Giuseppe Antonio Di Luna

Fiorella Artuso, Marco Mormando, Giuseppe A. Di Luna et al.

A recent trend in binary code analysis promotes the use of neural solutions based on instruction embedding models. An instruction embedding model is a neural network that transforms sequences of assembly instructions into embedding vectors. If the embedding network is trained such that the translation from code to vectors partially preserves the semantic, the network effectively represents an assembly code model. In this paper we present BinBert, a novel assembly code model. BinBert is built on a transformer pre-trained on a huge dataset of both assembly instruction sequences and symbolic execution information. BinBert can be applied to assembly instructions sequences and it is fine-tunable, i.e. it can be re-trained as part of a neural architecture on task-specific data. Through fine-tuning, BinBert learns how to apply the general knowledge acquired with pre-training to the specific task. We evaluated BinBert on a multi-task benchmark that we specifically designed to test the understanding of assembly code. The benchmark is composed of several tasks, some taken from the literature, and a few novel tasks that we designed, with a mix of intrinsic and downstream tasks. Our results show that BinBert outperforms state-of-the-art models for binary instruction embedding, raising the bar for binary code understanding.

Gianluca Capozzi, Daniele Cono D'Elia, Giuseppe Antonio Di Luna et al.

In recent years, binary analysis gained traction as a fundamental approach to inspect software and guarantee its security. Due to the exponential increase of devices running software, much research is now moving towards new autonomous solutions based on deep learning models, as they have been showing state-of-the-art performances in solving binary analysis problems. One of the hot topics in this context is binary similarity, which consists in determining if two functions in assembly code are compiled from the same source code. However, it is unclear how deep learning models for binary similarity behave in an adversarial context. In this paper, we study the resilience of binary similarity models against adversarial examples, showing that they are susceptible to both targeted and untargeted attacks (w.r.t. similarity goals) performed by black-box and white-box attackers. In more detail, we extensively test three current state-of-the-art solutions for binary similarity against two black-box greedy attacks, including a new technique that we call Spatial Greedy, and one white-box attack in which we repurpose a gradient-guided strategy used in attacks to image classifiers.

Gianluca Capozzi, Tong Tang, Jie Wan et al.

Binary function similarity, which often relies on learning-based algorithms to identify what functions in a pool are most similar to a given query function, is a sought-after topic in different communities, including machine learning, software engineering, and security. Its importance stems from the impact it has in facilitating several crucial tasks, from reverse engineering and malware analysis to automated vulnerability detection. Whereas recent work cast light around performance on this long-studied problem, the research landscape remains largely lackluster in understanding the resiliency of the state-of-the-art machine learning models against adversarial attacks. As security requires to reason about adversaries, in this work we assess the robustness of such models through a simple yet effective black-box greedy attack, which modifies the topology and the content of the control flow of the attacked functions. We demonstrate that this attack is successful in compromising all the models, achieving average attack success rates of 57.06% and 95.81% depending on the problem settings (targeted and untargeted attacks). Our findings are insightful: top performance on clean data does not necessarily relate to top robustness properties, which explicitly highlights performance-robustness trade-offs one should consider when deploying such models, calling for further research.

Giuseppe Antonio Di Luna, Davide Italiano, Luca Massarelli et al.

Despite the advancements in software testing, bugs still plague deployed software and result in crashes in production. When debugging issues -- sometimes caused by "heisenbugs" -- there is the need to interpret core dumps and reproduce the issue offline on the same binary deployed. This requires the entire toolchain (compiler, linker, debugger) to correctly generate and use debug information. Little attention has been devoted to checking that such information is correctly preserved by modern toolchains' optimization stages. This is particularly important as managing debug information in optimized production binaries is non-trivial, often leading to toolchain bugs that may hinder post-deployment debugging efforts. In this paper, we present Debug$^{2}$, a framework to find debug information bugs in modern toolchains. Our framework feeds random source programs to the target toolchain and surgically compares the debugging behavior of their optimized/unoptimized binary variants. Such differential analysis allows Debug$^{2}$ to check invariants at each debugging step and detect bugs from invariant violations. Our invariants are based on the (in)consistency of common debug entities, such as source lines, stack frames, and function arguments. We show that, while simple, this strategy yields powerful cross-toolchain and cross-language invariants, which can pinpoint several bugs in modern toolchains. We have used Debug$^{2}$ to find 23 bugs in the LLVM toolchain (clang/lldb), 8 bugs in the GNU toolchain (GCC/gdb), and 3 in the Rust toolchain (rustc/lldb) -- with 14 bugs already fixed by the developers.

Giuseppe A. Di Luna, Ryuhei Uehara, Giovanni Viglietta et al.

