Memory Protection Cache for Secure Peer-to-Peer Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current memory management systems in computer systems face bottlenecks and increased latency due to the need for host-reflected/bounced memory requests for memory protection checks, especially in disaggregated cloud platforms.

Innovation Solution

Implementing a distributed secure address translation service (ATS) architecture with a memory protection cache (MPC) closer to device memory, allowing for secure direct peer-to-peer memory access without host intervention, and synchronizing MPC information with a host's memory protection unit.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If host-reflected memory requests are used for memory protection checks, then security is maintained, but latency increases and performance deteriorates

Engineering Contradiction:
Improvememory protectionVSAvoidlatency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent introduces a Memory Protection Cache (MPC) as an intermediary component between devices and the host memory protection unit. The MPC stores memory protection information locally at the device level, enabling direct peer-to-peer memory access without requiring host reflection. This intermediary structure allows devices to perform protection checks independently, eliminating the latency associated with host-reflected requests while maintaining security through localized protection information.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If host-reflected memory requests are used for memory protection checks, then security is maintained, but device complexity increases

Engineering Contradiction:
Improvememory protectionVSAvoidmemory access path
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the memory protection function from the centralized host and places it at the device level through the Memory Protection Cache. By taking out the protection check capability from the host-reflection path and embedding it directly in devices, the system eliminates the need for complex host-mediated access paths. Devices can independently manage their own memory protection, simplifying the overall access architecture while maintaining robust security.

Inventive Principle:
Principle #2Taking out (Extraction)

3Productivity

If direct peer-to-peer memory access is enabled, then performance improves, but security control is weakened

Engineering Contradiction:
Improvememory access performanceVSAvoidsecurity control
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent enables devices to perform memory protection checks independently through locally stored Memory Protection Cache information. Each device serves its own security needs by using cached protection data to validate peer-to-peer access requests without requiring host intervention. This self-service approach allows high-performance direct memory access while maintaining security control, as devices autonomously verify access rights using pre-loaded protection information.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS12292840B2Secure direct peer-to-peer memory access requests between devices
Publication Date: 2025.05.06 INTEL CORP
  • US12292840B2 patent drawing
  • US12292840B2 patent drawing
  • US12292840B2 patent drawing

AI summary

An embodiment of an integrated circuit comprises circuitry to store memory protection information for a non-host memory in a memory protection cache, and perform one or more memory protection checks on a translated access request for the non-host memory based on the stored memory protection information. Other embodiments are disclosed and claimed.