#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
+#include <linux/jiffies.h>
+#include <linux/crc32.h>
+#include <linux/writeback.h>
+#include <linux/backing-dev.h>
/*
* Default IO end handler for temporary BJ_IO buffer_heads.
}
/*
- * When an ext3-ordered file is truncated, it is possible that many pages are
- * not sucessfully freed, because they are attached to a committing transaction.
+ * When an ext4 file is truncated, it is possible that some pages are not
+ * successfully freed, because they are attached to a committing transaction.
* 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
}
/*
- * Try to acquire jbd_lock_bh_state() against the buffer, when j_list_lock is
- * held. For ranking reasons we must trylock. If we lose, schedule away and
- * return 0. j_list_lock is dropped in this case.
- */
-static int inverted_lock(journal_t *journal, struct buffer_head *bh)
-{
- if (!jbd_trylock_bh_state(bh)) {
- spin_unlock(&journal->j_list_lock);
- schedule();
- return 0;
- }
- return 1;
-}
-
-/* Done it all: now write the commit record. We should have
+ * Done it all: now submit the commit record. We should have
* 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
*/
-static int journal_write_commit_record(journal_t *journal,
- transaction_t *commit_transaction)
+static int journal_submit_commit_record(journal_t *journal,
+ transaction_t *commit_transaction,
+ struct buffer_head **cbh,
+ __u32 crc32_sum)
{
struct journal_head *descriptor;
+ struct commit_header *tmp;
struct buffer_head *bh;
- int i, ret;
+ int ret;
int barrier_done = 0;
+ struct timespec now = current_kernel_time();
if (is_journal_aborted(journal))
return 0;
bh = jh2bh(descriptor);
- /* AKPM: buglet - add `i' to tmp! */
- for (i = 0; i < bh->b_size; i += 512) {
- journal_header_t *tmp = (journal_header_t*)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);
+ 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);
+ tmp->h_commit_sec = cpu_to_be64(now.tv_sec);
+ tmp->h_commit_nsec = cpu_to_be32(now.tv_nsec);
+
+ 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);
}
- JBUFFER_TRACE(descriptor, "write commit block");
+ JBUFFER_TRACE(descriptor, "submit commit block");
+ lock_buffer(bh);
+ get_bh(bh);
set_buffer_dirty(bh);
- if (journal->j_flags & JBD2_BARRIER) {
+ set_buffer_uptodate(bh);
+ bh->b_end_io = journal_end_buffer_io_sync;
+
+ if (journal->j_flags & JBD2_BARRIER &&
+ !JBD2_HAS_INCOMPAT_FEATURE(journal,
+ JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
set_buffer_ordered(bh);
barrier_done = 1;
}
- ret = sync_dirty_buffer(bh);
+ ret = submit_bh(WRITE, bh);
+ if (barrier_done)
+ clear_buffer_ordered(bh);
+
/* is it possible for another commit to fail at roughly
* the same time as this one? If so, we don't want to
* trust the barrier flag in the super, but instead want
spin_unlock(&journal->j_state_lock);
/* And try again, without the barrier */
- clear_buffer_ordered(bh);
+ lock_buffer(bh);
set_buffer_uptodate(bh);
set_buffer_dirty(bh);
- ret = sync_dirty_buffer(bh);
+ ret = submit_bh(WRITE, bh);
}
- put_bh(bh); /* One for getblk() */
- jbd2_journal_put_journal_head(descriptor);
+ *cbh = bh;
+ return ret;
+}
+
+/*
+ * This function along with journal_submit_commit_record
+ * allows to write the commit record asynchronously.
+ */
+static int journal_wait_on_commit_record(struct buffer_head *bh)
+{
+ int ret = 0;
- return (ret == -EIO);
+ clear_buffer_dirty(bh);
+ wait_on_buffer(bh);
+
+ if (unlikely(!buffer_uptodate(bh)))
+ ret = -EIO;
+ put_bh(bh); /* One for getblk() */
+ jbd2_journal_put_journal_head(bh2jh(bh));
+
+ return ret;
}
-static void journal_do_submit_data(struct buffer_head **wbuf, int bufs)
+/*
+ * 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 i;
+ 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),
+ .for_writepages = 1,
+ };
+
+ ret = generic_writepages(mapping, &wbc);
+ return ret;
+}
+
+/*
+ * 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.
