4 files changed, 307 insertions, 406 deletions

diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c
index 6914598022ce..91389c8aee8a 100644
--- a/fs/jbd2/checkpoint.c
+++ b/fs/jbd2/checkpoint.c
@@ -688,7 +688,6 @@ void __jbd2_journal_drop_transaction(journal_t *journal, transaction_t *transact
 
         J_ASSERT(transaction->t_state == T_FINISHED);
         J_ASSERT(transaction->t_buffers == NULL);
-        J_ASSERT(transaction->t_sync_datalist == NULL);
         J_ASSERT(transaction->t_forget == NULL);
         J_ASSERT(transaction->t_iobuf_list == NULL);
         J_ASSERT(transaction->t_shadow_list == NULL);
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index a2ed72f7ceee..f8b3be873226 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -22,6 +22,8 @@
 #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.
@@ -37,8 +39,8 @@ static void journal_end_buffer_io_sync(struct buffer_head *bh, int uptodate)
 }
 
 /*
- * 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
@@ -80,21 +82,6 @@ nope:
 }
 
 /*
- * 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 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
@@ -112,6 +99,7 @@ static int journal_submit_commit_record(journal_t *journal,
         struct buffer_head *bh;
         int ret;
         int barrier_done = 0;
+        struct timespec now = current_kernel_time();
 
         if (is_journal_aborted(journal))
                 return 0;
@@ -126,6 +114,8 @@ static int journal_submit_commit_record(journal_t *journal,
         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)) {
@@ -197,159 +187,104 @@ static int journal_wait_on_commit_record(struct buffer_head *bh)
 }
 
 /*
- * Wait for all submitted IO to complete.
+ * 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_wait_on_locked_list(journal_t *journal,
-                                       transaction_t *commit_transaction)
+static int journal_submit_inode_data_buffers(struct address_space *mapping)
 {
-        int ret = 0;
-        struct journal_head *jh;
-
-        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)))
-                                ret = -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);
-        }
+        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;
-  }
+}
 
-static void journal_do_submit_data(struct buffer_head **wbuf, int bufs)
+/*
+ * 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)
 {
-        int i;
+        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 (need_resched() || spin_needbreak(&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)
@@ -524,21 +459,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
          * 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 if commit
-         * record is to be written synchronously.
-         */
-        spin_lock(&journal->j_list_lock);
-        if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
-                JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT))
-                err = journal_wait_on_locked_list(journal,
-                                                commit_transaction);
-
-        spin_unlock(&journal->j_list_lock);
-
+        err = journal_submit_data_buffers(journal, commit_transaction);
         if (err)
                 jbd2_journal_abort(journal, err);
 
@@ -547,16 +468,6 @@ void jbd2_journal_commit_transaction(journal_t *journal)
         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:
@@ -574,6 +485,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
         J_ASSERT(commit_transaction->t_nr_buffers <=
                  commit_transaction->t_outstanding_credits);
 
+        err = 0;
         descriptor = NULL;
         bufs = 0;
         while (commit_transaction->t_buffers) {
@@ -748,15 +660,19 @@ start_journal_io:
                                                  &cbh, crc32_sum);
                 if (err)
                         __jbd2_journal_abort_hard(journal);
-
-                spin_lock(&journal->j_list_lock);
-                err = journal_wait_on_locked_list(journal,
-                                                commit_transaction);
-                spin_unlock(&journal->j_list_lock);
-                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
@@ -768,7 +684,7 @@ start_journal_io:
            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.
@@ -827,7 +743,7 @@ wait_for_iobuf:
 
         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:
@@ -854,7 +770,7 @@ wait_for_iobuf:
                 /* AKPM: bforget here */
         }
 
-        jbd_debug(3, "JBD: commit phase 6\n");
+        jbd_debug(3, "JBD: commit phase 5\n");
 
         if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
                 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
@@ -874,9 +790,9 @@ wait_for_iobuf:
            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);
@@ -997,7 +913,7 @@ restart_loop:
 
