aboutsummaryrefslogtreecommitdiffstats
path: root/fs/jbd2/commit.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/jbd2/commit.c')
-rw-r--r--fs/jbd2/commit.c211
1 files changed, 128 insertions, 83 deletions
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 27373f5792a4..885a7a6cc53e 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -62,6 +62,7 @@ static void journal_end_buffer_io_sync(struct buffer_head *bh, int uptodate)
*/
static void release_buffer_page(struct buffer_head *bh)
{
+ struct folio *folio;
struct page *page;
if (buffer_dirty(bh))
@@ -71,18 +72,19 @@ static void release_buffer_page(struct buffer_head *bh)
page = bh->b_page;
if (!page)
goto nope;
- if (page->mapping)
+ folio = page_folio(page);
+ if (folio->mapping)
goto nope;
/* OK, it's a truncated page */
- if (!trylock_page(page))
+ if (!folio_trylock(folio))
goto nope;
- get_page(page);
+ folio_get(folio);
__brelse(bh);
- try_to_free_buffers(page);
- unlock_page(page);
- put_page(page);
+ try_to_free_buffers(folio);
+ folio_unlock(folio);
+ folio_put(folio);
return;
nope:
@@ -120,8 +122,8 @@ static int journal_submit_commit_record(journal_t *journal,
{
struct commit_header *tmp;
struct buffer_head *bh;
- int ret;
struct timespec64 now;
+ blk_opf_t write_flags = REQ_OP_WRITE | REQ_SYNC;
*cbh = NULL;
@@ -153,13 +155,11 @@ static int journal_submit_commit_record(journal_t *journal,
if (journal->j_flags & JBD2_BARRIER &&
!jbd2_has_feature_async_commit(journal))
- ret = submit_bh(REQ_OP_WRITE,
- REQ_SYNC | REQ_PREFLUSH | REQ_FUA, bh);
- else
- ret = submit_bh(REQ_OP_WRITE, REQ_SYNC, bh);
+ write_flags |= REQ_PREFLUSH | REQ_FUA;
+ submit_bh(write_flags, bh);
*cbh = bh;
- return ret;
+ return 0;
}
/*
@@ -187,20 +187,48 @@ static int journal_wait_on_commit_record(journal_t *journal,
* use writepages() because with delayed allocation we may be doing
* block allocation in writepages().
*/
-static int journal_submit_inode_data_buffers(struct address_space *mapping,
- loff_t dirty_start, loff_t dirty_end)
+int jbd2_journal_submit_inode_data_buffers(struct jbd2_inode *jinode)
{
- int ret;
+ struct address_space *mapping = jinode->i_vfs_inode->i_mapping;
struct writeback_control wbc = {
.sync_mode = WB_SYNC_ALL,
.nr_to_write = mapping->nrpages * 2,
- .range_start = dirty_start,
- .range_end = dirty_end,
+ .range_start = jinode->i_dirty_start,
+ .range_end = jinode->i_dirty_end,
};
- ret = generic_writepages(mapping, &wbc);
- return ret;
+ /*
+ * 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.
+ */
+ return generic_writepages(mapping, &wbc);
+}
+
+/* Send all the data buffers related to an inode */
+int jbd2_submit_inode_data(struct jbd2_inode *jinode)
+{
+
+ if (!jinode || !(jinode->i_flags & JI_WRITE_DATA))
+ return 0;
+
+ trace_jbd2_submit_inode_data(jinode->i_vfs_inode);
+ return jbd2_journal_submit_inode_data_buffers(jinode);
+
+}
+EXPORT_SYMBOL(jbd2_submit_inode_data);
+
+int jbd2_wait_inode_data(journal_t *journal, struct jbd2_inode *jinode)
+{
+ if (!jinode || !(jinode->i_flags & JI_WAIT_DATA) ||
+ !jinode->i_vfs_inode || !jinode->i_vfs_inode->i_mapping)
+ return 0;
+ return filemap_fdatawait_range_keep_errors(
+ jinode->i_vfs_inode->i_mapping, jinode->i_dirty_start,
+ jinode->i_dirty_end);
}
+EXPORT_SYMBOL(jbd2_wait_inode_data);
/*
* Submit all the data buffers of inode associated with the transaction to
@@ -215,29 +243,20 @@ static int journal_submit_data_buffers(journal_t *journal,
{
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) {
- loff_t dirty_start = jinode->i_dirty_start;
- loff_t dirty_end = jinode->i_dirty_end;
-
if (!(jinode->i_flags & JI_WRITE_DATA))
continue;
- mapping = jinode->i_vfs_inode->i_mapping;
jinode->i_flags |= JI_COMMIT_RUNNING;
spin_unlock(&journal->j_list_lock);
- /*
- * 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.
