Blame view
fs/jbd2/commit.c
31.4 KB
470decc61
|
1 |
/* |
f7f4bccb7
|
2 |
* linux/fs/jbd2/commit.c |
470decc61
|
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
* * Written by Stephen C. Tweedie <sct@redhat.com>, 1998 * * Copyright 1998 Red Hat corp --- All Rights Reserved * * This file is part of the Linux kernel and is made available under * the terms of the GNU General Public License, version 2, or at your * option, any later version, incorporated herein by reference. * * Journal commit routines for the generic filesystem journaling code; * part of the ext2fs journaling system. */ #include <linux/time.h> #include <linux/fs.h> |
f7f4bccb7
|
18 |
#include <linux/jbd2.h> |
470decc61
|
19 20 21 22 |
#include <linux/errno.h> #include <linux/slab.h> #include <linux/mm.h> #include <linux/pagemap.h> |
8e85fb3f3
|
23 |
#include <linux/jiffies.h> |
818d276ce
|
24 |
#include <linux/crc32.h> |
cd1aac329
|
25 26 |
#include <linux/writeback.h> #include <linux/backing-dev.h> |
fd98496f4
|
27 |
#include <linux/bio.h> |
0e3d2a631
|
28 |
#include <linux/blkdev.h> |
39e3ac259
|
29 |
#include <linux/bitops.h> |
879c5e6b7
|
30 |
#include <trace/events/jbd2.h> |
39e3ac259
|
31 |
#include <asm/system.h> |
470decc61
|
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 |
/* * Default IO end handler for temporary BJ_IO buffer_heads. */ static void journal_end_buffer_io_sync(struct buffer_head *bh, int uptodate) { BUFFER_TRACE(bh, ""); if (uptodate) set_buffer_uptodate(bh); else clear_buffer_uptodate(bh); unlock_buffer(bh); } /* |
87c89c232
|
47 48 |
* When an ext4 file is truncated, it is possible that some pages are not * successfully freed, because they are attached to a committing transaction. |
470decc61
|
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 |
* After the transaction commits, these pages are left on the LRU, with no * ->mapping, and with attached buffers. These pages are trivially reclaimable * by the VM, but their apparent absence upsets the VM accounting, and it makes * the numbers in /proc/meminfo look odd. * * So here, we have a buffer which has just come off the forget list. Look to * see if we can strip all buffers from the backing page. * * Called under lock_journal(), and possibly under journal_datalist_lock. The * caller provided us with a ref against the buffer, and we drop that here. */ static void release_buffer_page(struct buffer_head *bh) { struct page *page; if (buffer_dirty(bh)) goto nope; if (atomic_read(&bh->b_count) != 1) goto nope; page = bh->b_page; if (!page) goto nope; if (page->mapping) goto nope; /* OK, it's a truncated page */ |
529ae9aaa
|
75 |
if (!trylock_page(page)) |
470decc61
|
76 77 78 79 80 81 82 83 84 85 86 87 88 89 |
goto nope; page_cache_get(page); __brelse(bh); try_to_free_buffers(page); unlock_page(page); page_cache_release(page); return; nope: __brelse(bh); } /* |
818d276ce
|
90 |
* Done it all: now submit the commit record. We should have |
470decc61
|
91 92 93 94 95 96 |
* cleaned up our previous buffers by now, so if we are in abort * mode we can now just skip the rest of the journal write * entirely. * * Returns 1 if the journal needs to be aborted or 0 on success */ |
818d276ce
|
97 98 99 100 |
static int journal_submit_commit_record(journal_t *journal, transaction_t *commit_transaction, struct buffer_head **cbh, __u32 crc32_sum) |
470decc61
|
101 102 |
{ struct journal_head *descriptor; |
818d276ce
|
103 |
struct commit_header *tmp; |
470decc61
|
104 |
struct buffer_head *bh; |
818d276ce
|
105 |
int ret; |
736603ab2
|
106 |
struct timespec now = current_kernel_time(); |
470decc61
|
107 |
|
6cba611e6
|
108 |
*cbh = NULL; |
470decc61
|
109 110 |
if (is_journal_aborted(journal)) return 0; |
f7f4bccb7
|
111 |
descriptor = jbd2_journal_get_descriptor_buffer(journal); |
470decc61
|
112 113 114 115 |
if (!descriptor) return 1; bh = jh2bh(descriptor); |
818d276ce
|
116 117 118 119 |
tmp = (struct commit_header *)bh->b_data; tmp->h_magic = cpu_to_be32(JBD2_MAGIC_NUMBER); tmp->h_blocktype = cpu_to_be32(JBD2_COMMIT_BLOCK); tmp->h_sequence = cpu_to_be32(commit_transaction->t_tid); |
736603ab2
|
120 121 |
tmp->h_commit_sec = cpu_to_be64(now.tv_sec); tmp->h_commit_nsec = cpu_to_be32(now.tv_nsec); |
818d276ce
|
122 123 124 125 126 127 |
if (JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM)) { tmp->h_chksum_type = JBD2_CRC32_CHKSUM; tmp->h_chksum_size = JBD2_CRC32_CHKSUM_SIZE; tmp->h_chksum[0] = cpu_to_be32(crc32_sum); |
470decc61
|
128 |
} |
818d276ce
|
129 130 |
JBUFFER_TRACE(descriptor, "submit commit block"); lock_buffer(bh); |
45a90bfd9
|
131 |
clear_buffer_dirty(bh); |
818d276ce
|
132 133 134 135 |
set_buffer_uptodate(bh); bh->b_end_io = journal_end_buffer_io_sync; if (journal->j_flags & JBD2_BARRIER && |
0e3d2a631
|
136 |
!JBD2_HAS_INCOMPAT_FEATURE(journal, |
9c35575bb
|
137 |
JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) |
721a9602e
|
138 |