         /* 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);
 
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index 2e24567c4a79..b26c6d9fe6ae 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -50,7 +50,6 @@ EXPORT_SYMBOL(jbd2_journal_unlock_updates);
 EXPORT_SYMBOL(jbd2_journal_get_write_access);
 EXPORT_SYMBOL(jbd2_journal_get_create_access);
 EXPORT_SYMBOL(jbd2_journal_get_undo_access);
-EXPORT_SYMBOL(jbd2_journal_dirty_data);
 EXPORT_SYMBOL(jbd2_journal_dirty_metadata);
 EXPORT_SYMBOL(jbd2_journal_release_buffer);
 EXPORT_SYMBOL(jbd2_journal_forget);
@@ -82,6 +81,10 @@ EXPORT_SYMBOL(jbd2_journal_blocks_per_page);
 EXPORT_SYMBOL(jbd2_journal_invalidatepage);
 EXPORT_SYMBOL(jbd2_journal_try_to_free_buffers);
 EXPORT_SYMBOL(jbd2_journal_force_commit);
+EXPORT_SYMBOL(jbd2_journal_file_inode);
+EXPORT_SYMBOL(jbd2_journal_init_jbd_inode);
+EXPORT_SYMBOL(jbd2_journal_release_jbd_inode);
+EXPORT_SYMBOL(jbd2_journal_begin_ordered_truncate);
 
 static int journal_convert_superblock_v1(journal_t *, journal_superblock_t *);
 static void __journal_abort_soft (journal_t *journal, int errno);
@@ -2195,6 +2198,54 @@ void jbd2_journal_put_journal_head(struct journal_head *jh)
 }
 
 /*
+ * Initialize jbd inode head
+ */
+void jbd2_journal_init_jbd_inode(struct jbd2_inode *jinode, struct inode *inode)
+{
+        jinode->i_transaction = NULL;
+        jinode->i_next_transaction = NULL;
+        jinode->i_vfs_inode = inode;
+        jinode->i_flags = 0;
+        INIT_LIST_HEAD(&jinode->i_list);
+}
+
+/*
+ * Function to be called before we start removing inode from memory (i.e.,
+ * clear_inode() is a fine place to be called from). It removes inode from
+ * transaction's lists.
+ */
+void jbd2_journal_release_jbd_inode(journal_t *journal,
+                                    struct jbd2_inode *jinode)
+{
+        int writeout = 0;
+
+        if (!journal)
+                return;
+restart:
+        spin_lock(&journal->j_list_lock);
+        /* Is commit writing out inode - we have to wait */
+        if (jinode->i_flags & JI_COMMIT_RUNNING) {
+                wait_queue_head_t *wq;
+                DEFINE_WAIT_BIT(wait, &jinode->i_flags, __JI_COMMIT_RUNNING);
+                wq = bit_waitqueue(&jinode->i_flags, __JI_COMMIT_RUNNING);
+                prepare_to_wait(wq, &wait.wait, TASK_UNINTERRUPTIBLE);
+                spin_unlock(&journal->j_list_lock);
+                schedule();
+                finish_wait(wq, &wait.wait);
+                goto restart;
+        }
+
+        /* Do we need to wait for data writeback? */
+        if (journal->j_committing_transaction == jinode->i_transaction)
+                writeout = 1;
+        if (jinode->i_transaction) {
+                list_del(&jinode->i_list);
+                jinode->i_transaction = NULL;
+        }
+        spin_unlock(&journal->j_list_lock);
+}
+
+/*
  * debugfs tunables
  */
 #ifdef CONFIG_JBD2_DEBUG
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index d6e006e67804..4f7cadbb19fa 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -41,7 +41,6 @@ static void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh);
  *      new transaction and we can't block without protecting against other
  *      processes trying to touch the journal while it is in transition.
  *
- * Called under j_state_lock
  */
 
 static transaction_t *
@@ -52,6 +51,7 @@ jbd2_get_transaction(journal_t *journal, transaction_t *transaction)
         transaction->t_tid = journal->j_transaction_sequence++;
         transaction->t_expires = jiffies + journal->j_commit_interval;
         spin_lock_init(&transaction->t_handle_lock);
+        INIT_LIST_HEAD(&transaction->t_inode_list);
 