- */
+ /* submit the inode data buffers. */
trace_jbd2_submit_inode_data(jinode->i_vfs_inode);
- err = journal_submit_inode_data_buffers(mapping, dirty_start,
- dirty_end);
- if (!ret)
- ret = err;
+ if (journal->j_submit_inode_data_buffers) {
+ err = journal->j_submit_inode_data_buffers(jinode);
+ if (!ret)
+ ret = err;
+ }
spin_lock(&journal->j_list_lock);
J_ASSERT(jinode->i_transaction == commit_transaction);
jinode->i_flags &= ~JI_COMMIT_RUNNING;
@@ -248,6 +267,15 @@ static int journal_submit_data_buffers(journal_t *journal,
return ret;
}
+int jbd2_journal_finish_inode_data_buffers(struct jbd2_inode *jinode)
+{
+ struct address_space *mapping = jinode->i_vfs_inode->i_mapping;
+
+ return filemap_fdatawait_range_keep_errors(mapping,
+ jinode->i_dirty_start,
+ jinode->i_dirty_end);
+}
+
/*
* Wait for data submitted for writeout, refile inodes to proper
* transaction if needed.
@@ -262,18 +290,16 @@ static int journal_finish_inode_data_buffers(journal_t *journal,
/* For locking, see the comment in journal_submit_data_buffers() */
spin_lock(&journal->j_list_lock);
list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) {
- loff_t dirty_start = jinode->i_dirty_start;
- loff_t dirty_end = jinode->i_dirty_end;
-
if (!(jinode->i_flags & JI_WAIT_DATA))
continue;
jinode->i_flags |= JI_COMMIT_RUNNING;
spin_unlock(&journal->j_list_lock);
- err = filemap_fdatawait_range_keep_errors(
- jinode->i_vfs_inode->i_mapping, dirty_start,
- dirty_end);
- if (!ret)
- ret = err;
+ /* wait for the inode data buffers writeout. */
+ if (journal->j_finish_inode_data_buffers) {
+ err = journal->j_finish_inode_data_buffers(jinode);
+ if (!ret)
+ ret = err;
+ }
spin_lock(&journal->j_list_lock);
jinode->i_flags &= ~JI_COMMIT_RUNNING;
smp_mb();
@@ -393,7 +419,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
/* Do we need to erase the effects of a prior jbd2_journal_flush? */
if (journal->j_flags & JBD2_FLUSHED) {
- jbd_debug(3, "super block updated\n");
+ jbd2_debug(3, "super block updated\n");
mutex_lock_io(&journal->j_checkpoint_mutex);
/*
* We hold j_checkpoint_mutex so tail cannot change under us.
