1 files changed, 45 insertions, 2 deletions

diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index c067a8cae63b..17f557f01cf0 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -331,6 +331,10 @@ void jbd2_journal_commit_transaction(journal_t *journal)
         struct buffer_head *cbh = NULL; /* For transactional checksums */
         __u32 crc32_sum = ~0;
         struct blk_plug plug;
+        /* Tail of the journal */
+        unsigned long first_block;
+        tid_t first_tid;
+        int update_tail;
 
         /*
          * First job: lock down the current transaction and wait for
@@ -340,7 +344,18 @@ 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_journal_update_superblock(journal, 1);
+                mutex_lock(&journal->j_checkpoint_mutex);
+                /*
+                 * We hold j_checkpoint_mutex so tail cannot change under us.
+                 * We don't need any special data guarantees for writing sb
+                 * since journal is empty and it is ok for write to be
+                 * flushed only with transaction commit.
+                 */
+                jbd2_journal_update_sb_log_tail(journal,
+                                                journal->j_tail_sequence,
+                                                journal->j_tail,
+                                                WRITE_SYNC);
+                mutex_unlock(&journal->j_checkpoint_mutex);
         } else {
                 jbd_debug(3, "superblock not updated\n");
         }
@@ -677,10 +692,30 @@ start_journal_io:
                 err = 0;
         }
 
+        /*
+         * Get current oldest transaction in the log before we issue flush
+         * to the filesystem device. After the flush we can be sure that
+         * blocks of all older transactions are checkpointed to persistent
+         * storage and we will be safe to update journal start in the
+         * superblock with the numbers we get here.
+         */
+        update_tail =
+                jbd2_journal_get_log_tail(journal, &first_tid, &first_block);
+
         write_lock(&journal->j_state_lock);
+        if (update_tail) {
+                long freed = first_block - journal->j_tail;
+
+                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)
+                        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
@@ -831,6 +866,14 @@ wait_for_iobuf:
         if (err)
                 jbd2_journal_abort(journal, err);
 
+        /*
+         * Now disk caches for filesystem device are flushed so we are safe to
+         * erase checkpointed transactions from the log by updating journal
+         * superblock.
+         */
+        if (update_tail)
+                jbd2_update_log_tail(journal, first_tid, first_block);
+
         /* 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
@@ -1048,7 +1091,7 @@ restart_loop:
         jbd_debug(1, "JBD2: commit %d complete, head %d\n",
                   journal->j_commit_sequence, journal->j_tail_sequence);
         if (to_free)
-                kfree(commit_transaction);
+                jbd2_journal_free_transaction(commit_transaction);
 
         wake_up(&journal->j_wait_done_commit);
 }