         /* Set up the commit timer for the new transaction. */
         journal->j_commit_timer.expires = round_jiffies(transaction->t_expires);
@@ -943,183 +943,6 @@ out:
 }
 
 /**
- * int jbd2_journal_dirty_data() -  mark a buffer as containing dirty data which
- *                             needs to be flushed before we can commit the
- *                             current transaction.
- * @handle: transaction
- * @bh: bufferhead to mark
- *
- * The buffer is placed on the transaction's data list and is marked as
- * belonging to the transaction.
- *
- * Returns error number or 0 on success.
- *
- * jbd2_journal_dirty_data() can be called via page_launder->ext3_writepage
- * by kswapd.
- */
-int jbd2_journal_dirty_data(handle_t *handle, struct buffer_head *bh)
-{
-        journal_t *journal = handle->h_transaction->t_journal;
-        int need_brelse = 0;
-        struct journal_head *jh;
-
-        if (is_handle_aborted(handle))
-                return 0;
-
-        jh = jbd2_journal_add_journal_head(bh);
-        JBUFFER_TRACE(jh, "entry");
-
-        /*
-         * The buffer could *already* be dirty.  Writeout can start
-         * at any time.
-         */
-        jbd_debug(4, "jh: %p, tid:%d\n", jh, handle->h_transaction->t_tid);
-
-        /*
-         * What if the buffer is already part of a running transaction?
-         *
-         * There are two cases:
-         * 1) It is part of the current running transaction.  Refile it,
-         *    just in case we have allocated it as metadata, deallocated
-         *    it, then reallocated it as data.
-         * 2) It is part of the previous, still-committing transaction.
-         *    If all we want to do is to guarantee that the buffer will be
-         *    written to disk before this new transaction commits, then
-         *    being sure that the *previous* transaction has this same
-         *    property is sufficient for us!  Just leave it on its old
-         *    transaction.
-         *
-         * In case (2), the buffer must not already exist as metadata
-         * --- that would violate write ordering (a transaction is free
-         * to write its data at any point, even before the previous
-         * committing transaction has committed).  The caller must
-         * never, ever allow this to happen: there's nothing we can do
-         * about it in this layer.
-         */
-        jbd_lock_bh_state(bh);
-        spin_lock(&journal->j_list_lock);
-
-        /* Now that we have bh_state locked, are we really still mapped? */
-        if (!buffer_mapped(bh)) {
-                JBUFFER_TRACE(jh, "unmapped buffer, bailing out");
-                goto no_journal;
-        }
-
-        if (jh->b_transaction) {
-                JBUFFER_TRACE(jh, "has transaction");
-                if (jh->b_transaction != handle->h_transaction) {
-                        JBUFFER_TRACE(jh, "belongs to older transaction");
-                        J_ASSERT_JH(jh, jh->b_transaction ==
-                                        journal->j_committing_transaction);
-
-                        /* @@@ IS THIS TRUE  ? */
-                        /*
-                         * Not any more.  Scenario: someone does a write()
-                         * in data=journal mode.  The buffer's transaction has
-                         * moved into commit.  Then someone does another
-                         * write() to the file.  We do the frozen data copyout
-                         * and set b_next_transaction to point to j_running_t.
-                         * And while we're in that state, someone does a
-                         * writepage() in an attempt to pageout the same area
-                         * of the file via a shared mapping.  At present that
-                         * calls jbd2_journal_dirty_data(), and we get right here.
-                         * It may be too late to journal the data.  Simply
-                         * falling through to the next test will suffice: the
-                         * data will be dirty and wil be checkpointed.  The
-                         * ordering comments in the next comment block still
-                         * apply.
-                         */
-                        //J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
-
-                        /*
-                         * If we're journalling data, and this buffer was
-                         * subject to a write(), it could be metadata, forget
-                         * or shadow against the committing transaction.  Now,
-                         * someone has dirtied the same darn page via a mapping
-                         * and it is being writepage()'d.
-                         * We *could* just steal the page from commit, with some
-                         * fancy locking there.  Instead, we just skip it -
-                         * don't tie the page's buffers to the new transaction
-                         * at all.
-                         * Implication: if we crash before the writepage() data
-                         * is written into the filesystem, recovery will replay
-                         * the write() data.
-                         */
-                        if (jh->b_jlist != BJ_None &&
-                                        jh->b_jlist != BJ_SyncData &&
