1 files changed, 34 insertions, 36 deletions

diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 8e2f8275a253..73ed02f061e1 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -107,7 +107,6 @@ static void jbd2_get_transaction(journal_t *journal,
 	transaction->t_start_time = ktime_get();
 	transaction->t_tid = journal->j_transaction_sequence++;
 	transaction->t_expires = jiffies + journal->j_commit_interval;
-	spin_lock_init(&transaction->t_handle_lock);
 	atomic_set(&transaction->t_updates, 0);
 	atomic_set(&transaction->t_outstanding_credits,
 		   jbd2_descriptor_blocks_per_trans(journal) +
@@ -139,26 +138,22 @@ static void jbd2_get_transaction(journal_t *journal,
 /*
  * Update transaction's maximum wait time, if debugging is enabled.
  *
- * In order for t_max_wait to be reliable, it must be protected by a
- * lock.  But doing so will mean that start_this_handle() can not be
- * run in parallel on SMP systems, which limits our scalability.  So
- * unless debugging is enabled, we no longer update t_max_wait, which
- * means that maximum wait time reported by the jbd2_run_stats
- * tracepoint will always be zero.
+ * t_max_wait is carefully updated here with use of atomic compare exchange.
+ * Note that there could be multiplre threads trying to do this simultaneously
+ * hence using cmpxchg to avoid any use of locks in this case.
+ * With this t_max_wait can be updated w/o enabling jbd2_journal_enable_debug.
  */
 static inline void update_t_max_wait(transaction_t *transaction,
 				     unsigned long ts)
 {
-#ifdef CONFIG_JBD2_DEBUG
-	if (jbd2_journal_enable_debug &&
-	    time_after(transaction->t_start, ts)) {
-		ts = jbd2_time_diff(ts, transaction->t_start);
-		spin_lock(&transaction->t_handle_lock);
-		if (ts > transaction->t_max_wait)
-			transaction->t_max_wait = ts;
-		spin_unlock(&transaction->t_handle_lock);
+	unsigned long oldts, newts;
+
+	if (time_after(transaction->t_start, ts)) {
+		newts = jbd2_time_diff(ts, transaction->t_start);
+		oldts = READ_ONCE(transaction->t_max_wait);
+		while (oldts < newts)
+			oldts = cmpxchg(&transaction->t_max_wait, oldts, newts);
 	}
-#endif
 }
 
 /*
@@ -690,7 +685,6 @@ int jbd2_journal_extend(handle_t *handle, int nblocks, int revoke_records)
 		DIV_ROUND_UP(
 			handle->h_revoke_credits_requested,
 			journal->j_revoke_records_per_block);
-	spin_lock(&transaction->t_handle_lock);
 	wanted = atomic_add_return(nblocks,
 				   &transaction->t_outstanding_credits);
 
@@ -698,7 +692,7 @@ int jbd2_journal_extend(handle_t *handle, int nblocks, int revoke_records)
 		jbd_debug(3, "denied handle %p %d blocks: "
 			  "transaction too large\n", handle, nblocks);
 		atomic_sub(nblocks, &transaction->t_outstanding_credits);
-		goto unlock;
+		goto error_out;
 	}
 
 	trace_jbd2_handle_extend(journal->j_fs_dev->bd_dev,
@@ -714,8 +708,6 @@ int jbd2_journal_extend(handle_t *handle, int nblocks, int revoke_records)
 	result = 0;
 
 	jbd_debug(3, "extended handle %p by %d\n", handle, nblocks);
-unlock:
-	spin_unlock(&transaction->t_handle_lock);
 error_out:
 	read_unlock(&journal->j_state_lock);
 	return result;
@@ -842,27 +834,35 @@ EXPORT_SYMBOL(jbd2_journal_restart);
  */
 void jbd2_journal_wait_updates(journal_t *journal)
 {
-	transaction_t *commit_transaction = journal->j_running_transaction;
+	DEFINE_WAIT(wait);
 
-	if (!commit_transaction)
-		return;
+	while (1) {
+		/*
+		 * Note that the running transaction can get freed under us if
+		 * this transaction is getting committed in
+		 * jbd2_journal_commit_transaction() ->
+		 * jbd2_journal_free_transaction(). This can only happen when we
+		 * release j_state_lock -> schedule() -> acquire j_state_lock.
+		 * Hence we should everytime retrieve new j_running_transaction
+		 * value (after j_state_lock release acquire cycle), else it may
+		 * lead to use-after-free of old freed transaction.
+		 */
+		transaction_t *transaction = journal->j_running_transaction;
 
-	spin_lock(&commit_transaction->t_handle_lock);
-	while (atomic_read(&commit_transaction->t_updates)) {
-		DEFINE_WAIT(wait);
+		if (!transaction)
+			break;
 
 		prepare_to_wait(&journal->j_wait_updates, &wait,
-					TASK_UNINTERRUPTIBLE);
-		if (atomic_read(&commit_transaction->t_updates)) {
-			spin_unlock(&commit_transaction->t_handle_lock);
-			write_unlock(&journal->j_state_lock);
-			schedule();
-			write_lock(&journal->j_state_lock);
-			spin_lock(&commit_transaction->t_handle_lock);
+				TASK_UNINTERRUPTIBLE);
+		if (!atomic_read(&transaction->t_updates)) {
+			finish_wait(&journal->j_wait_updates, &wait);
+			break;
 		}
+		write_unlock(&journal->j_state_lock);
+		schedule();
 		finish_wait(&journal->j_wait_updates, &wait);
+		write_lock(&journal->j_state_lock);
 	}
-	spin_unlock(&commit_transaction->t_handle_lock);
 }
 
 /**
@@ -877,8 +877,6 @@ void jbd2_journal_wait_updates(journal_t *journal)
  */
 void jbd2_journal_lock_updates(journal_t *journal)
 {
-	DEFINE_WAIT(wait);
-
 	jbd2_might_wait_for_commit(journal);
 
 	write_lock(&journal->j_state_lock);