У нас вы можете посмотреть бесплатно Efficient system-level support for CHERI Capabilities | Tariq Kurd / Mark Hill – Codasip или скачать в максимальном доступном качестве, видео которое было загружено на ютуб. Для загрузки выберите вариант из формы ниже:
Если кнопки скачивания не
загрузились
НАЖМИТЕ ЗДЕСЬ или обновите страницу
Если возникают проблемы со скачиванием видео, пожалуйста напишите в поддержку по адресу внизу
страницы.
Спасибо за использование сервиса ClipSaver.ru
Abstract: "To ensure that capabilities cannot be forged or manipulated by an adversary, CHERI requires that the system implements an out-of-band tag bit for every element of memory storage that can hold a capability. This tag, and the data it protects, must appear by all observers in the system to be updated atomically. When the tag can be co-located with the data this is relatively straightforward, when it cannot, due to integration with non-tag aware systems or system IPs, the tags must be stored elsewhere, and the situation is much more challenging. The primary challenges are: • Preserving the atomicity of tags and data stored in different locations in the presence of multiple bus initiators and downstream bus fabric and IPs, such as Dynamic Memory Controllers, that are aggressively reordering transactions. • Limiting the impact on the performance of the memory system, both in terms of bandwidth and latency, to a level which ensures performance is comparable to non-CHERI implementations. • Maximising design re-use. The design must support a wide-range of scenarios, such as integrating a CHERI-based Secure subsystem into a SoC which is not CHERI-aware, or a full CHERI-aware system, where every memory transaction in the SoC must pass through it. The presentation will describe the challenges associated with efficient handling of tags at a system level and how Codasip has met them in the development of the first commercially available System IP to solve this issue: the Capability Management Unit (CMU). Although the CMU is a universal IP that could be deployed to support any CHERI compliant core, we are currently demonstrating it as a companion to the X730 CHERI-enhanced dual-issue RVA22 compliant core. This work builds on and acknowledges the research previously done by the University of Cambridge on efficient handling of tagged memory." Filmed at the CHERI Blossoms 2025 conference on 2nd April 2025. You can download the slides for the talk and find out more information on the conference website at: https://cheri-alliance.org/events/che... Interested in joining the CHERI Alliance? Find out more at: https://cheri-alliance.org/memberships/