| Bug #109401 | Rows_query_log_event truncates query due to treating query as C string | ||
|---|---|---|---|
| Submitted: | 16 Dec 2022 2:01 | Modified: | 27 Aug 2024 15:37 |
| Reporter: | Jeremy Cole (Basic Quality Contributor) (OCA) | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Row Based Replication ( RBR ) | Severity: | S3 (Non-critical) |
| Version: | 8.0.31, 5.7.40, 8.0.33 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
| Tags: | binlog, binlog_rows_query_log_events, RBR | ||
[16 Dec 2022 10:52]
MySQL Verification Team
Hello Jeremy, Thank you for the report and feedback. Sincerely, Umesh
[27 Aug 2024 15:37]
Jon Stephens
Documented fix as follows in the MySQL 9.0.0 changelog:
When using row-based replication with
binlog_rows_query_log_events enabled, SQL statements are logged
in a Rows_query_log_event which is written before the
Table_map_event. Since SQL may contain any binary data including
embedded nulls (\0), using the C strlen() and snprintf() and
other similar functions which rely on processing up to the null
byte, it was possible for the query string to be truncated at
the first null byte, which resulted in processing incomplete
data.
We fix this by specifying the length of each
Rows_query_log_event such that it is no longer necessary to rely
on C style string methods for length calculations; in addition,
while processing the query, we now use functions which do not
rely on null termination. We also mark the first byte of a
Rows_query_log_event data body as unused.
Closed.
Description: When using RBR with the binlog_rows_query_log_events option enabled, SQL statements are logged in a Rows_query_log_event which is written before the Table_map_event. Since SQL may contain any binary data including embedded nulls (\0), it's inappropriate to use null-terminated strings or C string functions for query strings, and these binary log events are intended to use the event payload to determine the string length. However, in multiple places the C strlen() and snprintf() functions are used, causing the query string to be truncated at the first null byte. How to repeat: 1. Enable binary logging and binlog_rows_query_log_events. 2. Execute a query containing an embedded null (\0). 3. Observe that the query is truncated in the Rows_query_log_event event. For example, this can be reproduced based on the following case, creating a test MySQL instance with binlog_rows_query_log_events enabled: $ dbdeployer deploy single 8.0 --my-cnf-options binlog_rows_query_log_events A small test program is required to reproduce, because the mysql command doesn't easily allow writing a query with zero bytes in the query string, in this case I used ruby with the mysql2 gem: $ cat test.rb #!/usr/bin/env ruby require "mysql2" mysql = Mysql2::Client.new(socket: "/tmp/mysql_sandbox8031.sock", username: "msandbox", password: "msandbox") mysql.query("DROP TABLE IF EXISTS `test`.`t`") mysql.query("CREATE TABLE `test`.`t` (`name` VARCHAR(255))") mysql.query("INSERT INTO `test`.`t` (`name`) VALUES (\"foo\0bar\")") $ ruby test.rb The truncated query can be seen in the mysqlbinlog output, e.g.: $ mysqlbinlog -vvv --base64-output=decode-rows binlog.000001 ... # at 8893 #221215 17:53:20 server id 1 end_log_pos 8961 CRC32 0x8a780242 Rows_query # INSERT INTO `test`.`t` (`name`) VALUES ("foo # at 8961 #221215 17:53:20 server id 1 end_log_pos 9012 CRC32 0x200b11e6 Table_map: `test`.`t` mapped to number 109 # at 9012 #221215 17:53:20 server id 1 end_log_pos 9057 CRC32 0xd6114e06 Write_rows: table id 109 flags: STMT_END_F ### INSERT INTO `test`.`t` ### SET ### @1='foo\x00bar' /* VARSTRING(1020) meta=1020 nullable=1 is_null=0 */ ... To be sure that it's actually truncated in the binlog rather than just in mysqlbinlog's reading or printing of the query, you can verify with hexdump that the query itself is truncated: $ hexdump -C binlog.000001 ... 000022c0 63 1d 01 00 00 00 44 00 00 00 01 23 00 00 80 00 |c.....D....#....| 000022d0 2c 49 4e 53 45 52 54 20 49 4e 54 4f 20 60 74 65 |,INSERT INTO `te| 000022e0 73 74 60 2e 60 74 60 20 28 60 6e 61 6d 65 60 29 |st`.`t` (`name`)| 000022f0 20 56 41 4c 55 45 53 20 28 22 66 6f 6f 42 02 78 | VALUES ("fooB.x| 00002300 8a 10 5f 9b 63 13 01 00 00 00 33 00 00 00 34 23 |.._.c.....3...4#| ... Suggested fix: Audit all usage of Rows_query_log_event::m_rows_query and ensure it is properly handled in all cases and never treated as a null-terminated C string.