ret = submit_bh(WRITE_SYNC | WRITE_FLUSH_FUA, bh); |
9c35575bb
|
139 |
else |
721a9602e
|
140 |
ret = submit_bh(WRITE_SYNC, bh); |
9c35575bb
|
141 |
|
818d276ce
|
142 143 144 145 146 147 148 149 |
*cbh = bh; return ret; } /* * This function along with journal_submit_commit_record * allows to write the commit record asynchronously. */ |
fd98496f4
|
150 151 |
static int journal_wait_on_commit_record(journal_t *journal, struct buffer_head *bh) |
818d276ce
|
152 153 154 155 156 |
{ int ret = 0; clear_buffer_dirty(bh); wait_on_buffer(bh); |
470decc61
|
157 |
|
818d276ce
|
158 159 160 161 162 163 |
if (unlikely(!buffer_uptodate(bh))) ret = -EIO; put_bh(bh); /* One for getblk() */ jbd2_journal_put_journal_head(bh2jh(bh)); return ret; |
470decc61
|
164 |
} |
818d276ce
|
165 |
/* |
cd1aac329
|
166 167 168 169 170 171 172 173 174 175 176 177 178 |
* write the filemap data using writepage() address_space_operations. * We don't do block allocation here even for delalloc. We don't * use writepages() because with dealyed allocation we may be doing * block allocation in writepages(). */ static int journal_submit_inode_data_buffers(struct address_space *mapping) { int ret; struct writeback_control wbc = { .sync_mode = WB_SYNC_ALL, .nr_to_write = mapping->nrpages * 2, .range_start = 0, .range_end = i_size_read(mapping->host), |
cd1aac329
|
179 180 181 182 183 184 185 |
}; ret = generic_writepages(mapping, &wbc); return ret; } /* |
c851ed540
|
186 187 188 189 190 191 192 |
* Submit all the data buffers of inode associated with the transaction to * disk. * * We are in a committing transaction. Therefore no new inode can be added to * our inode list. We use JI_COMMIT_RUNNING flag to protect inode we currently * operate on from being released while we write out pages. */ |
cd1aac329
|
193 |
static int journal_submit_data_buffers(journal_t *journal, |
c851ed540
|
194 195 196 197 198 199 200 201 202 |
transaction_t *commit_transaction) { struct jbd2_inode *jinode; int err, ret = 0; struct address_space *mapping; spin_lock(&journal->j_list_lock); list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) { mapping = jinode->i_vfs_inode->i_mapping; |
39e3ac259
|
203 |
set_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); |
c851ed540
|
204 |
spin_unlock(&journal->j_list_lock); |
cd1aac329
|
205 206 207 208 209 210 |
/* * submit the inode data buffers. We use writepage * instead of writepages. Because writepages can do * block allocation with delalloc. We need to write * only allocated blocks here. */ |
879c5e6b7
|
211 |
trace_jbd2_submit_inode_data(jinode->i_vfs_inode); |
cd1aac329
|
212 |
err = journal_submit_inode_data_buffers(mapping); |
c851ed540
|
213 214 215 216 |
if (!ret) ret = err; spin_lock(&journal->j_list_lock); J_ASSERT(jinode->i_transaction == commit_transaction); |
39e3ac259
|
217 218 |
clear_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); smp_mb__after_clear_bit(); |
c851ed540
|
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 |
wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING); } spin_unlock(&journal->j_list_lock); return ret; } /* * Wait for data submitted for writeout, refile inodes to proper * transaction if needed. * */ static int journal_finish_inode_data_buffers(journal_t *journal, transaction_t *commit_transaction) { struct jbd2_inode *jinode, *next_i; int err, ret = 0; |
cd1aac329
|
235 |
/* For locking, see the comment in journal_submit_data_buffers() */ |
c851ed540
|
236 237 |
spin_lock(&journal->j_list_lock); list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) { |
39e3ac259
|
238 |
set_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); |
c851ed540
|
239 240 |
spin_unlock(&journal->j_list_lock); err = filemap_fdatawait(jinode->i_vfs_inode->i_mapping); |
e9e34f4e8
|
241 242 243 |
if (err) { /* * Because AS_EIO is cleared by |
94004ed72
|
244 |
* filemap_fdatawait_range(), set it again so |
e9e34f4e8
|
245 246 247 248 249 250 251 252 |
* that user process can get -EIO from fsync(). */ set_bit(AS_EIO, &jinode->i_vfs_inode->i_mapping->flags); if (!ret) ret = err; } |
c851ed540
|
253 |
spin_lock(&journal->j_list_lock); |
39e3ac259
|
254 255 |
clear_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); smp_mb__after_clear_bit(); |
c851ed540
|
256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 |
wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING); } /* Now refile inode to proper lists */ list_for_each_entry_safe(jinode, next_i, &commit_transaction->t_inode_list, i_list) { list_del(&jinode->i_list); if (jinode->i_next_transaction) { jinode->i_transaction = jinode->i_next_transaction; jinode->i_next_transaction = NULL; list_add(&jinode->i_list, &jinode->i_transaction->t_inode_list); } else { jinode->i_transaction = NULL; } } spin_unlock(&journal->j_list_lock); return ret; } |
818d276ce
|
276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 |
