Solution Overview

When Container Manager is started, it registers the fault subscription DCMI. If a fault occurs, the driver reports the event to Container Manager via this interface. Once the fault is resolved, the driver reports the rectification event to Container Manager through the same interface.

When an NPU fault occurs, the fault management framework collects the fault details and forwards them to the NPU driver's fault management framework. After receiving the fault details, the fault management framework reports them to Container Manager through the DCMI, as shown in Figure 1.

Figure 1 Fault detection mechanism