Ring Bus Buffer Threshold Adjustment for Deadlock Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In information processing systems with multiple nodes connected via a ring bus, deadlocks occur when a node becomes busy, causing requests to be piled up and leading to system failures.
Innovation Solution
The implementation of a buffer system with adjustable thresholds and a deadlock controller that suppresses request transmissions when a deadlock is detected, allowing for dynamic adjustment of buffer capacity and request prioritization to prevent deadlock states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If requests are stored in queues when a node becomes busy, then request buffering capacity is improved, but deadlock occurs when requests pile up in multiple nodes
Solution Approach 1:
The buffer threshold is dynamically adjusted between a first threshold (for normal operation) and a second threshold (for deadlock prevention). When the number of stored requests exceeds the first threshold, the buffer threshold is increased to the second threshold to allow more requests to be stored. When the number of stored requests falls below the second threshold, it is decreased back to the first threshold. This dynamic adjustment prevents deadlock while maintaining efficient request buffering.
Solution Approach 2:
The invention changes the parameter of buffer capacity (number of storable requests) based on system state. By switching between two threshold values (first threshold and second threshold), the system adapts its buffering capacity to prevent deadlock conditions while maintaining high utilization during normal operation.
2Quantity of substance
If buffer capacity is increased to prevent request loss, then request storage capability is improved, but system complexity increases due to threshold management
Solution Approach 1:
The buffer management is segmented into two distinct threshold levels: a first threshold for normal operation and a second threshold for deadlock prevention. This segmentation allows the system to operate efficiently under normal conditions while having a predefined mechanism to prevent deadlock, reducing the complexity of continuous monitoring and adjustment.
Solution Approach 2:
The system employs feedback control by monitoring the number of stored requests and automatically adjusting the buffer threshold based on this feedback. When the count exceeds the first threshold, the threshold is raised to the second threshold; when it falls below the second threshold, the threshold is lowered to the first threshold. This automated feedback mechanism simplifies threshold management compared to complex predictive algorithms.
Data Source
AI summary
A first information-processing-apparatus includes a buffer to have entries to store a first request-data received and transmitted to a second information-processing-apparatus, a memory, and a processor coupled to the memory and configured to transmit, to the second information-processing-apparatus, the first request-data and a second request-data to be transmitted to second or third information-processing-apparatus, when a state where a number of entries in which data is stored in the buffer is equal to or larger than a first threshold is continued for a time longer than a first time, and a state where transmissions of first and second request-data to the second information-processing-apparatus are suppressed is continued for a time longer than a second time, change a number of entries usable in the buffer into a second threshold larger than the first threshold, and when the number of entries usable is the second threshold, suppress a transmission of the second request-data.