@@ -407,19 +433,43 @@ void jbd2_journal_commit_transaction(journal_t *journal)
REQ_SYNC);
mutex_unlock(&journal->j_checkpoint_mutex);
} else {
- jbd_debug(3, "superblock not updated\n");
+ jbd2_debug(3, "superblock not updated\n");
}
J_ASSERT(journal->j_running_transaction != NULL);
J_ASSERT(journal->j_committing_transaction == NULL);
+ write_lock(&journal->j_state_lock);
+ journal->j_flags |= JBD2_FULL_COMMIT_ONGOING;
+ while (journal->j_flags & JBD2_FAST_COMMIT_ONGOING) {
+ DEFINE_WAIT(wait);
+
+ prepare_to_wait(&journal->j_fc_wait, &wait,
+ TASK_UNINTERRUPTIBLE);
+ write_unlock(&journal->j_state_lock);
+ schedule();
+ write_lock(&journal->j_state_lock);
+ finish_wait(&journal->j_fc_wait, &wait);
+ /*
+ * TODO: by blocking fast commits here, we are increasing
+ * fsync() latency slightly. Strictly speaking, we don't need
+ * to block fast commits until the transaction enters T_FLUSH
+ * state. So an optimization is possible where we block new fast
+ * commits here and wait for existing ones to complete
+ * just before we enter T_FLUSH. That way, the existing fast
+ * commits and this full commit can proceed parallely.
+ */
+ }
+ write_unlock(&journal->j_state_lock);
+
commit_transaction = journal->j_running_transaction;
trace_jbd2_start_commit(journal, commit_transaction);
- jbd_debug(1, "JBD2: starting commit of transaction %d\n",
+ jbd2_debug(1, "JBD2: starting commit of transaction %d\n",
commit_transaction->t_tid);
write_lock(&journal->j_state_lock);
+ journal->j_fc_off = 0;
J_ASSERT(commit_transaction->t_state == T_RUNNING);
commit_transaction->t_state = T_LOCKED;
@@ -434,24 +484,10 @@ void jbd2_journal_commit_transaction(journal_t *journal)
stats.run.rs_running = jbd2_time_diff(commit_transaction->t_start,
stats.run.rs_locked);
- spin_lock(&commit_transaction->t_handle_lock);
- while (atomic_read(&commit_transaction->t_updates)) {
- DEFINE_WAIT(wait);
+ // waits for any t_updates to finish
+ jbd2_journal_wait_updates(journal);
- prepare_to_wait(&journal->j_wait_updates, &wait,
- TASK_UNINTERRUPTIBLE);
- if (atomic_read(&commit_transaction->t_updates)) {
- spin_unlock(&commit_transaction->t_handle_lock);
- write_unlock(&journal->j_state_lock);
- schedule();
- write_lock(&journal->j_state_lock);
- spin_lock(&commit_transaction->t_handle_lock);
- }
- finish_wait(&journal->j_wait_updates, &wait);
- }
- spin_unlock(&commit_transaction->t_handle_lock);
commit_transaction->t_state = T_SWITCH;
- write_unlock(&journal->j_state_lock);
J_ASSERT (atomic_read(&commit_transaction->t_outstanding_credits) <=
journal->j_max_transaction_buffers);
@@ -471,6 +507,8 @@ void jbd2_journal_commit_transaction(journal_t *journal)
* has reserved. This is consistent with the existing behaviour
* that multiple jbd2_journal_get_write_access() calls to the same
* buffer are perfectly permissible.
+ * We use journal->j_state_lock here to serialize processing of
+ * t_reserved_list with eviction of buffers from journal_unmap_buffer().
