Processor Interrupt Loss Detection via Register Parameter Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In heterogeneous processor designs, dedicated processors often miss interrupts, leading to abnormal chip behavior, and existing methods fail to reliably detect these missed interrupts.
Innovation Solution
A processor detection method that determines the working state of a dedicated processor by analyzing parameters stored in specific registers, such as the fc_head and inst_num registers, to identify changes in data processing and instruction counts, thereby detecting interrupt misses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If dedicated processor is used to process specific tasks, then processing efficiency is improved, but interrupt reliability deteriorates (interrupts may be missed)
Solution Approach 1:
The patent implements a feedback mechanism by continuously monitoring register parameters (fc_head, inst_num) in the dedicated processor and comparing them against expected values. This feedback loop enables the system to detect when interrupts are missed by identifying discrepancies between actual processor state and expected state, thereby resolving the reliability issue while maintaining processing efficiency.
Solution Approach 2:
The patent introduces intermediary components including a general-purpose processor that acts as a mediator to monitor the dedicated processor's state through register readings, and uses intermediary data structures (recorded parameter values, comparison results) to detect interrupt misses. This intermediary approach allows interrupt monitoring without directly interfering with the dedicated processor's efficient operation.
2Reliability
If register parameters are monitored to detect interrupt misses, then detection reliability is improved, but system complexity increases
Solution Approach 1:
The patent applies self-service by having the dedicated processor automatically maintain and update its own register parameters (fc_head, inst_num) that reflect its processing state. The monitoring system simply reads these self-maintained parameters without needing to insert complex probe code into the dedicated processor, thereby achieving reliable detection with minimal added complexity.
Solution Approach 2:
The patent monitors changes in specific register parameters (fc_head indicating data processing state, inst_num indicating instruction count) to detect interrupt misses. By focusing on a small set of critical parameters rather than comprehensive system state monitoring, the solution achieves high detection reliability with low system complexity.
Data Source
AI summary
A method for detecting a processor comprises: determining a parameter stored in a first-type register in a detected processor, which is used to detect changes in the data processed by the detected processor, wherein the parameter relates to the a number of the data that needs to be processed currently by the detected processor, and relates to the a processing mode of the data that needs to be processed currently by the detected processor which can be characterized by using information of instructions or functions called during processing; determining a working state of the detected processor based on the parameter, and performing an interrupt missing handling operation if the working state is a first type of working state being an interrupt missing state. The method can conveniently and reliably realize the detection of interrupt loss, thereby effectively monitoring the situation of interrupt loss.