static __u32 jbd2_checksum_data(__u32 crc32_sum, struct buffer_head *bh) { struct page *page = bh->b_page; char *addr; __u32 checksum; addr = kmap_atomic(page, KM_USER0); checksum = crc32_be(crc32_sum, (void *)(addr + offset_in_page(bh->b_data)), bh->b_size); kunmap_atomic(addr, KM_USER0); return checksum; } static void write_tag_block(int tag_bytes, journal_block_tag_t *tag, |
18eba7aae
|
291 |
unsigned long long block) |
b517bea1c
|
292 293 |
{ tag->t_blocknr = cpu_to_be32(block & (u32)~0); |
cd02ff0b1
|
294 |
if (tag_bytes > JBD2_TAG_SIZE32) |
b517bea1c
|
295 296 |
tag->t_blocknr_high = cpu_to_be32((block >> 31) >> 1); } |
470decc61
|
297 |
/* |
f7f4bccb7
|
298 |
* jbd2_journal_commit_transaction |
470decc61
|
299 300 301 302 |
* * The primary function for committing a transaction to the log. This * function is called by the journal thread to begin a complete commit. */ |
f7f4bccb7
|
303 |
void jbd2_journal_commit_transaction(journal_t *journal) |
470decc61
|
304 |
{ |
8e85fb3f3
|
305 |
struct transaction_stats_s stats; |
470decc61
|
306 307 308 309 310 311 |
transaction_t *commit_transaction; struct journal_head *jh, *new_jh, *descriptor; struct buffer_head **wbuf = journal->j_wbuf; int bufs; int flags; int err; |
18eba7aae
|
312 |
unsigned long long blocknr; |
e07f7183a
|
313 314 |
ktime_t start_time; u64 commit_time; |
470decc61
|
315 316 317 318 319 320 |
char *tagp = NULL; journal_header_t *header; journal_block_tag_t *tag = NULL; int space_left = 0; int first_tag = 0; int tag_flag; |
fb68407b0
|
321 |
int i, to_free = 0; |
b517bea1c
|
322 |
int tag_bytes = journal_tag_bytes(journal); |
818d276ce
|
323 324 |
struct buffer_head *cbh = NULL; /* For transactional checksums */ __u32 crc32_sum = ~0; |
82f04ab47
|
325 |
struct blk_plug plug; |
470decc61
|
326 327 328 329 330 |
/* * First job: lock down the current transaction and wait for * all outstanding updates to complete. */ |
f7f4bccb7
|
331 332 |
/* Do we need to erase the effects of a prior jbd2_journal_flush? */ if (journal->j_flags & JBD2_FLUSHED) { |
470decc61
|
333 334 |
jbd_debug(3, "super block updated "); |
f7f4bccb7
|
335 |
jbd2_journal_update_superblock(journal, 1); |
470decc61
|
336 337 338 339 340 341 342 343 344 345 |
} else { jbd_debug(3, "superblock not updated "); } J_ASSERT(journal->j_running_transaction != NULL); J_ASSERT(journal->j_committing_transaction == NULL); commit_transaction = journal->j_running_transaction; J_ASSERT(commit_transaction->t_state == T_RUNNING); |
879c5e6b7
|
346 |
trace_jbd2_start_commit(journal, commit_transaction); |
f2a44523b
|
347 348 |
jbd_debug(1, "JBD2: starting commit of transaction %d ", |
470decc61
|
349 |
commit_transaction->t_tid); |
a931da6ac
|
350 |
write_lock(&journal->j_state_lock); |
470decc61
|
351 |
commit_transaction->t_state = T_LOCKED; |
879c5e6b7
|
352 |
trace_jbd2_commit_locking(journal, commit_transaction); |
bf6993276
|
353 354 355 356 |
stats.run.rs_wait = commit_transaction->t_max_wait; stats.run.rs_locked = jiffies; stats.run.rs_running = jbd2_time_diff(commit_transaction->t_start, stats.run.rs_locked); |
8e85fb3f3
|
357 |
|
470decc61
|
358 |
spin_lock(&commit_transaction->t_handle_lock); |
a51dca9cd
|
359 |
while (atomic_read(&commit_transaction->t_updates)) { |
470decc61
|
360 361 362 363 |
DEFINE_WAIT(wait); prepare_to_wait(&journal->j_wait_updates, &wait, TASK_UNINTERRUPTIBLE); |
a51dca9cd
|
364 |
if (atomic_read(&commit_transaction->t_updates)) { |
470decc61
|
365 |
spin_unlock(&commit_transaction->t_handle_lock); |
a931da6ac
|
366 |
write_unlock(&journal->j_state_lock); |
470decc61
|
367 |
schedule(); |
a931da6ac
|
368 |
write_lock(&journal->j_state_lock); |
470decc61
|
369 370 371 372 373 |
spin_lock(&commit_transaction->t_handle_lock); } finish_wait(&journal->j_wait_updates, &wait); } spin_unlock(&commit_transaction->t_handle_lock); |
a51dca9cd
|
374 |
J_ASSERT (atomic_read(&commit_transaction->t_outstanding_credits) <= |
470decc61
|
375 376 377 378 379 380 381 |
journal->j_max_transaction_buffers); /* * First thing we are allowed to do is to discard any remaining * BJ_Reserved buffers. Note, it is _not_ permissible to assume * that there are no such buffers: if a large filesystem * operation like a truncate needs to split itself over multiple |
f7f4bccb7
|
382 |
* transactions, then it may try to do a jbd2_journal_restart() while |
470decc61
|
383 384 385 386 387 388 389 |
* there are still BJ_Reserved buffers outstanding. These must * be released cleanly from the current transaction. * * In this case, the filesystem must still reserve write access * again before modifying the buffer in the new transaction, but * we do not require it to remember exactly which old buffers it * has reserved. This is consistent with the existing behaviour |
f7f4bccb7
|
390 |
* that multiple jbd2_journal_get_write_access() calls to the same |
25985edce
|
391 |
* buffer are perfectly permissible. |
470decc61
|
392 393 394 395 396 |
