Host Bridge Extended Error Handling Logic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In logical partitioned data processing systems, when a device error occurs in a PCI Host Bridge (PHB) that does not support extended error handling (EEH), it leads to the termination of all partitions sharing the bridge, resulting in data loss and system crashes, as errors propagate and require machine checks.
Innovation Solution
A method and system where the PHB dynamically switches between extended error handling and machine check modes based on the EEH support of devices below it, allowing devices to handle errors locally without propagating them, thus preventing partition termination and data loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the PHB generates a machine check when a device error occurs, then the system can detect and report the error, but all partitions sharing the PHB are terminated causing system crash
Solution Approach 1:
The patent introduces an intermediary mechanism (error handling logic in the PHB) that sits between the device error and the partition termination. When a device error occurs, the PHB intercepts the error and determines whether to generate a machine check based on the EEH capability of the offending device. This intermediary layer prevents direct propagation of errors to all partitions, allowing selective error handling that maintains partition continuity while still detecting and reporting errors when necessary.
Solution Approach 2:
The patent applies local quality by making the error handling behavior specific to each device rather than applying a uniform system-wide response. Each device is evaluated individually to determine if it supports EEH, and the PHB's error response is customized accordingly. Devices with EEH capability receive local error containment, while devices without EEH receive traditional machine check handling. This localized approach ensures that errors in one device do not unnecessarily affect other partitions.
2Reliability
If the PHB terminates all partitions when a device error occurs, then the system prevents crashing from error propagation, but critical data processing is lost
Solution Approach 1:
The PHB acts as an intermediary that filters and selectively propagates errors. Instead of uniformly terminating all partitions upon any device error, the PHB intercepts the error and uses EEH capability as a criterion to determine whether to allow error containment or propagate to machine check. This intermediary function prevents unnecessary information loss by maintaining partition continuity for devices with EEH capability, while still preserving system stability through traditional error handling when needed.
Solution Approach 2:
The patent changes the parameter of error propagation behavior based on the EEH capability parameter of the offending device. The system dynamically adjusts its error response parameter (machine check generation) according to the device's EEH support status. This parameter change allows the system to maintain stability while minimizing information loss by adapting the error handling strategy to the specific device characteristics rather than applying a fixed response.
3Device complexity
If the PHB operates in machine check mode for all devices, then error handling is simple and uniform, but devices with EEH capability cannot contain errors locally
Solution Approach 1:
The patent makes the PHB's error handling mode dynamic rather than static. The PHB continuously evaluates the EEH capability of devices and adjusts its error response behavior accordingly. When a device with EEH capability is detected, the PHB dynamically switches to a more sophisticated error containment mode for that specific device. When devices without EEH are detected, the PHB reverts to simple machine check mode. This dynamic adaptation allows the system to optimize error handling for each device type without requiring complex pre-configuration.
Solution Approach 2:
The system changes the error handling parameter (machine check generation vs. local containment) based on the detected EEH capability parameter of each device. The PHB monitors device characteristics and adjusts its operational parameter for error response. This parameter change enables the system to maintain simplicity for devices without EEH while activating advanced error containment for devices with EEH capability, thus resolving the contradiction between simplicity and functionality.
4Adaptability or versatility
If the system checks EEH support for all devices below the PHB, then the PHB can optimize error handling mode, but the initialization process becomes more complex
Solution Approach 1:
The patent applies preliminary action by performing EEH capability checks during the initialization phase before the system begins normal operation. The PHB proactively queries each device below it to determine EEH support status and configures the appropriate error handling mode in advance. This preliminary investigation allows the system to optimize error handling for the entire device hierarchy before workloads are assigned, avoiding the need for complex runtime decision-making and simplifying the operational phase.
Solution Approach 2:
The patent rushes through the EEH capability checking process during initialization by implementing efficient query mechanisms that quickly determine device support status. The system performs the necessary complexity of device enumeration and capability detection in a condensed initialization sequence, then skips ahead to optimized error handling operation. This approach concentrates the complexity into a brief setup phase rather than distributing it throughout system operation, making the overall process more manageable.
Data Source
AI summary
A method, system, and computer program product in a logical partitioned data processing system are disclosed for providing a host bridge that implements extended error handling (EEH). If all devices coupled to the host bridge implement EEH, the host bridge is initialized to operate in EEH mode. In EEH mode, the devices handle any error that occurs within the devices without reporting the error to the host bridge. All partitions that share the host bridge continue to operate without being terminated while the devices are handling the error. If at least one device does not implement EEH, the host bridge is initialized to operate in non-EEH mode. In non-EEH mode, a machine check is generated by the host bridge when an error occurs within one of the devices resulting in the termination of all partitions that share the host bridge in response to a receipt of the machine check.