-                                        jh->b_jlist != BJ_Locked) {
-                                JBUFFER_TRACE(jh, "Not stealing");
-                                goto no_journal;
-                        }
-
-                        /*
-                         * This buffer may be undergoing writeout in commit.  We
-                         * can't return from here and let the caller dirty it
-                         * again because that can cause the write-out loop in
-                         * commit to never terminate.
-                         */
-                        if (buffer_dirty(bh)) {
-                                get_bh(bh);
-                                spin_unlock(&journal->j_list_lock);
-                                jbd_unlock_bh_state(bh);
-                                need_brelse = 1;
-                                sync_dirty_buffer(bh);
-                                jbd_lock_bh_state(bh);
-                                spin_lock(&journal->j_list_lock);
-                                /* Since we dropped the lock... */
-                                if (!buffer_mapped(bh)) {
-                                        JBUFFER_TRACE(jh, "buffer got unmapped");
-                                        goto no_journal;
-                                }
-                                /* The buffer may become locked again at any
-                                   time if it is redirtied */
-                        }
-
-                        /* journal_clean_data_list() may have got there first */
-                        if (jh->b_transaction != NULL) {
-                                JBUFFER_TRACE(jh, "unfile from commit");
-                                __jbd2_journal_temp_unlink_buffer(jh);
-                                /* It still points to the committing
-                                 * transaction; move it to this one so
-                                 * that the refile assert checks are
-                                 * happy. */
-                                jh->b_transaction = handle->h_transaction;
-                        }
-                        /* The buffer will be refiled below */
-
-                }
-                /*
-                 * Special case --- the buffer might actually have been
-                 * allocated and then immediately deallocated in the previous,
-                 * committing transaction, so might still be left on that
-                 * transaction's metadata lists.
-                 */
-                if (jh->b_jlist != BJ_SyncData && jh->b_jlist != BJ_Locked) {
-                        JBUFFER_TRACE(jh, "not on correct data list: unfile");
-                        J_ASSERT_JH(jh, jh->b_jlist != BJ_Shadow);
-                        __jbd2_journal_temp_unlink_buffer(jh);
-                        jh->b_transaction = handle->h_transaction;
-                        JBUFFER_TRACE(jh, "file as data");
-                        __jbd2_journal_file_buffer(jh, handle->h_transaction,
-                                                BJ_SyncData);
-                }
-        } else {
-                JBUFFER_TRACE(jh, "not on a transaction");
-                __jbd2_journal_file_buffer(jh, handle->h_transaction, BJ_SyncData);
-        }
-no_journal:
-        spin_unlock(&journal->j_list_lock);
-        jbd_unlock_bh_state(bh);
-        if (need_brelse) {
-                BUFFER_TRACE(bh, "brelse");
-                __brelse(bh);
-        }
-        JBUFFER_TRACE(jh, "exit");
-        jbd2_journal_put_journal_head(jh);
-        return 0;
-}
-
-/**
  * int jbd2_journal_dirty_metadata() -  mark a buffer as containing dirty metadata
  * @handle: transaction to add buffer to.
  * @bh: buffer to mark
@@ -1541,10 +1364,10 @@ __blist_del_buffer(struct journal_head **list, struct journal_head *jh)
  * Remove a buffer from the appropriate transaction list.
  *
  * Note that this function can *change* the value of
- * bh->b_transaction->t_sync_datalist, t_buffers, t_forget,
- * t_iobuf_list, t_shadow_list, t_log_list or t_reserved_list.  If the caller
- * is holding onto a copy of one of thee pointers, it could go bad.
- * Generally the caller needs to re-read the pointer from the transaction_t.
+ * bh->b_transaction->t_buffers, t_forget, t_iobuf_list, t_shadow_list,
+ * t_log_list or t_reserved_list.  If the caller is holding onto a copy of one
+ * of these pointers, it could go bad.  Generally the caller needs to re-read
+ * the pointer from the transaction_t.
  *
  * Called under j_list_lock.  The journal may not be locked.
  */
@@ -1566,9 +1389,6 @@ void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh)
         switch (jh->b_jlist) {
         case BJ_None:
                 return;
-        case BJ_SyncData:
-                list = &transaction->t_sync_datalist;
-                break;
         case BJ_Metadata:
                 transaction->t_nr_buffers--;
                 J_ASSERT_JH(jh, transaction->t_nr_buffers >= 0);
@@ -1589,9 +1409,6 @@ void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh)
         case BJ_Reserved:
                 list = &transaction->t_reserved_list;
                 break;
-        case BJ_Locked:
-                list = &transaction->t_locked_list;
-                break;
         }
 
