UltraFuzz: Towards Resource-Saving in Distributed Fuzzing

Abstract

Recent research has sought to improve fuzzing performance via parallel computing. However, researchers focus on improving efficiency while ignoring the increasing cost of testing resources. Parallel fuzzing in the distributed environment amplifies the resource-wasting problem caused by the random nature of fuzzing. In the parallel mode, owing to the lack of an appropriate task dispatching scheme and timely fuzzing status synchronization among different fuzzing instances, task conflicts and workload imbalance occur, making the resourcewasting problem severe. In this paper, we design UltraFuzz, a fuzzer for resource-saving in distributed fuzzing. Based on centralized dynamic scheduling, UltraFuzz can dispatch tasks and schedule power globally and reasonably to avoid resourcewasting. Besides, UltraFuzz can elastically allocate computing power for fuzzing and seed evaluation, thereby avoiding the potential bottleneck of seed evaluation that blocks the fuzzing process. UltraFuzz was evaluated using real-world programs, and the results show that with the same testing resource, UltraFuzz outperforms state-of-the-art tools, such as AFL, AFL-P, PAFL, and EnFuzz. Most importantly, the experiment reveals certain results that seem counter-intuitive, namely that parallel fuzzing can achieve "super-linear acceleration" when compared with single-core fuzzing. We conduct additional experiments to reveal the deep reasons behind this phenomenon and dig deep into the inherent advantages of parallel fuzzing over serial fuzzing, including the global optimization of seed energy scheduling and the escape of local optimal seed. Additionally, 24 real-world vulnerabilities were discovered using UltraFuzz.

TAEF: A Task Allocation-Based Ensemble Fuzzing Framework for Optimizing the Advantages of Heterogeneous Fuzzers

“…The parallelization of fuzzers is a primary way to enhance the efficiency of fuzz testing, which can be classified into two major categories: parallel fuzzing with multiple instances of the same fuzzer [16,17,32,33] and ensemble fuzzing with paralleled instances of different fuzzers [15,34].…”

“…Global seeds scheduling: Both P-Fuzz [32] and UltraFuzz [33] are designed based on this type of approach. They implement a manager called the main node on top of the fuzzer and centralize the management and selection of seeds in the main node.…”

Particle Swarm Optimization with Enhanced Neighborhood Search Algorithm and Probabilistic Crossover Genetic Algorithm for Task Scheduling in Cloud Computing

“…When the tasks are submitted by the cloud user, the master processor collects them and distributes them to the slave nodes, where every node controls its own queue (X. Zhou et al, 2020). In contrast, distributed scheduling eliminates the need for a central processing node since local schedulers are in charge of managing incoming requests and keeping track of the state of all resources by regularly exchanging updates with them (Patel & Bhoi, 2013).…”