Nios® V Processor: Lockstep Implementation

ID 833274
Date 10/07/2024
Public
Document Table of Contents

2.2.1. System Phases

Figure 5. System Phases and Timing Relationship
Table 3.  System Phases and Timing Relationship
System Phases Description Timing Relationship
Detection

The self–checking comparator detects a fault in one of the two Nios® V processor CPUs or the fRSmartComp itself.

Detection Time Interval (DTI)
  • Related to faults occurring in the CPUs in the order of various clock cycles (≤5).
  • The time between the fault exiting the CPU (CPU boundary) and the fRSmartComp alarm generation.
Insulation

The subsystem composed of the two Nios® V processor CPUs and the fRSmartComp is optionally isolated from the rest of the integrated circuit. The CPUs are put in halt debug mode, stopping the execution of the application software. This may be performed to prevent data contamination toward the rest of the system.

Insulation Time Internal (ISTI)
  • The time between the fRSmartComp alarm generation and the Nios® V processor subsystem isolation.
Repair

The fRSmartComp performs actions (with the System Supervisor) implementing failure-control scenarios to recover the system from a faulty state.

Repair Time Interval (RTI)
  • Enforced within the fRSmartComp with a timeout-checking mechanism to control the duration of recovery actions.
  • The timeout ensures that a faulty condition never prevents the application from executing the planned failure control actions. Should the timeout expire, an alarm may eventually be fired.

The fRSmartComp provides the following functions to support the abilities described above:

  • CPU Fault detection: Comparators detecting improper behavior of the processors.
  • Timeout: The fRSmartComp checks for the correct evolution of the system. For instance, checks for System Supervisor access on fRSmartComp:
    • After asynchronous reset within a specific time.
    • After alarm generation within a specific time.
  • Counters: The fRSmartComp checks for the number of fRSmartComp restarts done in a specific time frame to avoid “endless loops” problems.
  • Specialized safety mechanisms: Diagnose latent faults internal to the fRSmartComp.

After fault detection, the fRSmartComp can execute programmable actions (i.e., issue interrupt to the System Supervisor). Upon receiving the interrupt, the system supervisor can carry out the failure control actions.

The following figure schematically represents the high-level process governing the fRSmartComp from the beginning of the first activation (after reset) to when it passes the control to the system Supervisor and goes into an inactive functional state. After an alarm generation, you must explicitly restart the fRSmartComp using the System Supervisor. It is crucial that the System Supervisor always have full control over the fRSmartComp.

The fRSmartComp is in “ON-LINE” mode until it detects an error and generates an alarm. After that, it transitions into “OFF-LINE” mode, implementing a failure control action (either asserting a safe state or executing a restart).

Figure 6. High-level Process of fRSmartComp