*/
while (commit_transaction->t_reserved_list) {
jh = commit_transaction->t_reserved_list;
@@ -490,6 +528,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
jbd2_journal_refile_buffer(journal, jh);
}
+ write_unlock(&journal->j_state_lock);
/*
* Now try to drop any written-back buffers from the journal's
* checkpoint lists. We do this *before* commit because it potentially
@@ -499,7 +538,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
__jbd2_journal_clean_checkpoint_list(journal, false);
spin_unlock(&journal->j_list_lock);
- jbd_debug(3, "JBD2: commit phase 1\n");
+ jbd2_debug(3, "JBD2: commit phase 1\n");
/*
* Clear revoked flag to reflect there is no revoked buffers
@@ -512,13 +551,13 @@ void jbd2_journal_commit_transaction(journal_t *journal)
*/
jbd2_journal_switch_revoke_table(journal);
+ write_lock(&journal->j_state_lock);
/*
* Reserved credits cannot be claimed anymore, free them
*/
atomic_sub(atomic_read(&journal->j_reserved_credits),
&commit_transaction->t_outstanding_credits);
- write_lock(&journal->j_state_lock);
trace_jbd2_commit_flushing(journal, commit_transaction);
stats.run.rs_flushing = jiffies;
stats.run.rs_locked = jbd2_time_diff(stats.run.rs_locked,
@@ -529,10 +568,10 @@ void jbd2_journal_commit_transaction(journal_t *journal)
journal->j_running_transaction = NULL;
start_time = ktime_get();
commit_transaction->t_log_start = journal->j_head;
- wake_up(&journal->j_wait_transaction_locked);
+ wake_up_all(&journal->j_wait_transaction_locked);
write_unlock(&journal->j_state_lock);
- jbd_debug(3, "JBD2: commit phase 2a\n");
+ jbd2_debug(3, "JBD2: commit phase 2a\n");
/*
* Now start flushing things to disk, in the order they appear
@@ -545,7 +584,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
blk_start_plug(&plug);
jbd2_journal_write_revoke_records(commit_transaction, &log_bufs);
- jbd_debug(3, "JBD2: commit phase 2b\n");
+ jbd2_debug(3, "JBD2: commit phase 2b\n");
/*
* Way to go: we have now written out all of the data for a
@@ -601,7 +640,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
if (!descriptor) {
J_ASSERT (bufs == 0);
- jbd_debug(4, "JBD2: get descriptor\n");
+ jbd2_debug(4, "JBD2: get descriptor\n");
descriptor = jbd2_journal_get_descriptor_buffer(
commit_transaction,
@@ -611,7 +650,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
continue;
}
- jbd_debug(4, "JBD2: got buffer %llu (%p)\n",
+ jbd2_debug(4, "JBD2: got buffer %llu (%p)\n",
(unsigned long long)descriptor->b_blocknr,
descriptor->b_data);
tagp = &descriptor->b_data[sizeof(journal_header_t)];
@@ -696,7 +735,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
commit_transaction->t_buffers == NULL ||
space_left < tag_bytes + 16 + csum_size) {
- jbd_debug(4, "JBD2: Submit %d IOs\n", bufs);
+ jbd2_debug(4, "JBD2: Submit %d IOs\n", bufs);
/* Write an end-of-descriptor marker before
submitting the IOs. "tag" still points to
@@ -722,7 +761,7 @@ start_journal_io:
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
bh->b_end_io = journal_end_buffer_io_sync;
- submit_bh(REQ_OP_WRITE, REQ_SYNC, bh);
+ submit_bh(REQ_OP_WRITE | REQ_SYNC, bh);
}
cond_resched();
@@ -760,14 +799,14 @@ start_journal_io:
if (first_block < journal->j_tail)
freed += journal->j_last - journal->j_first;
/* Update tail only if we free significant amount of space */
- if (freed < journal->j_maxlen / 4)
+ if (freed < jbd2_journal_get_max_txn_bufs(journal))
update_tail = 0;
}
J_ASSERT(commit_transaction->t_state == T_COMMIT);
commit_transaction->t_state = T_COMMIT_DFLUSH;
write_unlock(&journal->j_state_lock);
- /*
+ /*
* 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
@@ -775,7 +814,7 @@ start_journal_io:
if (commit_transaction->t_need_data_flush &&
(journal->j_fs_dev != journal->j_dev) &&
(journal->j_flags & JBD2_BARRIER))
- blkdev_issue_flush(journal->j_fs_dev, GFP_NOFS, NULL);
+ blkdev_issue_flush(journal->j_fs_dev);
/* Done it all: now write the commit record asynchronously. */
if (jbd2_has_feature_async_commit(journal)) {
@@ -798,7 +837,7 @@ start_journal_io:
so we incur less scheduling load.
