SQL Server 7.0 and 2000 provide a number of functions that enable database queries to generate text messages. In some cases, the functions create a text message and store it in a variable; in others, the functions directly display the message. Two vulnerabilities associated with these functions have been discovered.

The first vulnerability results because of a flaw in the functions themselves. Several of the functions don't adequately verify that the requested text will fit into the buffer that's supplied to hold it. A buffer overrun could occur as a result, and could be used either to run code in the security context of the SQL Server service or to cause the SQL Server service to fail. SQL Server can be configured to run in various security contexts, and by default runs as a domain user. The precise privileges the attacker could gain would depend on the specific security context that the service runs in.

The second vulnerability results because of a format string vulnerability in the C runtime functions that the SQL Server functions call when installed on Windows NT(r) 4.0, Windows(r) 2000 or Windows XP. Although format string vulnerabilities often can be exploited to run code of the attacker's choice, that is not true in this case. Because of the specific way this vulnerability occurs, the C Runtime code would always be overrun with the same values regardless of the attacker's inputs. As a result, this vulnerability could only be used as a denial of service.