Reconsidering custom memory allocation

Berger, Emery D.; Zorn, Benjamin G.; McKinley, Kathryn S.

doi:10.1145/582419.582421

Cited by 104 publications

(99 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To handle synchronization as described in Section 3.3 and address space layout differences as described in Section 3.7, someone has to implement the appropriate interposers. Besides highly application-dependent use of pointer values to index data structures, many real-world applications use custom synchronization mechanisms [42] as well as custom memory allocators [7] besides the standard glibc and pthread primitives (despite the cited literature demonstrating how bad that customization practice is). In all these cases, the application developers themselves are responsible (1) for ensuring that interposers can handle all cases correctly, and (2) for providing the interposers.…”

Section: Experimental Evaluationmentioning

confidence: 99%

GHUMVEE: Efficient, Effective, and Flexible Replication

Volckaert

Sutter

Baets

et al. 2013

Foundations and Practice of Security

View full text Add to dashboard Cite

Abstract. We present GHUMVEE, a multi-variant execution engine for software intrusion detection. GHUMVEE transparently executes and monitors diversified replicae of processes to thwart attacks relying on a predictable, single data layout. Unlike existing tools, GHUMVEE's interventions in the process' execution are not limited to system call invocations. Because of that design decision, GHUMVEE can handle complex, multi-threaded real-life programs that display non-deterministic behavior as a result of non-deterministic thread scheduling and as a result of pointer-value dependent behavior. This capability is demonstrated on GUI programs from the Gnome and KDE desktop environments.

show abstract

Section: Experimental Evaluationmentioning

confidence: 99%

GHUMVEE: Efficient, Effective, and Flexible Replication

Volckaert

Sutter

Baets

et al. 2013

Foundations and Practice of Security

View full text Add to dashboard Cite

show abstract

“…However, Berger et al suggest in [1] that region based approaches suffer from increased memory consumption due to retarded deallocation if the programmer is unwilling to adjust his or her programming style to suit the region approach and they propose a more generalised version of regions.…”

Section: Appaux(ythis):voidmentioning

confidence: 99%

Type-Based Amortised Heap-Space Analysis

Hofmann

Jost

2006

Programming Languages and Systems

View full text Add to dashboard Cite

Abstract. We present a type system for a compile-time analysis of heap-space requirements of Java style object-oriented programs with explicit deallocation.Our system is based on an amortised complexity analysis: the data is arbitrarily assigned a potential related to its size and layout; allocations must be "payed for" from this potential. The potential of each input then furnishes an upper bound on the heap space usage for the computation on this input.We successfully treat inheritance, downcast, update and aliasing. Example applications for the analysis include destination-passing style and doubly-linked lists.Type inference is explicitly not included; the contribution lies in the system itself and the nontrivial soundness theorem. This extended abstract elides most technical lemmas and proofs, even nontrivial ones, due to space limitations. A full version is available at the authors' web pages.

show abstract

“…We designed a custom memory manager for the single threaded version of TRIP, called rlalloc in this paper, in order to reduce the overhead in the allocation of elements of lists. The design of an efficient custom memory manager is an hard task as shown in the Berger's review [13]. Our memory manager splits a large memory chunk requested from the operating system in individual blocks of the size of the element of the recursive list representation.…”

Section: Memory Management For a Recursive List Representationmentioning

confidence: 99%

Development of TRIP: Fast Sparse Multivariate Polynomial Multiplication Using Burst Tries

Gastineau

Laskar

2006

Computational Science – ICCS 2006

View full text Add to dashboard Cite

Abstract. Flat vector representation of sparse multivariate polynomials is introduced in the computer algebra system TRIP with specific care to the cache memory. Burst tries are considered as an intermediate storage during the sparse multivariate polynomial multiplication by paying attention to the memory allocations. Timing and memory consumption are examined and compared with other recursive representations and other computer algebra systems.

show abstract

Reconsidering custom memory allocation

Cited by 104 publications

References 22 publications

GHUMVEE: Efficient, Effective, and Flexible Replication

GHUMVEE: Efficient, Effective, and Flexible Replication

Type-Based Amortised Heap-Space Analysis

Development of TRIP: Fast Sparse Multivariate Polynomial Multiplication Using Burst Tries

Contact Info

Product

Resources

About