*/ while (commit_transaction->t_reserved_list) { jh = commit_transaction->t_reserved_list; JBUFFER_TRACE(jh, "reserved, unused: refile"); /* |
f7f4bccb7
|
397 |
* A jbd2_journal_get_undo_access()+jbd2_journal_release_buffer() may |
470decc61
|
398 399 400 401 402 403 |
* leave undo-committed data. */ if (jh->b_committed_data) { struct buffer_head *bh = jh2bh(jh); jbd_lock_bh_state(bh); |
af1e76d6b
|
404 |
jbd2_free(jh->b_committed_data, bh->b_size); |
470decc61
|
405 406 407 |
jh->b_committed_data = NULL; jbd_unlock_bh_state(bh); } |
f7f4bccb7
|
408 |
jbd2_journal_refile_buffer(journal, jh); |
470decc61
|
409 410 411 412 413 414 415 416 |
} /* * Now try to drop any written-back buffers from the journal's * checkpoint lists. We do this *before* commit because it potentially * frees some memory */ spin_lock(&journal->j_list_lock); |
f7f4bccb7
|
417 |
__jbd2_journal_clean_checkpoint_list(journal); |
470decc61
|
418 |
spin_unlock(&journal->j_list_lock); |
f2a44523b
|
419 420 |
jbd_debug(3, "JBD2: commit phase 1 "); |
470decc61
|
421 422 423 424 |
/* * Switch to a new revoke table. */ |
f7f4bccb7
|
425 |
jbd2_journal_switch_revoke_table(journal); |
470decc61
|
426 |
|
879c5e6b7
|
427 |
trace_jbd2_commit_flushing(journal, commit_transaction); |
bf6993276
|
428 429 430 |
stats.run.rs_flushing = jiffies; stats.run.rs_locked = jbd2_time_diff(stats.run.rs_locked, stats.run.rs_flushing); |
8e85fb3f3
|
431 |
|
470decc61
|
432 433 434 |
commit_transaction->t_state = T_FLUSH; journal->j_committing_transaction = commit_transaction; journal->j_running_transaction = NULL; |
e07f7183a
|
435 |
start_time = ktime_get(); |
470decc61
|
436 437 |
commit_transaction->t_log_start = journal->j_head; wake_up(&journal->j_wait_transaction_locked); |
a931da6ac
|
438 |
write_unlock(&journal->j_state_lock); |
470decc61
|
439 |
|
f2a44523b
|
440 441 |
jbd_debug(3, "JBD2: commit phase 2 "); |
470decc61
|
442 443 |
/* |
470decc61
|
444 445 446 |
* Now start flushing things to disk, in the order they appear * on the transaction lists. Data blocks go first. */ |
cd1aac329
|
447 |
err = journal_submit_data_buffers(journal, commit_transaction); |
c851ed540
|
448 449 |
if (err) jbd2_journal_abort(journal, err); |
470decc61
|
450 |
|
82f04ab47
|
451 |
blk_start_plug(&plug); |
67c457a8c
|
452 |
jbd2_journal_write_revoke_records(journal, commit_transaction, |
82f04ab47
|
453 454 |
WRITE_SYNC); blk_finish_plug(&plug); |
470decc61
|
455 |
|
f2a44523b
|
456 457 |
jbd_debug(3, "JBD2: commit phase 2 "); |
470decc61
|
458 459 |
/* |
470decc61
|
460 461 462 463 |
* Way to go: we have now written out all of the data for a * transaction! Now comes the tricky part: we need to write out * metadata. Loop over the transaction's entire buffer list: */ |
a931da6ac
|
464 |
write_lock(&journal->j_state_lock); |
470decc61
|
465 |
commit_transaction->t_state = T_COMMIT; |
a931da6ac
|
466 |
write_unlock(&journal->j_state_lock); |
470decc61
|
467 |
|
879c5e6b7
|
468 |
trace_jbd2_commit_logging(journal, commit_transaction); |
bf6993276
|
469 470 471 |
stats.run.rs_logging = jiffies; stats.run.rs_flushing = jbd2_time_diff(stats.run.rs_flushing, stats.run.rs_logging); |
a51dca9cd
|
472 473 |
stats.run.rs_blocks = atomic_read(&commit_transaction->t_outstanding_credits); |
bf6993276
|
474 |
stats.run.rs_blocks_logged = 0; |
8e85fb3f3
|
475 |
|
1dfc3220d
|
476 |
J_ASSERT(commit_transaction->t_nr_buffers <= |
a51dca9cd
|
477 |
atomic_read(&commit_transaction->t_outstanding_credits)); |
1dfc3220d
|
478 |
|
87c89c232
|
479 |
err = 0; |
470decc61
|
480 481 |
descriptor = NULL; bufs = 0; |
82f04ab47
|
482 |
blk_start_plug(&plug); |
470decc61
|
483 484 485 486 487 488 489 |
while (commit_transaction->t_buffers) { /* Find the next buffer to be journaled... */ jh = commit_transaction->t_buffers; /* If we're in abort mode, we just un-journal the buffer and |
7ad7445f6
|
490 |
release it. */ |
470decc61
|
491 492 |
if (is_journal_aborted(journal)) { |
7ad7445f6
|
493 |
clear_buffer_jbddirty(jh2bh(jh)); |
470decc61
|
494 |
JBUFFER_TRACE(jh, "journal is aborting: refile"); |
e06c8227f
|
495 496 497 498 |
jbd2_buffer_abort_trigger(jh, jh->b_frozen_data ? jh->b_frozen_triggers : jh->b_triggers); |
f7f4bccb7
|
499 |
jbd2_journal_refile_buffer(journal, jh); |
470decc61
|
500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 |
/* If that was the last one, we need to clean up * any descriptor buffers which may have been * already allocated, even if we are now * aborting. */ if (!commit_transaction->t_buffers) goto start_journal_io; continue; } /* Make sure we have a descriptor block in which to record the metadata buffer. */ if (!descriptor) { struct buffer_head *bh; J_ASSERT (bufs == 0); |
f2a44523b
|
516 517 |
jbd_debug(4, "JBD2: get descriptor "); |
470decc61
|
518 |
|
f7f4bccb7
|
519 |
descriptor = jbd2_journal_get_descriptor_buffer(journal); |
470decc61
|
520 |
if (!descriptor) { |
a7fa2baf8
|
521 |
jbd2_journal_abort(journal, -EIO); |
