System and method for predictive processor failure recovery
View Patent ↗A system, method, and computer program product for reporting and recovering from an internal processor error in a multiprocessor system supporting system management mode. In accordance with the method of the present invention one or more replacement agents are allocated such as during system startup within the multiprocessor system. Machine specific error-reporting registers are monitored for one or more active processor agents during system operation. In response to detecting a faulty agent via the monitoring, a system management interrupt (SMI) request is issued to one or both the faulty agent and a selected replacement agent. In response to receiving the SMI request, the operating state of the faulty agent is copied to the selected replacement agent in system management mode. Operating system processing is then resumed using the replacement agent.
1. In a multiprocessor system having an operating system and supporting a system management mode that is transparent to the operating system, a method for recovering from an internal processor error comprising:
responsive to detecting a faulty processor agent:
issuing a system management interrupt (SMI) request;
operating the faulty processor agent and a replacement processor agent in the system management mode in response to said SMI request, and while operating the faulty processor agent and the replacement processor agent in system management mode:
saving an operating state of said faulty processor agent to system management memory that is accessible by said faulty processor agent only when said faulty processor agent is operating in system management mode;
copying the saved operating state to the replacement processor agent; and
re-assigning an interrupt control ID from the faulty processor agent to the replacement processor agent; and
resuming operating system processing from the system management mode utilizing the replacement agent.
2. The method of claim 1 , wherein said detecting a faulty processor agent comprises monitoring one or more internal error-reporting registers associated with the faulty processor agent.
3. The method of claim 1 , further comprising determining availability of one or more replacement agents within the multiprocessor system.
4. The method of claim 1 , said issuing an SMI request further comprising issuing an SMI request to the faulty processor agent.