Skip to content

Evil ELFs and Dishonest DWARFs (Jack Baker)

Engineers explore architecture and security mechanisms for Evil ELFs and Dishonest DWARFs (Jack Baker).

Summary

Abusing Debug Symbols

Jack Baker’s LayerOne talk “Evil ELFs and Dishonest DWARFs” subverts ELF debug information to hinder reverse engineering—not by misnaming functions but by breaking analyst tools. Targets include IDA Pro with Hex-Rays, Ghidra, radare2, and GDB dynamic analysis.

Breaking Static Analysis Trust

Static tools infer program structure without execution—disassembling variable-length x86, finding function boundaries, and decompiling types. Baker crafts malicious DWARF and symbol data exploiting those trust assumptions, causing truncated functions, wrong boundaries, and decompiler failures rather than exploiting memory corruption in the tools themselves.

Stripping Detection and Response

Because attacks rely on debug symbols, stripped binaries lack rich DWARF; Baker demonstrates dynamically detecting stripping and adapting techniques across stripped, basic symtab, and full debug symbol tiers. The research shows metadata poisoning can defeat analysts even when code behavior remains unchanged.

Related talks