logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

SILVA: A Scalable Incremental Layered Sparse Value-Flow Analysis

delete2025-10-04
delete0
PRE
AI
Y
Yu Wang
K
Ke Wang
L
Linzhang Wang
DOI：10.1145/3725214delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
Layered sparse value-flow analysis (SVFA) is a prominent static analysis for resolving program dependencies. Despite the significant progress, SVFA still suffers from scalability issue. In light of the natural, continuous evolution of software, we introduce SILVA, the first incremental layered SVFA that scales to large, real-world programs efficiently. At the core of SILVA lies a novel incremental pointer analysis and incremental Mod-Ref analysis. Our extensive experiments on large-scale, real-world C/C++ programs demonstrate its effectiveness: SILVA achieves nearly a 7× speedup over SVF, the state-of-the-art layered SVFA, without losing any precision. Moreover, our incremental pointer and Mod-Ref analysis algorithms are 12× and 5× faster than existing methods, respectively. Regarding the impact of the size of the code changes on SILVA’s effectiveness, we find that SILVA outperforms SVF for changes up to 10K lines—well beyond the typical scope of code commits in real-world software development.

Journal

A
ACM Transactions on Software Engineering and Methodology
IF:
6.2
Papers:
1.2K
Citations:
3.4K

Organization

No organization information available