Virtual Partition Machine Check Event Containment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine check events, such as machine check abort (MCA) errors, can cause system-wide disruptions and require costly downtime for error analysis and correction, affecting the entire hard partition and operating system.
Innovation Solution
The system firmware isolates and contains MCA errors within a local virtual partition, performing error processing and clean-up operations without disrupting other virtual partitions, allowing for directed reboots and minimizing system-wide impacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine check events are processed at the system level, then error handling completeness is improved, but system availability deteriorates due to mandatory system-wide shutdowns
Solution Approach 1:
The patent segments the hard partition into multiple virtual partitions (vPartitions), each with its own error handling mechanism. When an MCA error occurs, only the affected vPartition is reset, not the entire system. This segmentation allows independent error containment while maintaining system-wide availability.
Solution Approach 2:
The patent implements local error handling by assigning dedicated error handling resources and mechanisms to each virtual partition. Each vPartition has its own error state management, allowing localized error processing without affecting other partitions' operation.
2Measurement precision
If system-wide error processing is implemented, then error analysis thoroughness is improved, but downtime increases due to mandatory system shutdowns
Solution Approach 1:
The system divides the hard partition into independent virtual partitions, each capable of processing errors locally. This allows thorough error analysis within each vPartition without requiring system-wide shutdowns, as only the affected partition needs to be reset.
Solution Approach 2:
The patent introduces a virtual partition monitor as an intermediary layer between the error handling mechanism and the operating system. This monitor manages error states at the vPartition level, enabling detailed error analysis and controlled resets without impacting the entire system.
3Loss of information
If global machine check broadcast is used, then error notification completeness is improved, but system disruption increases
Solution Approach 1:
The patent replaces the global broadcast mechanism with partition-specific error notification. Each virtual partition maintains its own error state and notification mechanism, allowing complete error notification within the affected partition without disrupting other partitions.
Solution Approach 2:
The system implements local error notification by confining MCA error broadcasts to only the affected virtual partition. Each vPartition has dedicated error handling resources that process and notify about errors locally, avoiding system-wide disruption.
Data Source
AI summary
Embodiments include methods, apparatus, and systems for containing machine check events in a virtual partition. One embodiment is a method of software execution. The method divides a hard partition into first and second virtual partitions and attempts to correct an error in a firmware layer of the first virtual partition. If the error is not correctable, then the method reboots the first virtual partition without disrupting hardware resources in the second virtual partition.