When a fresh transaction begins, the tree mod log must be clean. Users of
the tree modification log must ensure they never span across transaction
boundaries.
We reset the sequence to 0 in this safe situation to make absolutely sure
overflow can't happen.
Signed-off-by: Jan Schmidt <list.bt...@jan-o-sch.net>
---
fs/btrfs/transaction.c | 9 +++++++++
1 files changed, 9 insertions(+), 0 deletions(-)
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index eb2bd82..3f50cba 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -122,6 +122,15 @@ loop:
cur_trans->delayed_refs.flushing = 0;
cur_trans->delayed_refs.run_delayed_start = 0;
cur_trans->delayed_refs.seq = 1;
+
+ /*
+ * although the tree mod log is per file system and not per transaction,
+ * the log must never go across transaction boundaries.
+ */
+ BUG_ON(!list_empty(&fs_info->tree_mod_seq_list));
+ BUG_ON(!RB_EMPTY_ROOT(&fs_info->tree_mod_log));
+ atomic_set(&fs_info->tree_mod_seq, 0);
+
init_waitqueue_head(&cur_trans->delayed_refs.seq_wait);
spin_lock_init(&cur_trans->commit_lock);
spin_lock_init(&cur_trans->delayed_refs.lock);
--
1.7.3.4
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