         __blist_del_buffer(list, jh);
@@ -1634,15 +1451,7 @@ __journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh)
                 goto out;
 
         spin_lock(&journal->j_list_lock);
-        if (jh->b_transaction != NULL && jh->b_cp_transaction == NULL) {
-                if (jh->b_jlist == BJ_SyncData || jh->b_jlist == BJ_Locked) {
-                        /* A written-back ordered data buffer */
-                        JBUFFER_TRACE(jh, "release data");
-                        __jbd2_journal_unfile_buffer(jh);
-                        jbd2_journal_remove_journal_head(bh);
-                        __brelse(bh);
-                }
-        } else if (jh->b_cp_transaction != NULL && jh->b_transaction == NULL) {
+        if (jh->b_cp_transaction != NULL && jh->b_transaction == NULL) {
                 /* written-back checkpointed metadata buffer */
                 if (jh->b_jlist == BJ_None) {
                         JBUFFER_TRACE(jh, "remove from checkpoint list");
@@ -1656,12 +1465,43 @@ out:
         return;
 }
 
+/*
+ * jbd2_journal_try_to_free_buffers() could race with
+ * jbd2_journal_commit_transaction(). The later might still hold the
+ * reference count to the buffers when inspecting them on
+ * t_syncdata_list or t_locked_list.
+ *
+ * jbd2_journal_try_to_free_buffers() will call this function to
+ * wait for the current transaction to finish syncing data buffers, before
+ * try to free that buffer.
+ *
+ * Called with journal->j_state_lock hold.
+ */
+static void jbd2_journal_wait_for_transaction_sync_data(journal_t *journal)
+{
+        transaction_t *transaction;
+        tid_t tid;
+
+        spin_lock(&journal->j_state_lock);
+        transaction = journal->j_committing_transaction;
+
+        if (!transaction) {
+                spin_unlock(&journal->j_state_lock);
+                return;
+        }
+
+        tid = transaction->t_tid;
+        spin_unlock(&journal->j_state_lock);
+        jbd2_log_wait_commit(journal, tid);
+}
 