+ */
+static int journal_submit_data_buffers(journal_t *journal,
+ transaction_t *commit_transaction)
+{
+ struct jbd2_inode *jinode;
+ int err, ret = 0;
+ struct address_space *mapping;
- for (i = 0; i < bufs; i++) {
- wbuf[i]->b_end_io = end_buffer_write_sync;
- /* We use-up our safety reference in submit_bh() */
- submit_bh(WRITE, wbuf[i]);
+ 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;
+ 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.
+ */
+ err = journal_submit_inode_data_buffers(mapping);
+ if (!ret)
+ ret = err;
+ spin_lock(&journal->j_list_lock);
+ J_ASSERT(jinode->i_transaction == commit_transaction);
+ jinode->i_flags &= ~JI_COMMIT_RUNNING;
+ wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING);
}
+ spin_unlock(&journal->j_list_lock);
+ return ret;
}
/*
- * Submit all the data buffers to disk
+ * Wait for data submitted for writeout, refile inodes to proper
+ * transaction if needed.
+ *
*/
-static void journal_submit_data_buffers(journal_t *journal,
- transaction_t *commit_transaction)
+static int journal_finish_inode_data_buffers(journal_t *journal,
+ transaction_t *commit_transaction)
{
- struct journal_head *jh;
- struct buffer_head *bh;
- int locked;
- int bufs = 0;
- struct buffer_head **wbuf = journal->j_wbuf;
+ struct jbd2_inode *jinode, *next_i;
+ int err, ret = 0;
- /*
- * Whenever we unlock the journal and sleep, things can get added
- * onto ->t_sync_datalist, so we have to keep looping back to
- * write_out_data until we *know* that the list is empty.
- *
- * Cleanup any flushed data buffers from the data list. Even in
- * abort mode, we want to flush this out as soon as possible.
- */
-write_out_data:
- cond_resched();
+ /* 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) {
+ jinode->i_flags |= JI_COMMIT_RUNNING;
+ spin_unlock(&journal->j_list_lock);
+ err = filemap_fdatawait(jinode->i_vfs_inode->i_mapping);
+ if (!ret)
+ ret = err;
+ spin_lock(&journal->j_list_lock);
+ jinode->i_flags &= ~JI_COMMIT_RUNNING;
+ wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING);
+ }
- while (commit_transaction->t_sync_datalist) {
- jh = commit_transaction->t_sync_datalist;
- bh = jh2bh(jh);
- locked = 0;
-
- /* Get reference just to make sure buffer does not disappear
- * when we are forced to drop various locks */
- get_bh(bh);
- /* If the buffer is dirty, we need to submit IO and hence
- * we need the buffer lock. We try to lock the buffer without
- * blocking. If we fail, we need to drop j_list_lock and do
- * blocking lock_buffer().