470decc61
|
522 523 524 525 |
continue; } bh = jh2bh(descriptor); |
f2a44523b
|
526 527 |
jbd_debug(4, "JBD2: got buffer %llu (%p) ", |
470decc61
|
528 529 |
(unsigned long long)bh->b_blocknr, bh->b_data); header = (journal_header_t *)&bh->b_data[0]; |
f7f4bccb7
|
530 531 |
header->h_magic = cpu_to_be32(JBD2_MAGIC_NUMBER); header->h_blocktype = cpu_to_be32(JBD2_DESCRIPTOR_BLOCK); |
470decc61
|
532 533 534 535 536 537 538 539 540 541 542 543 |
header->h_sequence = cpu_to_be32(commit_transaction->t_tid); tagp = &bh->b_data[sizeof(journal_header_t)]; space_left = bh->b_size - sizeof(journal_header_t); first_tag = 1; set_buffer_jwrite(bh); set_buffer_dirty(bh); wbuf[bufs++] = bh; /* Record it so that we can wait for IO completion later */ BUFFER_TRACE(bh, "ph3: file as descriptor"); |
f7f4bccb7
|
544 |
jbd2_journal_file_buffer(descriptor, commit_transaction, |
470decc61
|
545 546 547 548 |
BJ_LogCtl); } /* Where is the buffer to be written? */ |
f7f4bccb7
|
549 |
err = jbd2_journal_next_log_block(journal, &blocknr); |
470decc61
|
550 551 552 553 |
/* If the block mapping failed, just abandon the buffer and repeat this loop: we'll fall into the refile-on-abort condition above. */ if (err) { |
a7fa2baf8
|
554 |
jbd2_journal_abort(journal, err); |
470decc61
|
555 556 557 558 559 560 |
continue; } /* * start_this_handle() uses t_outstanding_credits to determine * the free space in the log, but this counter is changed |
f7f4bccb7
|
561 |
* by jbd2_journal_next_log_block() also. |
470decc61
|
562 |
*/ |
a51dca9cd
|
563 |
atomic_dec(&commit_transaction->t_outstanding_credits); |
470decc61
|
564 565 566 567 568 569 570 571 572 573 574 575 |
/* Bump b_count to prevent truncate from stumbling over the shadowed buffer! @@@ This can go if we ever get rid of the BJ_IO/BJ_Shadow pairing of buffers. */ atomic_inc(&jh2bh(jh)->b_count); /* Make a temporary IO buffer with which to write it out (this will requeue both the metadata buffer and the temporary IO buffer). new_bh goes on BJ_IO*/ set_bit(BH_JWrite, &jh2bh(jh)->b_state); /* |
f7f4bccb7
|
576 |
* akpm: jbd2_journal_write_metadata_buffer() sets |
470decc61
|
577 578 579 580 581 |
* new_bh->b_transaction to commit_transaction. * We need to clean this up before we release new_bh * (which is of type BJ_IO) */ JBUFFER_TRACE(jh, "ph3: write metadata"); |
f7f4bccb7
|
582 |
flags = jbd2_journal_write_metadata_buffer(commit_transaction, |
470decc61
|
583 |
jh, &new_jh, blocknr); |
e6ec116b6
|
584 585 586 587 |
if (flags < 0) { jbd2_journal_abort(journal, flags); continue; } |
470decc61
|
588 589 590 591 592 593 594 595 |
set_bit(BH_JWrite, &jh2bh(new_jh)->b_state); wbuf[bufs++] = jh2bh(new_jh); /* Record the new block's tag in the current descriptor buffer */ tag_flag = 0; if (flags & 1) |
f7f4bccb7
|
596 |
tag_flag |= JBD2_FLAG_ESCAPE; |
470decc61
|
597 |
if (!first_tag) |
f7f4bccb7
|
598 |
tag_flag |= JBD2_FLAG_SAME_UUID; |
470decc61
|
599 600 |
tag = (journal_block_tag_t *) tagp; |
b517bea1c
|
601 |
write_tag_block(tag_bytes, tag, jh2bh(jh)->b_blocknr); |
470decc61
|
602 |
tag->t_flags = cpu_to_be32(tag_flag); |
b517bea1c
|
603 604 |
tagp += tag_bytes; space_left -= tag_bytes; |
470decc61
|
605 606 607 608 609 610 611 612 613 614 615 616 617 |
if (first_tag) { memcpy (tagp, journal->j_uuid, 16); tagp += 16; space_left -= 16; first_tag = 0; } /* If there's no more to do, or if the descriptor is full, let the IO rip! */ if (bufs == journal->j_wbufsize || commit_transaction->t_buffers == NULL || |
b517bea1c
|
618 |
space_left < tag_bytes + 16) { |
470decc61
|
619 |
|
f2a44523b
|
620 621 |
jbd_debug(4, "JBD2: Submit %d IOs ", bufs); |
470decc61
|
622 623 624 625 |
/* Write an end-of-descriptor marker before submitting the IOs. "tag" still points to the last tag we set up. */ |
f7f4bccb7
|
626 |
tag->t_flags |= cpu_to_be32(JBD2_FLAG_LAST_TAG); |
470decc61
|
627 628 629 630 |
start_journal_io: for (i = 0; i < bufs; i++) { struct buffer_head *bh = wbuf[i]; |
818d276ce
|
631 632 633 634 635 636 637 638 |
/* * Compute checksum. */ if (JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM)) { crc32_sum = jbd2_checksum_data(crc32_sum, bh); } |
470decc61
|
639 640 641 642 |
lock_buffer(bh); clear_buffer_dirty(bh); set_buffer_uptodate(bh); bh->b_end_io = journal_end_buffer_io_sync; |
82f04ab47
|
643 |
submit_bh(WRITE_SYNC, bh); |
470decc61
|
644 645 |
} cond_resched(); |
bf6993276
|
646 |
stats.run.rs_blocks_logged += bufs; |
470decc61
|
647 648 649 650 651 652 653 |
/* Force a new descriptor to be generated next time round the loop. */ descriptor = NULL; bufs = 0; } } |
f73bee498
|
654 655 656 657 658 659 660 661 662 663 |
err = journal_finish_inode_data_buffers(journal, commit_transaction); if (err) { printk(KERN_WARNING "JBD2: Detected IO errors while flushing file data " "on %s ", journal->j_devname); if (journal->j_flags & JBD2_ABORT_ON_SYNCDATA_ERR) jbd2_journal_abort(journal, err); err = 0; } |
