Btrfs: fix race in WAIT_SYNC ioctl

We check whether transid is already committed via last_trans_committed and
then search through trans_list for pending transactions.  If
last_trans_committed is updated by btrfs_commit_transaction after we check
it (there is no locking), we will fail to find the committed transaction
and return EINVAL to the caller.  This has been observed occasionally by
ceph-osd (which uses this ioctl heavily).

Fix by rechecking whether the provided transid <= last_trans_committed
after the search fails, and if so return 0.

Signed-off-by: Sage Weil <sage@redhat.com>
Signed-off-by: Chris Mason <clm@fb.com>

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index a47b1000a6e5ebbb9f46be289df5e83e6491889c..86ef8d79c19feb549fb0f22bda737d4e3e6d47e9 100644 (file)
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -609,7 +609,6 @@ int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid)
                if (transid <= root->fs_info->last_trans_committed)
                        goto out;
 
-               ret = -EINVAL;
                /* find specified transaction */
                spin_lock(&root->fs_info->trans_lock);
                list_for_each_entry(t, &root->fs_info->trans_list, list) {
@@ -625,9 +624,16 @@ int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid)
                        }
                }
                spin_unlock(&root->fs_info->trans_lock);
-               /* The specified transaction doesn't exist */
-               if (!cur_trans)
+
+               /*
+                * The specified transaction doesn't exist, or we
+                * raced with btrfs_commit_transaction
+                */
+               if (!cur_trans) {
+                       if (transid > root->fs_info->last_trans_committed)
+                               ret = -EINVAL;
                        goto out;
+               }
        } else {
                /* find newest transaction that is committing | committed */
                spin_lock(&root->fs_info->trans_lock);