Interprocedural Dominance Graph Construction for Scalable Code Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for computing interprocedural dominators in software programs are either complex or sacrifice simplicity and ease of implementation, and they often require full program analysis, which is inefficient for large programs or incremental changes.
Innovation Solution
A method and system that compute interprocedural dominance graphs by combining intraprocedural dominator graphs, identifying interprocedural dominance candidates, and computing interprocedural dominator sets, allowing for parallel processing and incremental analysis, which reduces the need for full program reanalysis and improves efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods for computing interprocedural dominators are used, then dominance relationships can be identified, but the computational complexity increases and full program reanalysis is required
Solution Approach 1:
The patent divides the program into procedure-specific dominator graphs, computing dominators locally within each procedure rather than performing a single global analysis. This segmentation allows the overall dominance computation to be broken into smaller, more manageable pieces that can be processed independently and then combined.
Solution Approach 2:
The patent transitions from traditional single-dimension dominator computation to a multi-dimensional approach by considering interprocedural edges and dominance relationships across procedure boundaries. This adds a new dimension to the analysis that captures cross-procedure dominance without requiring complete reanalysis of all procedures.
2Measurement precision
If full program analysis is performed to compute interprocedural dominators, then complete dominance information is obtained, but the analysis time increases significantly for large programs
Solution Approach 1:
The patent performs preliminary computation of procedure-specific dominator graphs before combining them into the interprocedural dominance graph. This preliminary action allows much of the dominance computation to be done in advance for each procedure, reducing the work needed during the final interprocedural combination phase.
Solution Approach 2:
The patent computes dominator relationships partially within each procedure first, then supplements this with interprocedural edge analysis. This partial action approach avoids the need to perform complete global analysis from scratch, achieving sufficient dominance information with reduced computational effort.
3Measurement precision
If traditional dominator algorithms are used, then dominance relationships are computed, but simplicity and ease of implementation are sacrificed
Solution Approach 1:
The patent segments the dominator computation into procedure-specific graphs that can be computed using simpler local algorithms, then combines these with interprocedural edge information. This segmentation maintains implementation simplicity by allowing each procedure to be analyzed independently with straightforward algorithms.
Solution Approach 2:
The patent merges procedure-specific dominator graphs with interprocedural edge information to create the complete interprocedural dominance graph. This combining approach integrates local simplicity with global accuracy, maintaining ease of implementation while achieving comprehensive dominance relationships.
Data Source
AI summary
According to an aspect of some embodiments of the present invention there is provided a computerized method of analyzing code of a software program for dominance relationships between a plurality of functions of the software program, the method comprising: receiving source code of a software program, the source code having a plurality of functions; identifying a plurality of intraprocedural dominator graphs each for another of the plurality of functions; combining the plurality of intraprocedural dominator graphs to create an interprocedural dominance graph with edges that logically connect between nodes of the plurality of functions; identifying a plurality of interprocedural dominance relations between nodes in different functions of the plurality of functions using the interprocedural dominance graph; and analyzing the software program according to the plurality of interprocedural dominance relations.