A swarm of anonymous oblivious mobile robots, operating in deterministic Look-Compute-Move cycles, is confined within a circular track. All robots agree on the clockwise direction (chirality), they are activated by an adversarial semi-synchronous scheduler (SSYNCH), and an active robot always reaches the destination point it computes (rigidity). Robots have limited visibility: each robot can see only the points on the circle that have an angular distance strictly smaller than a constant $\vartheta$ from the robot's current location, where $0<\vartheta\leqπ$ (angles are expressed in radians). We study the Gathering problem for such a swarm of robots: that is, all robots are initially in distinct locations on the circle, and their task is to reach the same point on the circle in a finite number of turns, regardless of the way they are activated by the scheduler. Note that, due to the anonymity of the robots, this task is impossible if the initial configuration is rotationally symmetric; hence, we have to make the assumption that the initial configuration be rotationally asymmetric. We prove that, if $\vartheta=π$ (i.e., each robot can see the entire circle except its antipodal point), there is a distributed algorithm that solves the Gathering problem for swarms of any size. By contrast, we also prove that, if $\vartheta\leq π/2$, no distributed algorithm solves the Gathering problem, regardless of the size of the swarm, even under the assumption that the initial configuration is rotationally asymmetric and the visibility graph of the robots is connected. The latter impossibility result relies on a probabilistic technique based on random perturbations, which is novel in the context of anonymous mobile robots. Such a technique is of independent interest, and immediately applies to other Pattern-Formation problems.

Fiorella Artuso, Giuseppe Antonio Di Luna, Luca Massarelli et al.

In this paper we investigate the problem of automatically naming pieces of assembly code. Where by naming we mean assigning to an assembly function a string of words that would likely be assigned by a human reverse engineer. We formally and precisely define the framework in which our investigation takes place. That is we define the problem, we provide reasonable justifications for the choices that we made for the design of training and the tests. We performed an analysis on a large real-world corpora constituted by nearly 9 millions of functions taken from more than 22k softwares. In such framework we test baselines coming from the field of Natural Language Processing (e.g., Seq2Seq networks and Transformer). Interestingly, our evaluation shows promising results beating the state-of-the-art and reaching good performance. We investigate the applicability of tine-tuning (i.e., taking a model already trained on a large generic corpora and retraining it for a specific task). Such technique is popular and well-known in the NLP field. Our results confirm that fine-tuning is effective even when neural networks are applied to binaries. We show that a model, pre-trained on the aforementioned corpora, when fine-tuned has higher performances on specific domains (such as predicting names in system utilites, malware, etc).

Shantanu Das, Giuseppe A. Di Luna, Paola Flocchini et al.

We consider a distributed system of n identical mobile robots operating in the two dimensional Euclidian plane. As in the previous studies, we consider the robots to be anonymous, oblivious, dis-oriented, and without any communication capabilities, operating based on the Look-Compute-Move model where the next location of a robot depends only on its view of the current configuration. Even in this seemingly weak model, most formation problems which require constructing specific configurations, can be solved quite easily when the robots are fully synchronized with each other. In this paper we introduce and study a new class of problems which, unlike the formation problems so far, cannot always be solved even in the fully synchronous model with atomic and rigid moves. This class of problems requires the robots to permute their locations in the plane. In particular, we are interested in implementing two special types of permutations -- permutations without any fixed points and permutations of order $n$. The former (called MOVE-ALL) requires each robot to visit at least two of the initial locations, while the latter (called VISIT-ALL) requires every robot to visit each of the initial locations in a periodic manner. We provide a characterization of the solvability of these problems, showing the main challenges in solving this class of problems for mobile robots. We also provide algorithms for the feasible cases, in particular distinguishing between one-step algorithms (where each configuration must be a permutation of the original configuration) and multi-step algorithms (which allow intermediate configurations). These results open a new research direction in mobile distributed robotics which has not been investigated before.

Luca Massarelli, Giuseppe Antonio Di Luna, Fabio Petroni et al.

The binary similarity problem consists in determining if two functions are similar by only considering their compiled form. Advanced techniques for binary similarity recently gained momentum as they can be applied in several fields, such as copyright disputes, malware analysis, vulnerability detection, etc., and thus have an immediate practical impact. Current solutions compare functions by first transforming their binary code in multi-dimensional vector representations (embeddings), and then comparing vectors through simple and efficient geometric operations. However, embeddings are usually derived from binary code using manual feature extraction, that may fail in considering important function characteristics, or may consider features that are not important for the binary similarity problem. In this paper we propose SAFE, a novel architecture for the embedding of functions based on a self-attentive neural network. SAFE works directly on disassembled binary functions, does not require manual feature extraction, is computationally more efficient than existing solutions (i.e., it does not incur in the computational overhead of building or manipulating control flow graphs), and is more general as it works on stripped binaries and on multiple architectures. We report the results from a quantitative and qualitative analysis that show how SAFE provides a noticeable performance improvement with respect to previous solutions. Furthermore, we show how clusters of our embedding vectors are closely related to the semantic of the implemented algorithms, paving the way for further interesting applications (e.g. semantic-based binary function search).

