Work environment

Expected behavior

When completing the bug injection step, your yield should NOT be 0.

Actual behavior

Currently, the yield is consistently 0 exploitable bugs.

Steps to reproduce the behavior

Run LAVA end-to-end, and you will see 0 bugs with exploits at the end.

• Use code markdown CODE to make your code visible

Additional Logs, screenshots, source code, configuration dump, ...

At first, using PANDA v1.8.76, it can pass one injection, but once a certain AST node name, *(((**(((*f)._chain)))._chain)) is passed, Clang can't handle that complicated AST node name on direct code injection. So it would crash.

The obvious attempt was to ban AST node names with **, as seen in here, but it seems that this is the wrong approach? See the PR for better notes on how this was fixed.