 /**
  * int jbd2_journal_try_to_free_buffers() - try to free page buffers.
  * @journal: journal for operation
  * @page: to try and free
- * @unused_gfp_mask: unused
+ * @gfp_mask: we use the mask to detect how hard should we try to release
+ * buffers. If __GFP_WAIT and __GFP_FS is set, we wait for commit code to
+ * release the buffers.
  *
  *
  * For all the buffers on this page,
@@ -1690,9 +1530,11 @@ out:
  * journal_try_to_free_buffer() is changing its state.  But that
  * cannot happen because we never reallocate freed data as metadata
  * while the data is part of a transaction.  Yes?
+ *
+ * Return 0 on failure, 1 on success
  */
 int jbd2_journal_try_to_free_buffers(journal_t *journal,
-                                struct page *page, gfp_t unused_gfp_mask)
+                                struct page *page, gfp_t gfp_mask)
 {
         struct buffer_head *head;
         struct buffer_head *bh;
@@ -1708,7 +1550,8 @@ int jbd2_journal_try_to_free_buffers(journal_t *journal,
                 /*
                  * We take our own ref against the journal_head here to avoid
                  * having to add tons of locking around each instance of
-                 * jbd2_journal_remove_journal_head() and jbd2_journal_put_journal_head().
+                 * jbd2_journal_remove_journal_head() and
+                 * jbd2_journal_put_journal_head().
                  */
                 jh = jbd2_journal_grab_journal_head(bh);
                 if (!jh)
@@ -1721,7 +1564,28 @@ int jbd2_journal_try_to_free_buffers(journal_t *journal,
                 if (buffer_jbd(bh))
                         goto busy;
         } while ((bh = bh->b_this_page) != head);
+
         ret = try_to_free_buffers(page);
+
+        /*
+         * There are a number of places where jbd2_journal_try_to_free_buffers()
+         * could race with jbd2_journal_commit_transaction(), the later still
+         * holds the reference to the buffers to free while processing them.
+         * try_to_free_buffers() failed to free those buffers. Some of the
+         * caller of releasepage() request page buffers to be dropped, otherwise
+         * treat the fail-to-free as errors (such as generic_file_direct_IO())
+         *
+         * So, if the caller of try_to_release_page() wants the synchronous
+         * behaviour(i.e make sure buffers are dropped upon return),
+         * let's wait for the current transaction to finish flush of
+         * dirty data buffers, then try to free those buffers again,
+         * with the journal locked.
+         */
+        if (ret == 0 && (gfp_mask & __GFP_WAIT) && (gfp_mask & __GFP_FS)) {
+                jbd2_journal_wait_for_transaction_sync_data(journal);
+                ret = try_to_free_buffers(page);
+        }
+
 busy:
         return ret;
 }
@@ -1823,6 +1687,7 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
         if (!buffer_jbd(bh))
                 goto zap_buffer_unlocked;
 
+        /* OK, we have data buffer in journaled mode */
         spin_lock(&journal->j_state_lock);
         jbd_lock_bh_state(bh);
         spin_lock(&journal->j_list_lock);
@@ -1886,15 +1751,6 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
                 }
         } else if (transaction == journal->j_committing_transaction) {
                 JBUFFER_TRACE(jh, "on committing transaction");
-                if (jh->b_jlist == BJ_Locked) {
-                        /*
-                         * The buffer is on the committing transaction's locked
-                         * list.  We have the buffer locked, so I/O has
-                         * completed.  So we can nail the buffer now.
-                         */
-                        may_free = __dispose_buffer(jh, transaction);
-                        goto zap_buffer;
-                }
                 /*
                  * If it is committing, we simply cannot touch it.  We
                  * can remove it's next_transaction pointer from the
@@ -2027,9 +1883,6 @@ void __jbd2_journal_file_buffer(struct journal_head *jh,
                 J_ASSERT_JH(jh, !jh->b_committed_data);
                 J_ASSERT_JH(jh, !jh->b_frozen_data);
                 return;
-        case BJ_SyncData:
-                list = &transaction->t_sync_datalist;
-                break;
         case BJ_Metadata:
                 transaction->t_nr_buffers++;
                 list = &transaction->t_buffers;
@@ -2049,9 +1902,6 @@ void __jbd2_journal_file_buffer(struct journal_head *jh,
         case BJ_Reserved:
                 list = &transaction->t_reserved_list;
                 break;
-        case BJ_Locked:
-                list =  &transaction->t_locked_list;
-                break;
         }
 