Roberto Baldoni, Giuseppe Antonio Di Luna, Luca Massarelli et al.

In this paper we consider the binary similarity problem that consists in determining if two binary functions are similar only considering their compiled form. This problem is know to be crucial in several application scenarios, such as copyright disputes, malware analysis, vulnerability detection, etc. The current state-of-the-art solutions in this field work by creating an embedding model that maps binary functions into vectors in $\mathbb{R}^{n}$. Such embedding model captures syntactic and semantic similarity between binaries, i.e., similar binary functions are mapped to points that are close in the vector space. This strategy has many advantages, one of them is the possibility to precompute embeddings of several binary functions, and then compare them with simple geometric operations (e.g., dot product). In [32] functions are first transformed in Annotated Control Flow Graphs (ACFGs) constituted by manually engineered features and then graphs are embedded into vectors using a deep neural network architecture. In this paper we propose and test several ways to compute annotated control flow graphs that use unsupervised approaches for feature learning, without incurring a human bias. Our methods are inspired after techniques used in the natural language processing community (e.g., we use word2vec to encode assembly instructions). We show that our approach is indeed successful, and it leads to better performance than previous state-of-the-art solutions. Furthermore, we report on a qualitative analysis of functions embeddings. We found interesting cases in which embeddings are clustered according to the semantic of the original binary function.

Giuseppe A. Di Luna, Paola Flocchini, Nicola Santoro et al.

In this paper we investigate the computational power of a set of mobile robots with limited visibility. At each iteration, a robot takes a snapshot of its surroundings, uses the snapshot to compute a destination point, and it moves toward its destination. Each robot is punctiform and memoryless, it operates in $\mathbb{R}^m$, it has a local reference system independent of the other robots' ones, and is activated asynchronously by an adversarial scheduler. Moreover, robots are non-rigid, in that they may be stopped by the scheduler at each move before reaching their destination (but are guaranteed to travel at least a fixed unknown distance before being stopped). We show that despite these strong limitations, it is possible to arrange $3m+3k$ of these weak entities in $\mathbb{R}^m$ to simulate the behavior of a stronger robot that is rigid (i.e., it always reaches its destination) and is endowed with $k$ registers of persistent memory, each of which can store a real number. We call this arrangement a TuringMobile. In its simplest form, a TuringMobile consisting of only three robots can travel in the plane and store and update a single real number. We also prove that this task is impossible with fewer than three robots. Among the applications of the TuringMobile, we focused on Near-Gathering (all robots have to gather in a small-enough disk) and Pattern Formation (of which Gathering is a special case) with limited visibility. Interestingly, our investigation implies that both problems are solvable in Euclidean spaces of any dimension, even if the visibility graph of the robots is initially disconnected, provided that a small amount of these robots are arranged to form a TuringMobile. In the special case of the plane, a basic TuringMobile of only three robots is sufficient.

Giuseppe A. Di Luna, Paola Flocchini, Nicola Santoro et al.

Shape formation is a basic distributed problem for systems of computational mobile entities. Intensively studied for systems of autonomous mobile robots, it has recently been investigated in the realm of programmable matter. Namely, it has been studied in the geometric Amoebot model, where the anonymous entities, called particles, operate on a hexagonal tessellation of the plane and have limited computational power (they have constant memory), strictly local interaction and communication capabilities (only with particles in neighboring nodes of the grid), and limited motorial capabilities (from a grid node to an empty neighboring node); their activation is controlled by an adversarial scheduler. Recent investigations have shown how, starting from a well-structured configuration in which the particles form a (not necessarily complete) triangle, the particles can form a large class of shapes. This result has been established under several assumptions: agreement on the clockwise direction (i.e., chirality), a sequential activation schedule, and randomization (i.e., particles can flip coins). In this paper we provide a characterization of which shapes can be formed deterministically starting from any simply connected initial configuration of $n$ particles. As a byproduct, if randomization is allowed, then any input shape can be formed from any initial (simply connected) shape by our algorithm, provided that $n$ is large enough. Our algorithm works without chirality, proving that chirality is computationally irrelevant for shape formation. Furthermore, it works under a strong adversarial scheduler, not necessarily sequential. We also consider the complexity of shape formation in terms of both the number of rounds and of moves performed by the particles. We prove that our solution has a complexity of $O(n^2)$ rounds and moves: this number of moves is also asymptotically optimal.

