SynopsisThe remote Scientific Linux host is missing one or more security updates.
DescriptionMySQL did not require privileges such as 'SELECT' for the source table in a 'CREATE TABLE LIKE' statement. An authenticated user could obtain sensitive information, such as the table structure. (CVE-2007-3781)
A flaw was discovered in MySQL that allowed an authenticated user to gain update privileges for a table in another database, via a view that refers to the external table. (CVE-2007-3782)
MySQL did not require the 'DROP' privilege for 'RENAME TABLE' statements. An authenticated user could use this flaw to rename arbitrary tables. (CVE-2007-2691)
A flaw was discovered in the mysql_change_db function when returning from SQL SECURITY INVOKER stored routines. An authenticated user could use this flaw to gain database privileges. (CVE-2007-2692)
MySQL allowed an authenticated user to bypass logging mechanisms via SQL queries that contain the NULL character, which were not properly handled by the mysql_real_query function. (CVE-2006-0903)
MySQL allowed an authenticated user to access a table through a previously created MERGE table, even after the user's privileges were revoked from the original table, which might violate intended security policy. This is addressed by allowing the MERGE storage engine to be disabled, which can be done by running mysqld with the '--skip-merge' option. (CVE-2006-4031)
MySQL evaluated arguments in the wrong security context, which allowed an authenticated user to gain privileges through a routine that had been made available using 'GRANT EXECUTE'. (CVE-2006-4227)
Multiple flaws in MySQL allowed an authenticated user to cause the MySQL daemon to crash via crafted SQL queries. This only caused a temporary denial of service, as the MySQL daemon is automatically restarted after the crash. (CVE-2006-7232, CVE-2007-1420, CVE-2007-2583)
As well, these updated packages fix the following bugs :
- a separate counter was used for 'insert delayed' statements, which caused rows to be discarded. In these updated packages, 'insert delayed' statements no longer use a separate counter, which resolves this issue.
- due to a bug in the Native POSIX Thread Library, in certain situations, 'flush tables' caused a deadlock on tables that had a read lock. The mysqld daemon had to be killed forcefully. Now, 'COND_refresh' has been replaced with 'COND_global_read_lock', which resolves this issue.
- mysqld crashed if a query for an unsigned column type contained a negative value for a 'WHERE [column] NOT IN' subquery.
- in master and slave server situations, specifying 'on duplicate key update' for 'insert' statements did not update slave servers.
- in the mysql client, empty strings were displayed as 'NULL'. For example, running 'insert into [table-name] values (' ');' resulted in a 'NULL' entry being displayed when querying the table using 'select * from [table-name];'.
- a bug in the optimizer code resulted in certain queries executing much slower than expected.
- on 64-bit PowerPC architectures, MySQL did not calculate the thread stack size correctly, which could have caused MySQL to crash when overly-complex queries were used.
Note: these updated packages upgrade MySQL to version 5.0.45. For a full list of bug fixes and enhancements, refer to the MySQL release notes: http://dev.mysql.com/doc/refman/5.0/en/releasenotes-cs-5-0.html
SolutionUpdate the affected packages.