         __blist_add_buffer(list, jh);
@@ -2141,3 +1991,88 @@ void jbd2_journal_refile_buffer(journal_t *journal, struct journal_head *jh)
         spin_unlock(&journal->j_list_lock);
         __brelse(bh);
 }
+
+/*
+ * File inode in the inode list of the handle's transaction
+ */
+int jbd2_journal_file_inode(handle_t *handle, struct jbd2_inode *jinode)
+{
+        transaction_t *transaction = handle->h_transaction;
+        journal_t *journal = transaction->t_journal;
+
+        if (is_handle_aborted(handle))
+                return -EIO;
+
+        jbd_debug(4, "Adding inode %lu, tid:%d\n", jinode->i_vfs_inode->i_ino,
+                        transaction->t_tid);
+
+        /*
+         * First check whether inode isn't already on the transaction's
+         * lists without taking the lock. Note that this check is safe
+         * without the lock as we cannot race with somebody removing inode
+         * from the transaction. The reason is that we remove inode from the
+         * transaction only in journal_release_jbd_inode() and when we commit
+         * the transaction. We are guarded from the first case by holding
+         * a reference to the inode. We are safe against the second case
+         * because if jinode->i_transaction == transaction, commit code
+         * cannot touch the transaction because we hold reference to it,
+         * and if jinode->i_next_transaction == transaction, commit code
+         * will only file the inode where we want it.
+         */
+        if (jinode->i_transaction == transaction ||
+            jinode->i_next_transaction == transaction)
+                return 0;
+
+        spin_lock(&journal->j_list_lock);
+
+        if (jinode->i_transaction == transaction ||
+            jinode->i_next_transaction == transaction)
+                goto done;
+
+        /* On some different transaction's list - should be
+         * the committing one */
+        if (jinode->i_transaction) {
+                J_ASSERT(jinode->i_next_transaction == NULL);
+                J_ASSERT(jinode->i_transaction ==
+                                        journal->j_committing_transaction);
+                jinode->i_next_transaction = transaction;
+                goto done;
+        }
+        /* Not on any transaction list... */
+        J_ASSERT(!jinode->i_next_transaction);
+        jinode->i_transaction = transaction;
+        list_add(&jinode->i_list, &transaction->t_inode_list);
+done:
+        spin_unlock(&journal->j_list_lock);
+
+        return 0;
+}
+
+/*
+ * This function must be called when inode is journaled in ordered mode
+ * before truncation happens. It starts writeout of truncated part in
+ * case it is in the committing transaction so that we stand to ordered
+ * mode consistency guarantees.
+ */
+int jbd2_journal_begin_ordered_truncate(struct jbd2_inode *inode,
+                                        loff_t new_size)
+{
+        journal_t *journal;
+        transaction_t *commit_trans;
+        int ret = 0;
+
+        if (!inode->i_transaction && !inode->i_next_transaction)
+                goto out;
+        journal = inode->i_transaction->t_journal;
+        spin_lock(&journal->j_state_lock);
+        commit_trans = journal->j_committing_transaction;
+        spin_unlock(&journal->j_state_lock);
+        if (inode->i_transaction == commit_trans) {
+                ret = filemap_fdatawrite_range(inode->i_vfs_inode->i_mapping,
+                        new_size, LLONG_MAX);
+                if (ret)
+                        jbd2_journal_abort(journal, ret);
+        }
+out:
+        return ret;
+}