jbd2: check 'jh->b_transaction' before removing it from checkpoint

commit 590a809 upstream.

Following process will corrupt ext4 image:
Step 1:
jbd2_journal_commit_transaction
 __jbd2_journal_insert_checkpoint(jh, commit_transaction)
 // Put jh into trans1->t_checkpoint_list
 journal->j_checkpoint_transactions = commit_transaction
 // Put trans1 into journal->j_checkpoint_transactions

Step 2:
do_get_write_access
 test_clear_buffer_dirty(bh) // clear buffer dirty，set jbd dirty
 __jbd2_journal_file_buffer(jh, transaction) // jh belongs to trans2

Step 3:
drop_cache
 journal_shrink_one_cp_list
  jbd2_journal_try_remove_checkpoint
   if (!trylock_buffer(bh))  // lock bh, true
   if (buffer_dirty(bh))     // buffer is not dirty
   __jbd2_journal_remove_checkpoint(jh)
   // remove jh from trans1->t_checkpoint_list

Step 4:
jbd2_log_do_checkpoint
 trans1 = journal->j_checkpoint_transactions
 // jh is not in trans1->t_checkpoint_list
 jbd2_cleanup_journal_tail(journal)  // trans1 is done

Step 5: Power cut, trans2 is not committed, jh is lost in next mounting.

Fix it by checking 'jh->b_transaction' before remove it from checkpoint.

Cc: [email protected]
Fixes: 46f881b ("jbd2: fix a race when checking checkpoint buffer busy")
Signed-off-by: Zhihao Cheng <[email protected]>
Signed-off-by: Zhang Yi <[email protected]>
Reviewed-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>

fs/jbd2/checkpoint.c

@@ -639,6 +639,8 @@ int jbd2_journal_try_remove_checkpoint(struct journal_head *jh)
 {
 	struct buffer_head *bh = jh2bh(jh);
 
+	if (jh->b_transaction)
+		return -EBUSY;
 	if (!trylock_buffer(bh))
 		return -EBUSY;
 	if (buffer_dirty(bh)) {