bbd2be369
|
664 665 666 667 |
write_lock(&journal->j_state_lock); J_ASSERT(commit_transaction->t_state == T_COMMIT); commit_transaction->t_state = T_COMMIT_DFLUSH; write_unlock(&journal->j_state_lock); |
cc3e1bea5
|
668 669 670 671 672 |
/* * If the journal is not located on the file system device, * then we must flush the file system device before we issue * the commit record */ |
81be12c81
|
673 |
if (commit_transaction->t_need_data_flush && |
cc3e1bea5
|
674 675 |
(journal->j_fs_dev != journal->j_dev) && (journal->j_flags & JBD2_BARRIER)) |
dd3932edd
|
676 |
blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL); |
818d276ce
|
677 |
|
cc3e1bea5
|
678 |
/* Done it all: now write the commit record asynchronously. */ |
818d276ce
|
679 |
if (JBD2_HAS_INCOMPAT_FEATURE(journal, |
0e3d2a631
|
680 |
JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) { |
818d276ce
|
681 682 683 684 |
err = journal_submit_commit_record(journal, commit_transaction, &cbh, crc32_sum); if (err) __jbd2_journal_abort_hard(journal); |
e9e34f4e8
|
685 |
} |
c851ed540
|
686 |
|
82f04ab47
|
687 |
blk_finish_plug(&plug); |
470decc61
|
688 689 690 691 692 693 694 695 696 697 |
/* Lo and behold: we have just managed to send a transaction to the log. Before we can commit it, wait for the IO so far to complete. Control buffers being written are on the transaction's t_log_list queue, and metadata buffers are on the t_iobuf_list queue. Wait for the buffers in reverse order. That way we are less likely to be woken up until all IOs have completed, and so we incur less scheduling load. */ |
f2a44523b
|
698 699 |
jbd_debug(3, "JBD2: commit phase 3 "); |
470decc61
|
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 |
/* * akpm: these are BJ_IO, and j_list_lock is not needed. * See __journal_try_to_free_buffer. */ wait_for_iobuf: while (commit_transaction->t_iobuf_list != NULL) { struct buffer_head *bh; jh = commit_transaction->t_iobuf_list->b_tprev; bh = jh2bh(jh); if (buffer_locked(bh)) { wait_on_buffer(bh); goto wait_for_iobuf; } if (cond_resched()) goto wait_for_iobuf; if (unlikely(!buffer_uptodate(bh))) err = -EIO; clear_buffer_jwrite(bh); JBUFFER_TRACE(jh, "ph4: unfile after journal write"); |
f7f4bccb7
|
724 |
jbd2_journal_unfile_buffer(journal, jh); |
470decc61
|
725 726 727 |
/* * ->t_iobuf_list should contain only dummy buffer_heads |
f7f4bccb7
|
728 |
* which were created by jbd2_journal_write_metadata_buffer(). |
470decc61
|
729 730 |
*/ BUFFER_TRACE(bh, "dumping temporary bh"); |
f7f4bccb7
|
731 |
jbd2_journal_put_journal_head(jh); |
470decc61
|
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 |
__brelse(bh); J_ASSERT_BH(bh, atomic_read(&bh->b_count) == 0); free_buffer_head(bh); /* We also have to unlock and free the corresponding shadowed buffer */ jh = commit_transaction->t_shadow_list->b_tprev; bh = jh2bh(jh); clear_bit(BH_JWrite, &bh->b_state); J_ASSERT_BH(bh, buffer_jbddirty(bh)); /* The metadata is now released for reuse, but we need to remember it against this transaction so that when we finally commit, we can do any checkpointing required. */ JBUFFER_TRACE(jh, "file as BJ_Forget"); |
f7f4bccb7
|
748 |
jbd2_journal_file_buffer(jh, commit_transaction, BJ_Forget); |
229309cae
|
749 750 751 752 753 754 755 |
/* * Wake up any transactions which were waiting for this IO to * complete. The barrier must be here so that changes by * jbd2_journal_file_buffer() take effect before wake_up_bit() * does the waitqueue check. */ smp_mb(); |
470decc61
|
756 757 758 759 760 761 |
wake_up_bit(&bh->b_state, BH_Unshadow); JBUFFER_TRACE(jh, "brelse shadowed buffer"); __brelse(bh); } J_ASSERT (commit_transaction->t_shadow_list == NULL); |
f2a44523b
|
762 763 |
jbd_debug(3, "JBD2: commit phase 4 "); |
470decc61
|
764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 |
/* Here we wait for the revoke record and descriptor record buffers */ wait_for_ctlbuf: while (commit_transaction->t_log_list != NULL) { struct buffer_head *bh; jh = commit_transaction->t_log_list->b_tprev; bh = jh2bh(jh); if (buffer_locked(bh)) { wait_on_buffer(bh); goto wait_for_ctlbuf; } if (cond_resched()) goto wait_for_ctlbuf; if (unlikely(!buffer_uptodate(bh))) err = -EIO; BUFFER_TRACE(bh, "ph5: control buffer writeout done: unfile"); clear_buffer_jwrite(bh); |
f7f4bccb7
|
784 785 |
jbd2_journal_unfile_buffer(journal, jh); jbd2_journal_put_journal_head(jh); |
470decc61
|
786 787 788 |
__brelse(bh); /* One for getblk */ /* AKPM: bforget here */ } |
77e841de8
|
789 790 |
if (err) jbd2_journal_abort(journal, err); |
f2a44523b
|
791 792 |
jbd_debug(3, "JBD2: commit phase 5 "); |
bbd2be369
|
793 794 795 796 |
write_lock(&journal->j_state_lock); J_ASSERT(commit_transaction->t_state == T_COMMIT_DFLUSH); commit_transaction->t_state = T_COMMIT_JFLUSH; write_unlock(&journal->j_state_lock); |
470decc61
|
797 |
|
818d276ce
|
798 |
if (!JBD2_HAS_INCOMPAT_FEATURE(journal, |
0e3d2a631
|
799 |
JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) { |
818d276ce
|
800 801 802 803 804 |
err = journal_submit_commit_record(journal, commit_transaction, &cbh, crc32_sum); if (err) __jbd2_journal_abort_hard(journal); } |
6cba611e6
|
805 |
if (cbh) |
fd98496f4
|
806 |
err = journal_wait_on_commit_record(journal, cbh); |
f73bee498
|
807 808 809 |
if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT) && journal->j_flags & JBD2_BARRIER) { |
dd3932edd
|
810 |
blkdev_issue_flush(journal->j_dev, GFP_KERNEL, NULL); |
f73bee498
|
811 |
} |
470decc61
|
812 813 |
if (err) |
a7fa2baf8
|
814 |
jbd2_journal_abort(journal, err); |
470decc61
|
815 816 817 818 819 |
/* End of a transaction! Finally, we can do checkpoint processing: any buffers committed as a result of this transaction can be removed from any checkpoint list it was on before. */ |
f2a44523b
|
820 821 |
jbd_debug(3, "JBD2: commit phase 6 "); |
470decc61
|
822 |
|
c851ed540
|
823 |
J_ASSERT(list_empty(&commit_transaction->t_inode_list)); |
470decc61
|
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 |
J_ASSERT(commit_transaction->t_buffers == NULL); J_ASSERT(commit_transaction->t_checkpoint_list == NULL); J_ASSERT(commit_transaction->t_iobuf_list == NULL); J_ASSERT(commit_transaction->t_shadow_list == NULL); J_ASSERT(commit_transaction->t_log_list == NULL); restart_loop: /* * As there are other places (journal_unmap_buffer()) adding buffers * to this list we have to be careful and hold the j_list_lock. */ spin_lock(&journal->j_list_lock); while (commit_transaction->t_forget) { transaction_t *cp_transaction; struct buffer_head *bh; |
de1b79413
|
839 |
int try_to_free = 0; |
470decc61
|
840 841 842 843 |
jh = commit_transaction->t_forget; spin_unlock(&journal->j_list_lock); bh = jh2bh(jh); |
de1b79413
|
844 845 846 847 848 |
/* * Get a reference so that bh cannot be freed before we are * done with it. */ get_bh(bh); |
470decc61
|
849 |
jbd_lock_bh_state(bh); |
23e2af351
|
850 |
J_ASSERT_JH(jh, jh->b_transaction == commit_transaction); |
470decc61
|
851 852 853 854 855 856 857 858 859 860 |
/* * If there is undo-protected committed data against * this buffer, then we can remove it now. If it is a * buffer needing such protection, the old frozen_data * field now points to a committed version of the * buffer, so rotate that field to the new committed * data. * * Otherwise, we can just throw away the frozen data now. |
e06c8227f
|
861 862 863 |
* * We also know that the frozen data has already fired * its triggers if they exist, so we can clear that too. |
470decc61
|
864 865 |
*/ if (jh->b_committed_data) { |
af1e76d6b
|
866 |
jbd2_free(jh->b_committed_data, bh->b_size); |
470decc61
|
867 868 869 870 |
jh->b_committed_data = NULL; if (jh->b_frozen_data) { jh->b_committed_data = jh->b_frozen_data; jh->b_frozen_data = NULL; |
e06c8227f
|
871 |
jh->b_frozen_triggers = NULL; |
470decc61
|
872 873 |
} } else if (jh->b_frozen_data) { |
af1e76d6b
|
874 |
jbd2_free(jh->b_frozen_data, bh->b_size); |
470decc61
|
875 |
jh->b_frozen_data = NULL; |
e06c8227f
|
876 |
jh->b_frozen_triggers = NULL; |
470decc61
|
877 878 879 880 881 882 |
} spin_lock(&journal->j_list_lock); cp_transaction = jh->b_cp_transaction; if (cp_transaction) { JBUFFER_TRACE(jh, "remove from old cp transaction"); |
8e85fb3f3
|
883 |
cp_transaction->t_chp_stats.cs_dropped++; |
f7f4bccb7
|
884 |
__jbd2_journal_remove_checkpoint(jh); |
470decc61
|
885 886 887 888 |
} /* Only re-checkpoint the buffer_head if it is marked * dirty. If the buffer was added to the BJ_Forget list |
f7f4bccb7
|
889 |
* by jbd2_journal_forget, it may no longer be dirty and |
470decc61
|
890 891 892 893 894 895 |
* there's no point in keeping a checkpoint record for * it. */ /* A buffer which has been freed while still being * journaled by a previous transaction may end up still * being dirty here, but we want to avoid writing back |
ba869023e
|
896 897 898 899 |
* that buffer in the future after the "add to orphan" * operation been committed, That's not only a performance * gain, it also stops aliasing problems if the buffer is * left behind for writeback and gets reallocated for another |
470decc61
|
900 |
* use in a different page. */ |
ba869023e
|
901 |
if (buffer_freed(bh) && !jh->b_next_transaction) { |
470decc61
|
902 903 904 905 906 907 |
clear_buffer_freed(bh); clear_buffer_jbddirty(bh); } if (buffer_jbddirty(bh)) { JBUFFER_TRACE(jh, "add to new checkpointing trans"); |
f7f4bccb7
|
908 |
__jbd2_journal_insert_checkpoint(jh, commit_transaction); |
7ad7445f6
|
909 910 |
if (is_journal_aborted(journal)) clear_buffer_jbddirty(bh); |
470decc61
|
911 912 |
} else { J_ASSERT_BH(bh, !buffer_dirty(bh)); |
de1b79413
|
913 914 |
/* * The buffer on BJ_Forget list and not jbddirty means |
470decc61
|
915 916 917 918 919 |
* it has been freed by this transaction and hence it * could not have been reallocated until this * transaction has committed. *BUT* it could be * reallocated once we have written all the data to * disk and before we process the buffer on BJ_Forget |