*/
- jbd_debug(3, "JBD2: commit phase 3\n");
+ jbd2_debug(3, "JBD2: commit phase 3\n");
while (!list_empty(&io_bufs)) {
struct buffer_head *bh = list_entry(io_bufs.prev,
@@ -841,7 +880,7 @@ start_journal_io:
J_ASSERT (commit_transaction->t_shadow_list == NULL);
- jbd_debug(3, "JBD2: commit phase 4\n");
+ jbd2_debug(3, "JBD2: commit phase 4\n");
/* Here we wait for the revoke record and descriptor record buffers */
while (!list_empty(&log_bufs)) {
@@ -865,7 +904,7 @@ start_journal_io:
if (err)
jbd2_journal_abort(journal, err);
- jbd_debug(3, "JBD2: commit phase 5\n");
+ jbd2_debug(3, "JBD2: commit phase 5\n");
write_lock(&journal->j_state_lock);
J_ASSERT(commit_transaction->t_state == T_COMMIT_DFLUSH);
commit_transaction->t_state = T_COMMIT_JFLUSH;
@@ -882,7 +921,7 @@ start_journal_io:
stats.run.rs_blocks_logged++;
if (jbd2_has_feature_async_commit(journal) &&
journal->j_flags & JBD2_BARRIER) {
- blkdev_issue_flush(journal->j_dev, GFP_NOFS, NULL);
+ blkdev_issue_flush(journal->j_dev);
}
if (err)
@@ -904,7 +943,7 @@ start_journal_io:
transaction can be removed from any checkpoint list it was on
before. */
- jbd_debug(3, "JBD2: commit phase 6\n");
+ jbd2_debug(3, "JBD2: commit phase 6\n");
J_ASSERT(list_empty(&commit_transaction->t_inode_list));
J_ASSERT(commit_transaction->t_buffers == NULL);
@@ -997,9 +1036,10 @@ restart_loop:
* journalled data) we need to unmap buffer and clear
* more bits. We also need to be careful about the check
* because the data page mapping can get cleared under
- * out hands, which alse need not to clear more bits
- * because the page and buffers will be freed and can
- * never be reused once we are done with them.
+ * our hands. Note that if mapping == NULL, we don't
+ * need to make buffer unmapped because the page is
+ * already detached from the mapping and buffers cannot
+ * get reused.
*/
mapping = READ_ONCE(bh->b_page->mapping);
if (mapping && !sb_is_blkdev_sb(mapping->host->i_sb)) {
@@ -1080,7 +1120,7 @@ restart_loop:
/* Done with this transaction! */
- jbd_debug(3, "JBD2: commit phase 7\n");
+ jbd2_debug(3, "JBD2: commit phase 7\n");
J_ASSERT(commit_transaction->t_state == T_COMMIT_JFLUSH);
@@ -1118,12 +1158,16 @@ restart_loop:
if (journal->j_commit_callback)
journal->j_commit_callback(journal, commit_transaction);
+ if (journal->j_fc_cleanup_callback)
+ journal->j_fc_cleanup_callback(journal, 1, commit_transaction->t_tid);
trace_jbd2_end_commit(journal, commit_transaction);
- jbd_debug(1, "JBD2: commit %d complete, head %d\n",
+ jbd2_debug(1, "JBD2: commit %d complete, head %d\n",
journal->j_commit_sequence, journal->j_tail_sequence);
write_lock(&journal->j_state_lock);
+ journal->j_flags &= ~JBD2_FULL_COMMIT_ONGOING;
+ journal->j_flags &= ~JBD2_FAST_COMMIT_ONGOING;
spin_lock(&journal->j_list_lock);
commit_transaction->t_state = T_FINISHED;
/* Check if the transaction can be dropped now that we are finished */
@@ -1135,6 +1179,7 @@ restart_loop:
spin_unlock(&journal->j_list_lock);
write_unlock(&journal->j_state_lock);
wake_up(&journal->j_wait_done_commit);
+ wake_up(&journal->j_fc_wait);
/*
* Calculate overall stats
Copyright © 1996 – 2023 Jason A. Donenfeld. All Rights Reverse Engineered.