- */
- if (buffer_dirty(bh)) {
- if (test_set_buffer_locked(bh)) {
- BUFFER_TRACE(bh, "needs blocking lock");
- spin_unlock(&journal->j_list_lock);
- /* Write out all data to prevent deadlocks */
- journal_do_submit_data(wbuf, bufs);
- bufs = 0;
- lock_buffer(bh);
- spin_lock(&journal->j_list_lock);
- }
- locked = 1;
- }
- /* We have to get bh_state lock. Again out of order, sigh. */
- if (!inverted_lock(journal, bh)) {
- jbd_lock_bh_state(bh);
- spin_lock(&journal->j_list_lock);
- }
- /* Someone already cleaned up the buffer? */
- if (!buffer_jbd(bh)
- || jh->b_transaction != commit_transaction
- || jh->b_jlist != BJ_SyncData) {
- jbd_unlock_bh_state(bh);
- if (locked)
- unlock_buffer(bh);
- BUFFER_TRACE(bh, "already cleaned up");
- put_bh(bh);
- continue;
- }
- if (locked && test_clear_buffer_dirty(bh)) {
- BUFFER_TRACE(bh, "needs writeout, adding to array");
- wbuf[bufs++] = bh;
- __jbd2_journal_file_buffer(jh, commit_transaction,
- BJ_Locked);
- jbd_unlock_bh_state(bh);
- if (bufs == journal->j_wbufsize) {
- spin_unlock(&journal->j_list_lock);
- journal_do_submit_data(wbuf, bufs);
- bufs = 0;
- goto write_out_data;
- }
- } else if (!locked && buffer_locked(bh)) {
- __jbd2_journal_file_buffer(jh, commit_transaction,
- BJ_Locked);
- jbd_unlock_bh_state(bh);
- put_bh(bh);
+ /* 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 {
- BUFFER_TRACE(bh, "writeout complete: unfile");
- __jbd2_journal_unfile_buffer(jh);
- jbd_unlock_bh_state(bh);
- if (locked)
- unlock_buffer(bh);
- jbd2_journal_remove_journal_head(bh);
- /* Once for our safety reference, once for
- * jbd2_journal_remove_journal_head() */
- put_bh(bh);
- put_bh(bh);
- }
-
- if (lock_need_resched(&journal->j_list_lock)) {
- spin_unlock(&journal->j_list_lock);
- goto write_out_data;
+ jinode->i_transaction = NULL;
}
}
spin_unlock(&journal->j_list_lock);
- journal_do_submit_data(wbuf, bufs);
+
+ return ret;
+}
+
+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 inline void write_tag_block(int tag_bytes, journal_block_tag_t *tag,
+static void write_tag_block(int tag_bytes, journal_block_tag_t *tag,
unsigned long long block)
{
tag->t_blocknr = cpu_to_be32(block & (u32)~0);
*/
void jbd2_journal_commit_transaction(journal_t *journal)
{
+ struct transaction_stats_s stats;
transaction_t *commit_transaction;
struct journal_head *jh, *new_jh, *descriptor;
struct buffer_head **wbuf = journal->j_wbuf;
int tag_flag;
int i;
int tag_bytes = journal_tag_bytes(journal);
+ struct buffer_head *cbh = NULL; /* For transactional checksums */
+ __u32 crc32_sum = ~0;
/*
* First job: lock down the current transaction and wait for
spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_LOCKED;
+ stats.u.run.rs_wait = commit_transaction->t_max_wait;
+ stats.u.run.rs_locked = jiffies;
+ stats.u.run.rs_running = jbd2_time_diff(commit_transaction->t_start,
+ stats.u.run.rs_locked);
+
spin_lock(&commit_transaction->t_handle_lock);
while (commit_transaction->t_updates) {
DEFINE_WAIT(wait);
*/
jbd2_journal_switch_revoke_table(journal);
+ stats.u.run.rs_flushing = jiffies;
+ stats.u.run.rs_locked = jbd2_time_diff(stats.u.run.rs_locked,
+ stats.u.run.rs_flushing);
+
commit_transaction->t_state = T_FLUSH;
journal->j_committing_transaction = commit_transaction;
journal->j_running_transaction = NULL;
jbd_debug (3, "JBD: commit phase 2\n");
- /*
- * First, drop modified flag: all accesses to the buffers
- * will be tracked for a new trasaction only -bzzz
- */
- spin_lock(&journal->j_list_lock);
- if (commit_transaction->t_buffers) {
- new_jh = jh = commit_transaction->t_buffers->b_tnext;
- do {
- J_ASSERT_JH(new_jh, new_jh->b_modified == 1 ||
- new_jh->b_modified == 0);
- new_jh->b_modified = 0;
- new_jh = new_jh->b_tnext;
- } while (new_jh != jh);
- }
- spin_unlock(&journal->j_list_lock);
-
/*
* Now start flushing things to disk, in the order they appear
* on the transaction lists. Data blocks go first.
*/
- err = 0;
- journal_submit_data_buffers(journal, commit_transaction);
-
- /*
- * Wait for all previously submitted IO to complete.