de1b79413
|
920 921 922 923 |
* list. */ if (!jh->b_next_transaction) try_to_free = 1; |
470decc61
|
924 |
} |
de1b79413
|
925 926 927 928 929 930 931 |
JBUFFER_TRACE(jh, "refile or unfile buffer"); __jbd2_journal_refile_buffer(jh); jbd_unlock_bh_state(bh); if (try_to_free) release_buffer_page(bh); /* Drops bh reference */ else __brelse(bh); |
470decc61
|
932 933 934 935 |
cond_resched_lock(&journal->j_list_lock); } spin_unlock(&journal->j_list_lock); /* |
f5a7a6b0d
|
936 937 938 939 |
* This is a bit sleazy. We use j_list_lock to protect transition * of a transaction into T_FINISHED state and calling * __jbd2_journal_drop_transaction(). Otherwise we could race with * other checkpointing code processing the transaction... |
470decc61
|
940 |
*/ |
a931da6ac
|
941 |
write_lock(&journal->j_state_lock); |
470decc61
|
942 943 944 945 946 947 948 |
spin_lock(&journal->j_list_lock); /* * Now recheck if some buffers did not get attached to the transaction * while the lock was dropped... */ if (commit_transaction->t_forget) { spin_unlock(&journal->j_list_lock); |
a931da6ac
|
949 |
write_unlock(&journal->j_state_lock); |
470decc61
|
950 951 952 953 |
goto restart_loop; } /* Done with this transaction! */ |
f2a44523b
|
954 955 |
jbd_debug(3, "JBD2: commit phase 7 "); |
470decc61
|
956 |
|
bbd2be369
|
957 |
J_ASSERT(commit_transaction->t_state == T_COMMIT_JFLUSH); |
470decc61
|
958 |
|
8e85fb3f3
|
959 |
commit_transaction->t_start = jiffies; |
bf6993276
|
960 961 |
stats.run.rs_logging = jbd2_time_diff(stats.run.rs_logging, commit_transaction->t_start); |
8e85fb3f3
|
962 963 |
/* |
bf6993276
|
964 |
* File the transaction statistics |
8e85fb3f3
|
965 |
*/ |
8e85fb3f3
|
966 |
stats.ts_tid = commit_transaction->t_tid; |
8dd420466
|
967 968 |
stats.run.rs_handle_count = atomic_read(&commit_transaction->t_handle_count); |
bf6993276
|
969 970 |
trace_jbd2_run_stats(journal->j_fs_dev->bd_dev, commit_transaction->t_tid, &stats.run); |
8e85fb3f3
|
971 972 973 974 |
/* * Calculate overall stats */ |
bf6993276
|
975 |
spin_lock(&journal->j_history_lock); |
8e85fb3f3
|
976 |
journal->j_stats.ts_tid++; |
bf6993276
|
977 978 979 980 981 982 983 984 |
journal->j_stats.run.rs_wait += stats.run.rs_wait; journal->j_stats.run.rs_running += stats.run.rs_running; journal->j_stats.run.rs_locked += stats.run.rs_locked; journal->j_stats.run.rs_flushing += stats.run.rs_flushing; journal->j_stats.run.rs_logging += stats.run.rs_logging; journal->j_stats.run.rs_handle_count += stats.run.rs_handle_count; journal->j_stats.run.rs_blocks += stats.run.rs_blocks; journal->j_stats.run.rs_blocks_logged += stats.run.rs_blocks_logged; |
8e85fb3f3
|
985 |
spin_unlock(&journal->j_history_lock); |
470decc61
|
986 987 988 989 |
commit_transaction->t_state = T_FINISHED; J_ASSERT(commit_transaction == journal->j_committing_transaction); journal->j_commit_sequence = commit_transaction->t_tid; journal->j_committing_transaction = NULL; |
e07f7183a
|
990 |
commit_time = ktime_to_ns(ktime_sub(ktime_get(), start_time)); |
470decc61
|
991 |
|
e07f7183a
|
992 993 994 995 996 997 998 999 1000 |
/* * weight the commit time higher than the average time so we don't * react too strongly to vast changes in the commit time */ if (likely(journal->j_average_commit_time)) journal->j_average_commit_time = (commit_time + journal->j_average_commit_time*3) / 4; else journal->j_average_commit_time = commit_time; |
a931da6ac
|
1001 |
write_unlock(&journal->j_state_lock); |
6c20ec850
|
1002 |
|
f89b77950
|
1003 1004 |
if (commit_transaction->t_checkpoint_list == NULL && commit_transaction->t_checkpoint_io_list == NULL) { |
f7f4bccb7
|
1005 |
__jbd2_journal_drop_transaction(journal, commit_transaction); |
fb68407b0
|
1006 |
to_free = 1; |
470decc61
|
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 |
} else { if (journal->j_checkpoint_transactions == NULL) { journal->j_checkpoint_transactions = commit_transaction; commit_transaction->t_cpnext = commit_transaction; commit_transaction->t_cpprev = commit_transaction; } else { commit_transaction->t_cpnext = journal->j_checkpoint_transactions; commit_transaction->t_cpprev = commit_transaction->t_cpnext->t_cpprev; commit_transaction->t_cpnext->t_cpprev = commit_transaction; commit_transaction->t_cpprev->t_cpnext = commit_transaction; } } spin_unlock(&journal->j_list_lock); |
fb68407b0
|
1024 1025 |
if (journal->j_commit_callback) journal->j_commit_callback(journal, commit_transaction); |
879c5e6b7
|
1026 |
trace_jbd2_end_commit(journal, commit_transaction); |
f2a44523b
|
1027 1028 |
jbd_debug(1, "JBD2: commit %d complete, head %d ", |
470decc61
|
1029 |
journal->j_commit_sequence, journal->j_tail_sequence); |
fb68407b0
|
1030 1031 |
if (to_free) kfree(commit_transaction); |
470decc61
|
1032 1033 1034 |
wake_up(&journal->j_wait_done_commit); } |