arXiv subjects

Petr Tůma

Publications and source records attributed to Petr Tůma.

5 recordsLinked to original sources

arXiv2026

MapReplay: Trace-Driven Benchmark Generation for Java HashMap

Hash-based maps, particularly java.util.HashMap, are pervasive in Java applications and the JVM, making their performance critical. Evaluating optimizations is challenging because performance depends on factors such as operation patterns, key distributions, and resizing behavior. Microbenchmarks are fast and repeatable but often oversimplify workloads, failing to capture the realistic usage patterns. Application benchmarks (e.g., DaCapo, Renaissance) provide realistic usages but are more expensive to run, prone to variability, and dominated by non-HashMap computations, making map-related performance changes difficult to observe. To address this challenge, we propose MapReplay, a benchmarking methodology that combines the realism of application benchmarks with the efficiency of microbenchmarks. MapReplay traces HashMap API usages generating a replay workload that reproduces the same operation sequence while faithfully reconstructing internal map states. This enables realistic and efficient evaluation of alternative implementations under realistic usage patterns. Applying MapReplay to DaCapo-Chopin and Renaissance, the resulting suite, MapReplayBench, reproduces application-level performance trends while reducing experimentation time and revealing insights difficult to obtain from full benchmarks.

cs.PL↗

arXiv2020

Database Traffic Interception for Graybox Detection of Stored and Context-Sensitive XSS

XSS is a security vulnerability that permits injecting malicious code into the client side of a web application. In the simplest situations, XSS vulnerabilities arise when a web application includes the user input in the web output without due sanitization. Such simple XSS vulnerabilities can be detected fairly reliably with blackbox scanners, which inject malicious payload into sensitive parts of HTTP requests and look for the reflected values in the web output. Contemporary blackbox scanners are not effective against stored XSS vulnerabilities, where the malicious payload in an HTTP response originates from the database storage of the web application, rather than from the associated HTTP request. Similarly, many blackbox scanners do not systematically handle context-sensitive XSS vulnerabilities, where the user input is included in the web output after a transformation that prevents the scanner from recognizing the original value, but does not sanitize the value sufficiently. Among the combination of two basic data sources (stored vs reflected) and two basic vulnerability patterns (context sensitive vs not so), only one is therefore tested systematically by state-of-the-art blackbox scanners. Our work focuses on systematic coverage of the three remaining combinations. We present a graybox mechanism that extends a general purpose database to cooperate with our XSS scanner, reporting and injecting the test inputs at the boundary between the database and the web application. Furthermore, we design a mechanism for identifying the injected inputs in the web output even after encoding by the web application, and check whether the encoding sanitizes the injected inputs correctly in the respective browser context. We evaluate our approach on eight mature and technologically diverse web applications, discovering previously unknown and exploitable XSS flaws in each of those applications.

cs.CR↗

arXiv2020

DjangoChecker: Applying Extended Taint Tracking and Server Side Parsing for Detection of Context-Sensitive XSS Flaws

Cross-site scripting (XSS) flaws are a class of security flaws that permit the injection of malicious code into a web application. In simple situations, these flaws can be caused by missing input sanitizations. Sometimes, however, all application inputs are sanitized, but the sanitizations are not appropriate for the browser contexts of the sanitized values. Using an incorrect sanitizer can make the application look protected, when it is in fact vulnerable as if no sanitization was used, creating a context-sensitive XSS flaw. To discover context-sensitive XSS flaws, we introduce DjangoChecker. DjangoChecker combines extended dynamic taint tracking with a model browser for context analysis. We demonstrate the practical application of DjangoChecker on eight mature web applications based on Django, discovering previously unknown flaws in seven of the eight applications, including highly severe flaws that allow arbitrary JavaScript execution in the seven flawed applications.

cs.CR↗

arXiv2020

Duet Benchmarking: Improving Measurement Accuracy in the Cloud

We investigate the duet measurement procedure, which helps improve the accuracy of performance comparison experiments conducted on shared machines by executing the measured artifacts in parallel and evaluating their relative performance together, rather than individually. Specifically, we analyze the behavior of the procedure in multiple cloud environments and use experimental evidence to answer multiple research questions concerning the assumption underlying the procedure. We demonstrate improvements in accuracy ranging from 2.3x to 12.5x (5.03x on average) for the tested ScalaBench (and DaCapo) workloads, and from 23.8x to 82.4x (37.4x on average) for the SPEC CPU 2017 workloads.

cs.PF↗

arXiv2019

On Evaluating the Renaissance Benchmarking Suite: Variety, Performance, and Complexity

The recently proposed Renaissance suite is composed of modern, real-world, concurrent, and object-oriented workloads that exercise various concurrency primitives of the JVM. Renaissance was used to compare performance of two stateof-the-art, production-quality JIT compilers (HotSpot C2 and Graal), and to show that the performance differences are more significant than on existing suites such as DaCapo and SPECjvm2008. In this technical report, we give an overview of the experimental setup that we used to assess the variety and complexity of the Renaissance suite, as well as its amenability to new compiler optimizations. We then present the obtained measurements in detail.

cs.PL↗