Giuseppe A. Di Luna, Paola Flocchini, Nicola Santoro et al.

The Meeting problem for $k\geq 2$ searchers in a polygon $P$ (possibly with holes) consists in making the searchers move within $P$, according to a distributed algorithm, in such a way that at least two of them eventually come to see each other, regardless of their initial positions. The polygon is initially unknown to the searchers, and its edges obstruct both movement and vision. Depending on the shape of $P$, we minimize the number of searchers $k$ for which the Meeting problem is solvable. Specifically, if $P$ has a rotational symmetry of order $σ$ (where $σ=1$ corresponds to no rotational symmetry), we prove that $k=σ+1$ searchers are sufficient, and the bound is tight. Furthermore, we give an improved algorithm that optimally solves the Meeting problem with $k=2$ searchers in all polygons whose barycenter is not in a hole (which includes the polygons with no holes). Our algorithms can be implemented in a variety of standard models of mobile robots operating in Look-Compute-Move cycles. For instance, if the searchers have memory but are anonymous, asynchronous, and have no agreement on a coordinate system or a notion of clockwise direction, then our algorithms work even if the initial memory contents of the searchers are arbitrary and possibly misleading. Moreover, oblivious searchers can execute our algorithms as well, encoding information by carefully positioning themselves within the polygon. This code is computable with basic arithmetic operations, and each searcher can geometrically construct its own destination point at each cycle using only a compass. We stress that such memoryless searchers may be located anywhere in the polygon when the execution begins, and hence the information they initially encode is arbitrary. Our algorithms use a self-stabilizing map construction subroutine which is of independent interest.

G. A. Di Luna, P. Flocchini, S. Gan Chaudhuri et al.

Consider a finite set of identical computational entities that can move freely in the Euclidean plane operating in Look-Compute-Move cycles. Let p(t) denote the location of entity p at time t; entity p can see entity q at time t if at that time no other entity lies in the line segment p(t)q(t). We consider the basic problem called Mutual Visibility: starting from arbitrary distinct locations, within finite time the entities must reach, without collisions, a configuration where they all see each other. This problem must be solved by each entity autonomously executing the same algorithm. We study this problem in the "luminous robots" model; in this generalization of the standard model of oblivious robots, each entity, called "robot", has an externally visible persistent light which can assume colors from a fixed set. The case where the number of colors is c=1 corresponds to the classical model without lights. In this paper we investigate under what conditions luminous robots can solve Mutual Visibility without collisions and at what cost (i.e., with how many colors). We establish a spectrum of results, depending on the power of the adversary, on the number c of colors, and on the a-priori knowledge the robots have about the system. Among such results, we prove that Mutual Visibility can always be solved without collisions in SSynch with c=2 colors and in ASynch with c=3 colors. If an adversary can interrupt and stop a robot moving to its computed destination, Mutual Visibility is still always solvable without collisions in SSynch with c=3 colors, and, if the robots agree on the direction of one axis, also in ASynch. All the results are obtained constructively by means of novel protocols. As a byproduct of our solutions, we provide the first obstructed-visibility solutions to two classical problems for oblivious robots: Collision-less Convergence to a point and Circle Formation.

G. A. Di Luna, P. Flocchini, S. Gan Chaudhuri et al.

Robots with lights is a model of autonomous mobile computational entities operating in the plane in Look-Compute-Move cycles: each agent has an externally visible light which can assume colors from a fixed set; the lights are persistent (i.e., the color is not erased at the end of a cycle), but otherwise the agents are oblivious. The investigation of computability in this model, initially suggested by Peleg, is under way, and several results have been recently established. In these investigations, however, an agent is assumed to be capable to see through another agent. In this paper we start the study of computing when visibility is obstructable, and investigate the most basic problem for this setting, Complete Visibility: The agents must reach within finite time a configuration where they can all see each other and terminate. We do not make any assumption on a-priori knowledge of the number of agents, on rigidity of movements nor on chirality. The local coordinate system of an agent may change at each activation. Also, by definition of lights, an agent can communicate and remember only a constant number of bits in each cycle. In spite of these weak conditions, we prove that Complete Visibility is always solvable, even in the asynchronous setting, without collisions and using a small constant number of colors. The proof is constructive. We also show how to extend our protocol for Complete Visibility so that, with the same number of colors, the agents solve the (non-uniform) Circle Formation problem with obstructed visibility.