Row Hammer Mitigation via Activation-Biased Caching Policy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional row hammer detection techniques in DRAM devices are probabilistic and cannot guarantee perfect prevention of data corruption, making them vulnerable to malicious attacks that cause data corruption due to frequent row activations.
Innovation Solution
A row hammer detector integrated with a cache memory that implements a row activation-biased caching policy, tracking row activations and issuing refresh commands to mitigate row hammer attacks by adjusting caching policies based on detected thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional probabilistic row hammer detection techniques are used, then device complexity is reduced, but reliability of row hammer mitigation deteriorates
Solution Approach 1:
The system performs preliminary tracking of row activation counts before row hammer events occur. Counters are maintained for each row to monitor activation frequency in advance, allowing the system to detect potential row hammer conditions before data corruption happens, rather than reacting after the fact.
Solution Approach 2:
The system implements feedback mechanisms where row activation counts are continuously monitored and fed back to the control logic. When a row exceeds the threshold, this feedback triggers mitigation actions such as refreshing victim rows or adjusting cache policies, creating a closed-loop system that adapts to detected conditions.
2Reliability
If perfect tracking of row hammer events is implemented, then reliability of data protection is improved, but use of energy increases
Solution Approach 1:
The system applies different tracking intensities to different rows based on their characteristics. High-priority rows that contain critical data receive more intensive monitoring and protection, while low-priority rows use simpler tracking. This localized approach to quality of protection reduces overall energy consumption while maintaining reliability for critical data.
Solution Approach 2:
The system dynamically changes tracking parameters such as counter update frequency and threshold values based on system conditions. During normal operation, tracking may be less intensive, but when row hammer patterns are detected, the system increases tracking granularity and adjusts parameters to provide stronger protection, optimizing energy usage across different operational states.
3Reliability
If aggressive row hammer mitigation is applied, then reliability against attacks is improved, but productivity of memory operations deteriorates
Solution Approach 1:
The system applies partial mitigation actions rather than always executing full countermeasures. When a row approaches the threshold, the system may take preventive actions on only the most vulnerable victim rows or apply less intrusive cache policies, rather than shutting down all memory operations. This partial action maintains attack resistance while preserving memory throughput.
Solution Approach 2:
The mitigation strategy dynamically adjusts based on the severity and pattern of detected row hammer activity. The system can transition between different operational modes from normal operation to enhanced protection, adjusting cache replacement policies, refresh frequencies, and access patterns in real-time based on current threat levels, thereby maintaining productivity while providing adaptive security.
Data Source
AI summary
A system comprising a row hammer mitigation circuitry and a cache memory that collaborate to mitigate row hammer attacks on a memory media device is described. The cache memory biases cache policy based on row access count information maintained by the row hammer mitigation circuit. The row hammer mitigation circuitry may be implemented in a memory controller. The memory media device may be DRAM. Corresponding methods are also described.