fabsx00 / bjoern-old Goto Github PK

For source code, storing the id of the corresponding function in each node has shown to be helpful in creating queries with good performance. We need this for the step 'functionToAPISymbols' in particular, that is used for similar function detection using 'bjoern-tools'.

I commented out the destructors for BjoernFunctionNode and BjoernInstructionNode as they seem to corrupt memory. This issue had gone unnoticed because the destructor of the base class wasn't virtual, and hence, the destructors were never called.

Hi!

The url of boost is wrong. It should be

http://www.boost.org/users/history/version_1_53_0.html

Looking up nodes by their address to determine edge destinations when writing CSVs is an idea that is fundamentally flawed as
(1) Functions and their entry basic blocks have the same address
(2) Basic blocks have the same address as their first instruction

In contrast, the BjoernNodes are actually unique, so I'm going to lookup by BjoernNode instead.