Опубликована: Июнь 1, 2020
Most Java static analysis frameworks provide an intermediate presentation (IR) of Bytecode to facilitate the development analyses. While such IRs are often based on three-address code, transformation itself is a great opportunity apply optimizations transformed as constant propagation.
Язык: Английский
Journal of Systems and Software, Год журнала: 2019, Номер 156, С. 312 - 327
Опубликована: Июль 9, 2019
Язык: Английский
Процитировано
46
Опубликована: Июль 10, 2019
Call graphs are widely used; in particular for advanced control- and data-flow analyses. Even though many call graph algorithms with different precision scalability properties have been proposed, a comprehensive understanding of sources unsoundness, their relevance, the capabilities existing this respect is missing. To address problem, we propose Judge, toolchain that helps unsoundness improving soundness graphs. In several experiments, use Judge an extensive test suite related to (a) compute capability profiles implementations Soot, WALA, DOOP, OPAL, (b) determine prevalence language features APIs affect modern Java Bytecode, (c) compare OPAL – highlighting important differences implementations, (d) evaluate necessary effort achieve project-specific reasonable sound We show soundness-relevant features/APIs frequently used support them differs vastly, up point where comparing computed by same base (e.g., RTA) but frameworks bogus. also can users establishing effort.
Язык: Английский
Процитировано
38
Опубликована: Июнь 27, 2020
Static analyses have problems modelling dynamic language features soundly while retaining acceptable precision. The problem is well-understood in theory, but there little evidence on how this impacts the analysis of real-world programs. We studied issue for call graph construction a set 31 Java programs using an oracle actual program behaviour recorded from executions built-in and synthesised test cases with high coverage, measured recall that being achieved by various static algorithms configurations, investigated which lead to false negatives.
Язык: Английский
Процитировано
32
Journal of Systems and Software, Год журнала: 2024, Номер 212, С. 111971 - 111971
Опубликована: Март 12, 2024
A core principle of open science is the clear, concise and accessible publication empirical data, including "raw" observational data as well processed results. However, in software engineering there are no established standards (de jure or de facto) for representing "opening" observations collected test-driven experiments — that is, involving execution subjects controlled scenarios. Execution therefore usually represented ad hoc ways, often making it abstruse difficult to access without significant manual effort. In this paper we present new structures designed address problem by clearly defining, correlating stimuli responses used execute experiments. To demonstrate their utility, show how they can be promote repetition, replication reproduction experimental evaluations AI-based code completion tools. We also proposed facilitate incremental expansion sets, thus repurposing addressing research questions.
Язык: Английский
Процитировано
3
Опубликована: Июль 16, 2018
Call graphs are at the core of many static analyses ranging from detection unused methods to advanced control-and data-flow analyses. Therefore, a comprehensive understanding precision and recall respective is crucial enable an assessment which call-graph construction algorithms suited in analysis scenario. For example, malware often obfuscated tries hide its intent by using Reflection. that do not represent reflective method calls are, therefore, limited use when analyzing such apps.
Язык: Английский
Процитировано
23
Lecture notes in computer science, Год журнала: 2018, Номер unknown, С. 69 - 88
Опубликована: Янв. 1, 2018
Язык: Английский
Процитировано
21
Опубликована: Июль 13, 2020
Current Java static analyzers, operating either on the source or bytecode level, exhibit unsoundness for programs that contain native code. We show Native Interface (JNI) specification, which is used by to interoperate with code, principled enough permit reasoning about effects of code program execution when it comes call-backs. Our approach consists disassembling binaries, recovering symbol information corresponds method signatures, and producing a model statically exercising these call-backs appropriate mock objects. The manages recover virtually all calls in both Android desktop applications—(a) achieving 100% native-to-application call-graph recall large applications (Chrome, Instagram) (b) capturing full call-back behavior XCorpus suite programs.
Язык: Английский
Процитировано
19
Опубликована: Июль 12, 2018
The dynamic proxy API is one of Java's most widely-used features, permitting principled run-time code generation and link- ing. Dynamic proxies can implement any set interfaces for- ward method calls to a special object that handles them reflectively. flexibility proxies, however, comes at the cost having dynamically generated layer bytecode cannot be penetrated by current static analyses. In this paper, we observe stylized enough permit analysis. We show how semantics modeled in straightforward manner as logical rules Doop analysis framework. This concise enables Doop's standard analyses process behind proxies. evaluate our approach analyzing XCorpus, corpus real-world Java programs: fully handle 95% its reported creation sites. Our handling results significant portions previously unreachable or incompletely- code.
Язык: Английский
Процитировано
20
IEEE Transactions on Software Engineering, Год журнала: 2021, Номер 48(9), С. 3613 - 3625
Опубликована: Авг. 4, 2021
The
use
of
vulnerable
open-source
dependencies
is
a
known
problem
in
today's
software
development.
Several
vulnerability
scanners
to
detect
known-vulnerable
appeared
the
last
decade,
however,
there
exists
no
case
study
investigating
impact
development
practices,
e.g.,
forking,
patching,
re-bundling,
on
their
performance.
This
paper
studies
(i)
types
modifications
that
may
affect
and
(ii)
performance
scanners.
Through
an
empirical
7,024
Java
projects
developed
at
Язык: Английский
Процитировано
13
Empirical Software Engineering, Год журнала: 2023, Номер 28(2)
Опубликована: Март 1, 2023
Abstract Machine Learning for Source Code () is an active research field in which extensive experimentation needed to discover how best use source code’s richly structured information. With this mind, we introduce : An Extensible Java Dataset Applications, a large-scale, diverse, and high-quality dataset targeted at . Our goal with lower the barrier entry by providing building blocks experiment code models tasks. comes considerable amount of pre-processed information such as metadata, representations (e.g., tokens, ASTs, graphs), several properties metrics, static analysis results) 50,000 projects from dataset, over 1.2 million classes 8 methods. also extensible allowing users add new evaluate tasks on them. Thus, becomes workbench that researchers can novel operating code. To demonstrate utility report results two empirical studies our data, ultimately showing significant work lies ahead design context-aware reason broader network entities software project—the very task designed help with.
Язык: Английский
Процитировано
4