Method of implementing precise, localized hardware-error workarounds under centralized control

摘要

In a processor, a localized workaround is activated upon the sensing of a problematic condition occurring on said processor, and then control of the deactivation of the localized workaround is superseded by a centralized controller. In a preferred embodiment, the centralized controller monitors forward progress of the processor and maintains the workaround in an active condition until a threshold level of forward progress has occurred. Optionally, the localized workaround may be re-activated while under centralized control, resetting the notion of forward progress. Using the present invention, localized workarounds perform effectively while having a minimal impact on processor performance.