Re: DB2UDB hack

The code base for DB2 on Linux, Unix, & Windows is effectively the same [thus the DB2 LUW moniker]. If an error is in common code, then it could easily span those servers. FWiW, a visual for the common code in DB2 LUW:
http://www.ibm.com/developerworks/db2/library/techarticle/dm-0611zikopoulos/#section1

According to the defect, a /DB2JDS service listening on TCP port/ would have to be both available and active on the [DB2 for] i5/OS. I am not aware of that service, but it both could and may exist. Regardless, I am doubtful that "a specially crafted packet" being "improperly processed" would both result in "a standard stack overflow situation" and then enabling an unauthenticated [database] access to perform "remote code execution" of "arbitrary code" within the i5/OS framework. In the framework of i5/OS a code stream is encapsulated in an object against which only supported methods are allowed -- thus removing much of the /arbitrary/ from the equation. The program object can be addressed, but the code stream can not be addressed.

Apparently DB2JDS is the "DB2 JDBC Applet Server" which is "started with the command: db2jstrt <port number>" and used for a Type3 driver; may run as db2jd as a [service?] name -- without the 's'. The following article mentions that function in scenarios 9 and 7, and notes that Type3 is not even supported by i5/OS, which I believe is still correct:
http://www.ibm.com/developerworks/db2/library/techarticle/dm-0401chong

Regards, Chuck