ip_alloc_sem); brelse(di_bh); ocfs2_meta_unlock(inode, 1); return ret; } |
7307de805
|
1488 1489 1490 |
int ocfs2_write_end_nolock(struct address_space *mapping, loff_t pos, unsigned len, unsigned copied, struct page *page, void *fsdata) |
3a307ffc2
|
1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 |
{ int i; unsigned from, to, start = pos & (PAGE_CACHE_SIZE - 1); struct inode *inode = mapping->host; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_write_ctxt *wc = fsdata; struct ocfs2_dinode *di = (struct ocfs2_dinode *)wc->w_di_bh->b_data; handle_t *handle = wc->w_handle; struct page *tmppage; if (unlikely(copied < len)) { if (!PageUptodate(wc->w_target_page)) copied = 0; ocfs2_zero_new_buffers(wc->w_target_page, start+copied, start+len); } flush_dcache_page(wc->w_target_page); for(i = 0; i < wc->w_num_pages; i++) { tmppage = wc->w_pages[i]; if (tmppage == wc->w_target_page) { from = wc->w_target_from; to = wc->w_target_to; BUG_ON(from > PAGE_CACHE_SIZE || to > PAGE_CACHE_SIZE || to < from); } else { /* * Pages adjacent to the target (if any) imply * a hole-filling write in which case we want * to flush their entire range. */ from = 0; to = PAGE_CACHE_SIZE; } if (ocfs2_should_order_data(inode)) walk_page_buffers(wc->w_handle, page_buffers(tmppage), from, to, NULL, ocfs2_journal_dirty_data); block_commit_write(tmppage, from, to); } pos += copied; if (pos > inode->i_size) { i_size_write(inode, pos); mark_inode_dirty(inode); } inode->i_blocks = ocfs2_inode_sector_count(inode); di->i_size = cpu_to_le64((u64)i_size_read(inode)); inode->i_mtime = inode->i_ctime = CURRENT_TIME; di->i_mtime = di->i_ctime = cpu_to_le64(inode->i_mtime.tv_sec); di->i_mtime_nsec = di->i_ctime_nsec = cpu_to_le32(inode->i_mtime.tv_nsec); |
3a307ffc2
|
1548 1549 1550 |
ocfs2_journal_dirty(handle, wc->w_di_bh); ocfs2_commit_trans(osb, handle); |
59a5e416d
|
1551 |
|
b27b7cbcf
|
1552 |
ocfs2_run_deallocs(osb, &wc->w_dealloc); |
607d44aa3
|
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 |
ocfs2_free_write_ctxt(wc); return copied; } int ocfs2_write_end(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned copied, struct page *page, void *fsdata) { int ret; struct inode *inode = mapping->host; ret = ocfs2_write_end_nolock(mapping, pos, len, copied, page, fsdata); |
3a307ffc2
|
1566 1567 1568 |
ocfs2_data_unlock(inode, 1); up_write(&OCFS2_I(inode)->ip_alloc_sem); ocfs2_meta_unlock(inode, 1); |
9517bac6c
|
1569 |
|
607d44aa3
|
1570 |
return ret; |
9517bac6c
|
1571 |
} |
f5e54d6e5
|
1572 |
const struct address_space_operations ocfs2_aops = { |
ccd979bdb
|
1573 1574 |
.readpage = ocfs2_readpage, .writepage = ocfs2_writepage, |
ccd979bdb
|
1575 1576 |
.bmap = ocfs2_bmap, .sync_page = block_sync_page, |
03f981cf2
|
1577 1578 1579 1580 |
.direct_IO = ocfs2_direct_IO, .invalidatepage = ocfs2_invalidatepage, .releasepage = ocfs2_releasepage, .migratepage = buffer_migrate_page, |
ccd979bdb
|
1581 |
}; |