Hi, the two patches below fix waiting on transaction in ext3_sync_file. The problem is that log_start_commit() returns 1 only if it has really started committing. If the transaction is already undergoing commit, it will return 0 and thus we incorrectly proceed although we should have waited for commit to finish. This problem would be easily solved by always waiting for transaction commit. But we also need to know whether the commit really happened after we have entered ext3_sync_file - otherwise we have to send a barrier request to disk. Sending it in all cases has unnecessarily bad performance. Attached patch addresses the issue by providing a function journal_trans_will_send_data_barrier which takes transaction ID and returns whether the final stage of commit of this transaction has not happened yet. Ext3 then uses this function to check whether it needs to send an additional barrier or not. Any comments or review welcome. Honza -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