- */
- spin_lock(&journal->j_list_lock);
- while (commit_transaction->t_locked_list) {
- struct buffer_head *bh;
-
- jh = commit_transaction->t_locked_list->b_tprev;
- bh = jh2bh(jh);
- get_bh(bh);
- if (buffer_locked(bh)) {
- spin_unlock(&journal->j_list_lock);
- wait_on_buffer(bh);
- if (unlikely(!buffer_uptodate(bh)))
- err = -EIO;
- spin_lock(&journal->j_list_lock);
- }
- if (!inverted_lock(journal, bh)) {
- put_bh(bh);
- spin_lock(&journal->j_list_lock);
- continue;
- }
- if (buffer_jbd(bh) && jh->b_jlist == BJ_Locked) {
- __jbd2_journal_unfile_buffer(jh);
- jbd_unlock_bh_state(bh);
- jbd2_journal_remove_journal_head(bh);
- put_bh(bh);
- } else {
- jbd_unlock_bh_state(bh);
- }
- put_bh(bh);
- cond_resched_lock(&journal->j_list_lock);
- }
- spin_unlock(&journal->j_list_lock);
-
+ err = journal_submit_data_buffers(journal, commit_transaction);
if (err)
jbd2_journal_abort(journal, err);
jbd_debug(3, "JBD: commit phase 2\n");
- /*
- * If we found any dirty or locked buffers, then we should have
- * looped back up to the write_out_data label. If there weren't
- * any then journal_clean_data_list should have wiped the list
- * clean by now, so check that it is in fact empty.
- */
- J_ASSERT (commit_transaction->t_sync_datalist == NULL);
-
- jbd_debug (3, "JBD: commit phase 3\n");
-
/*
* 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:
*/
+ spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_COMMIT;
+ spin_unlock(&journal->j_state_lock);
+ stats.u.run.rs_logging = jiffies;
+ stats.u.run.rs_flushing = jbd2_time_diff(stats.u.run.rs_flushing,
+ stats.u.run.rs_logging);
+ stats.u.run.rs_blocks = commit_transaction->t_outstanding_credits;
+ stats.u.run.rs_blocks_logged = 0;
+
+ J_ASSERT(commit_transaction->t_nr_buffers <=
+ commit_transaction->t_outstanding_credits);
+
+ err = 0;
descriptor = NULL;
bufs = 0;
while (commit_transaction->t_buffers) {
start_journal_io:
for (i = 0; i < bufs; i++) {
struct buffer_head *bh = wbuf[i];
+ /*
+ * Compute checksum.
+ */
+ if (JBD2_HAS_COMPAT_FEATURE(journal,
+ JBD2_FEATURE_COMPAT_CHECKSUM)) {
+ crc32_sum =
+ jbd2_checksum_data(crc32_sum, bh);
+ }
+
lock_buffer(bh);
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
submit_bh(WRITE, bh);
}
cond_resched();
+ stats.u.run.rs_blocks_logged += bufs;
/* Force a new descriptor to be generated next
time round the loop. */
}
}
+ /* Done it all: now write the commit record asynchronously. */
+
+ if (JBD2_HAS_INCOMPAT_FEATURE(journal,
+ JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+ err = journal_submit_commit_record(journal, commit_transaction,
+ &cbh, crc32_sum);
+ if (err)
+ __jbd2_journal_abort_hard(journal);
+ }
+
+ /*
+ * This is the right place to wait for data buffers both for ASYNC
+ * and !ASYNC commit. If commit is ASYNC, we need to wait only after
+ * the commit block went to disk (which happens above). If commit is
+ * SYNC, we need to wait for data buffers before we start writing
+ * commit block, which happens below in such setting.
+ */
+ err = journal_finish_inode_data_buffers(journal, commit_transaction);
+ if (err)
+ jbd2_journal_abort(journal, err);
+
/* 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
so we incur less scheduling load.
*/
- jbd_debug(3, "JBD: commit phase 4\n");
+ jbd_debug(3, "JBD: commit phase 3\n");
/*
* akpm: these are BJ_IO, and j_list_lock is not needed.
