NVall: A Crash-Resistant and Kernel-Compatible Memory Allocator for NVRAM

Nguyen DT, Wiedemann O, Nolte J, Schröder-Preikschat W (2023)


Publication Language: English

Publication Type: Conference contribution, Conference Contribution

Publication year: 2023

Publisher: Gesellschaft für Informatik e.V.

Series: Tagungsband des FG-BS Herbsttreffens 2023

Pages Range: 1-7

Event location: Bamberg DE

URI: https://dl.gi.de/items/8d0686f6-a88e-4e96-af34-2743d49b99a4

DOI: 10.18420/fgbs2023h-02

Abstract

Byte-addressable non-volatile memory is essentially persistent, but slower main memory that needs to be managed accordingly. Typical memory allocators for volatile memory are highly efficient today, but usually never had to be designed to keep their state in main memory consistent at all times against the background of system crashes. In this paper we present NVall, a crash-resistant kernel-level memory allocator for non-volatile RAM (NVRAM). The allocator works in a transactional manner, uses existing volatile memory to improve the performance of normal operation and is able to recover its volatile state from persistent data after a system crash. We implemented the allocator for the FreeBSD kernel and compare its performance against the standard (non-crash-resistant) in-kernel allocator of FreeBSD.

Authors with CRIS profile

Related research project(s)

Involved external institutions

How to cite

APA:

Nguyen, D.T., Wiedemann, O., Nolte, J., & Schröder-Preikschat, W. (2023). NVall: A Crash-Resistant and Kernel-Compatible Memory Allocator for NVRAM. In Proceedings of the FG-BS Herbsttreffens 2023 (pp. 1-7). Bamberg, DE: Gesellschaft für Informatik e.V..

MLA:

Nguyen, Dustin Tien, et al. "NVall: A Crash-Resistant and Kernel-Compatible Memory Allocator for NVRAM." Proceedings of the FG-BS Herbsttreffens 2023, Bamberg Gesellschaft für Informatik e.V., 2023. 1-7.

BibTeX: Download