The invention provides an interrupt monitoring method and device for an
operating system kernel in a multi-kernel
processing system, and the method comprises the steps: collecting interrupt
event data of a corresponding CPU core through a monitoring unit; executing interrupt
storm detection and interrupt
processing delay monitoring based on the collected interrupt
event data; when an interrupt
storm event is detected or an interrupt
processing delay abnormity is monitored, generating diagnosis information which comprises at least one of an interrupt number, a device identifier, a
call stack and a process context; and outputting a monitoring result and diagnosis information through a configuration interface and a query interface provided by the
ProcFs file system. According to the method, through a distributed low-overhead architecture and an intelligent triggering mechanism, real-time, accurate, operable and maintainable production-level monitoring of interruption
storm and processing time
delay of the multi-
core system is realized.