Moon Slayer - Chapter 25

“large amount of rewritten source code when querying the database”????? Please, someone explain this to me. What I currently have is this sequence of events:

1. A bad entry makes its way to the DB, either by someone inputting something they shouldn’t have (like “five” for an input that expects number “5”), or a disk error.
2. Some function queries the DB, expecting an integer (5), and instead gets a string (in C/C++ semantics, it’s a pointer to a character – char*)
3. There is some sort of dispatch table (an array of funtion pointers/other unique identifiers that may be converted to function pointers. Again, in c/c++ semantics that would be – return_type (dispatch_table**)(arguments). ).
4. The an element of the dispatch table gets indexed using the corrupted result from the DB (that would be – dispatch_table[unfiltered_db_result] )
5. Now, we have a pointer to a “random” place in memory, created by adding the two pointers (the string and the array, which is what indexing is, just adding something to the array pointer).
6. Something calls the pointer.
7. A “random” place in memory is currently executed with some arguments, modifying some other things.
8. The original thread that called the function pointer resides close to the kernel, thus bypassing detection (assuming there is an, idk, ANTIVIRUS?)

That is the only way I can think of that may “rewrite the source code” by “querying the DB”. If someone has their own opinion, please free me from the pain I feel while reading these segments.