Context-Sensitive Data-Flow

Release:

6.1.2

Date:

Oct 28, 2026

Context-sensitive data-flow graphs for C programs, tracking how values flow through the program under a calling context (the call_i / ret_i families), as in “Taming Transitive Redundancy for Context-Free Language Reachability”.

Graph	Num Nodes	Num Edges	Reachable pairs	Size (MB)	Download
xz	30492	37173	358834	1.10	xz.tar.gz 📥
nab	31215	37484	739646	1.95	nab.tar.gz 📥
leela	47665	63996	662466	2.01	leela.tar.gz 📥
x264	138702	201034	20259480	48.87	x264.tar.gz 📥
parest	233900	307850	1342540	5.17	parest.tar.gz 📥
imagick	331177	445544	12687034	35.41	imagick.tar.gz 📥
povray	346034	581210	34599413	2.45	povray.tar.gz 📥
cactus	359200	580297	47806209	2.74	cactus.tar.gz 📥
omnetpp	463454	958487	8424500	6.42	omnetpp.tar.gz 📥
perlbench	605864	1114892	297504186	4.64	perlbench.tar.gz 📥

Canonical grammars

Productions with index \(i\) are duplicated for each call site in the analyzed program. The start nonterminal is \(A\).

\[\begin{split}A \, \rightarrow \, A \, A \mid a \mid \varepsilon \, \\ A \, \rightarrow \, call_i \, A \, ret_i \, \\\end{split}\]

RSM (stored as aa.rsm in each archive):

start: A
[box A]
start: 0
final: 0
0 --a--> 0
0 --call_i--> p_i
p_i --A--> q_i
q_i --ret_i--> 0