J_ASSERT (commit_transaction->t_shadow_list == NULL);
- jbd_debug(3, "JBD: commit phase 5\n");
+ jbd_debug(3, "JBD: commit phase 4\n");
/* Here we wait for the revoke record and descriptor record buffers */
wait_for_ctlbuf:
/* AKPM: bforget here */
}
- jbd_debug(3, "JBD: commit phase 6\n");
+ jbd_debug(3, "JBD: commit phase 5\n");
- if (journal_write_commit_record(journal, commit_transaction))
- err = -EIO;
+ if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
+ JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+ err = journal_submit_commit_record(journal, commit_transaction,
+ &cbh, crc32_sum);
+ if (err)
+ __jbd2_journal_abort_hard(journal);
+ }
+ if (!err && !is_journal_aborted(journal))
+ err = journal_wait_on_commit_record(cbh);
if (err)
jbd2_journal_abort(journal, err);
transaction can be removed from any checkpoint list it was on
before. */
- jbd_debug(3, "JBD: commit phase 7\n");
+ jbd_debug(3, "JBD: commit phase 6\n");
- J_ASSERT(commit_transaction->t_sync_datalist == NULL);
+ J_ASSERT(list_empty(&commit_transaction->t_inode_list));
J_ASSERT(commit_transaction->t_buffers == NULL);
J_ASSERT(commit_transaction->t_checkpoint_list == NULL);
J_ASSERT(commit_transaction->t_iobuf_list == NULL);
cp_transaction = jh->b_cp_transaction;
if (cp_transaction) {
JBUFFER_TRACE(jh, "remove from old cp transaction");
+ cp_transaction->t_chp_stats.cs_dropped++;
__jbd2_journal_remove_checkpoint(jh);
}
}
spin_unlock(&journal->j_list_lock);
/*
- * This is a bit sleazy. We borrow j_list_lock to protect
- * journal->j_committing_transaction in __jbd2_journal_remove_checkpoint.
- * Really, __jbd2_journal_remove_checkpoint should be using j_state_lock but
- * it's a bit hassle to hold that across __jbd2_journal_remove_checkpoint
+ * 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...
*/
spin_lock(&journal->j_state_lock);
spin_lock(&journal->j_list_lock);
/* Done with this transaction! */
- jbd_debug(3, "JBD: commit phase 8\n");
+ jbd_debug(3, "JBD: commit phase 7\n");
J_ASSERT(commit_transaction->t_state == T_COMMIT);
+ commit_transaction->t_start = jiffies;
+ stats.u.run.rs_logging = jbd2_time_diff(stats.u.run.rs_logging,
+ commit_transaction->t_start);
+
+ /*
+ * File the transaction for history
+ */
+ stats.ts_type = JBD2_STATS_RUN;
+ stats.ts_tid = commit_transaction->t_tid;
+ stats.u.run.rs_handle_count = commit_transaction->t_handle_count;
+ spin_lock(&journal->j_history_lock);
+ memcpy(journal->j_history + journal->j_history_cur, &stats,
+ sizeof(stats));
+ if (++journal->j_history_cur == journal->j_history_max)
+ journal->j_history_cur = 0;
+
+ /*
+ * Calculate overall stats
+ */
+ journal->j_stats.ts_tid++;
+ journal->j_stats.u.run.rs_wait += stats.u.run.rs_wait;
+ journal->j_stats.u.run.rs_running += stats.u.run.rs_running;
+ journal->j_stats.u.run.rs_locked += stats.u.run.rs_locked;
+ journal->j_stats.u.run.rs_flushing += stats.u.run.rs_flushing;
+ journal->j_stats.u.run.rs_logging += stats.u.run.rs_logging;
+ journal->j_stats.u.run.rs_handle_count += stats.u.run.rs_handle_count;
+ journal->j_stats.u.run.rs_blocks += stats.u.run.rs_blocks;
+ journal->j_stats.u.run.rs_blocks_logged += stats.u.run.rs_blocks_logged;
+ spin_unlock(&journal->j_history_lock);
+
commit_transaction->t_state = T_FINISHED;
J_ASSERT(commit_transaction == journal->j_committing_transaction);
journal->j_commit_sequence = commit_transaction->t_tid;