反编译 2024：反编译研究复苏的一年

Summary: 2024 was a resurgent year for decompilation. Academic publications made up nearly 30% of all top decompilation publications. It saw a talk tour by Cristina Cifuentes, a world-renowned decompilation panel at Recon 2024, and a high top-tier publication record. Three emerging trends in 2024 decompilation research are defining "good" decompilation, the rise of AI in decompilation, and the application of theory with open-source implementations. Other aspects like updates to decompilers and great talks also flourished. Overall, it was a special year and an optimistic start for future decompilation research.
Main Points:
- 2024 had a significant increase in academic decompilation publications.
- Cristina Cifuentes' tour and the Recon 2024 panel were notable events.
- Three trends in decompilation research: defining good decompilation, AI in decompilation, and theory application.
- Many decompilers saw updates and new industry approaches.
- Encouragement to contribute to the Decompilation Wiki and be involved in a new decompilation workshop.
Key Information:
- Graph shows the increase in decompilation works published since 1993.
- Details of each trend: Dramko's work on decompiler fidelity, SAILR for finding mismatches, R2I for simple decompilation, D-Helix for semantic accuracy, DeGPT and ReSym using LLMs, VarBERT and TyGr for specific predictions.
- Open-source implementations of 2024 papers and their integration into real-world decompilers.
- Acknowledgment of Dustin Fraze's work on DARPA's CHESS program.
Important Details:
- Decompilation research was previously saturated in non-top-tier conferences but is now maturing.
- AI is improving specific tasks in decompilation but not replacing decompilers.
- The Decompilation Wiki is a platform for organizing and guiding the field.
- There was a meeting of legends at the Recon 2024 panel with diverse perspectives.