Contents

Applications on the HPC safety island (MCU)

The safety island runs predictable control and monitoring software that checks the main processor and coordinates a safe response when something goes wrong.

What runs here

The safety island is an MCU subsystem inside an HPC system-on-chip (SoC). Its application software performs explicitly allocated safety functions: bounded control loops, checks on application-processor outputs, and decisions about degraded operation or safe states. The exact allocation depends on the vehicle safety concept; local actuator control may remain on another ECU.

Component Role
Real-time control application Executes allocated control algorithms with defined deadlines and validated inputs.
Application-processor supervisor Checks progress, communication freshness, and relevant output plausibility.
Command validator Checks requested actions against permitted ranges, rates, operating modes, and available independent information.
Safety reaction coordinator Selects and requests the specified degraded mode, output inhibition, or safe-state transition.

These applications use underlying services. AUTOSAR Watchdog Manager (WdgM), communication protection, and fault storage belong to lower layers. Application code supplies meaningful checkpoints and interprets faults in the context of vehicle behavior. WdgM provides alive, deadline, and logical supervision; these mechanisms alone do not establish that an algorithm’s result is correct. AUTOSAR Watchdog Manager specification

Why it looks like this

Safety reactions need bounded execution even when the application processor is overloaded, restarting, or unavailable. A small, statically configured application set makes execution times, memory use, and fault responses easier to analyze. Local SRAM can reduce dependence on shared external memory, subject to the actual SoC architecture.

Lockstep cores and error-correcting memory support hardware fault detection. They do not establish software correctness or automatically make an application ASIL D. Software safety requirements, architecture, implementation, and verification still need appropriate evidence. ISO 26262-6:2018

Integration can reduce external components and communication overhead, but independence must be demonstrated. Shared power, clocks, interconnects, reset controls, and peripherals can create dependencies. Isolation capabilities are device-specific; consult the supplier’s safety manual and assumptions of use. TI safety and isolation overview

Keeping the monitor small also limits verification effort and coupling to frequently updated feature software. Any feature update that changes monitored behavior, timing, or interfaces still requires safety impact analysis.

How it connects

There is no higher software layer in this grid: these applications consume vehicle requirements and coordinate with peer applications on the main processor. Below them, AUTOSAR Classic software components can use the Runtime Environment (RTE); a safety RTOS implementation may expose different interfaces. Platform services provide supervision, communication checks, and diagnostic reporting, while drivers reach hardware.

Communication with the application processor may use shared memory, mailboxes, or another interprocessor channel. Define message ownership, atomicity, freshness, timeouts, startup sequencing, and behavior during resets. Where allocated, AUTOSAR E2E protection supplies communication error detection; the receiving application must define its response to invalid data. E2E protection does not prove semantic correctness or authenticate the sender. AUTOSAR E2E Library specification

Other nodes receive validated commands, health status, or safety requests through configured vehicle-network paths. The safety reaction needs an effective route to the actuator or responsible controller when the main processor fails. Resetting the HPC is only suitable when the resulting vehicle behavior satisfies the safety concept.

Compared with other node classes

  • HPC application processor: Runs feature services, HMI, or ADAS applications; the island supervises selected behavior and executes allocated bounded safety functions.
  • Zone controller: Focuses on local I/O and gateway logic, potentially enforcing actuator safety requests received from the HPC.
  • MCU-based ECU: May own an entire function, such as brake control; the island’s allocation is tied to the HPC safety architecture.
  • Smart node / LIN slave: Usually implements a narrow fixed function with fewer resources and less system-wide supervisory responsibility.

Best practices

  • Derive each monitor and reaction from a safety requirement, including detection and reaction time budgets within the fault tolerant time interval.
  • Check meaningful progress and output plausibility using sufficiently independent information; document diagnostic limitations.
  • Analyze worst-case execution and communication delays under shared-resource contention and fault conditions.
  • Define startup, shutdown, restart, and recovery states, including who has authority over outputs.
  • Inject relevant faults: stalled tasks, stale messages, implausible commands, processor resets, and unavailable communication paths.
  • Version interfaces and calibrations together, and assess compatibility whenever either processor’s software changes.

Common pitfalls

  • Treating physical core separation as proof of freedom from interference.
  • Accepting a heartbeat as evidence that the feature is computing correct results.
  • Requiring the failed processor to execute its own safety reaction.
  • Assuming immediate shutdown is always safe; some functions need controlled degradation or continued operation.

What it means for your discipline

  • Functional safety Core

    Allocates monitoring and control requirements, defines safe reactions and timing budgets, and verifies independence and fault coverage.

  • HPC & application software Core

    Implements control algorithms, plausibility checks, and safety state machines with bounded execution and explicit handling of invalid inputs.

  • Base software Partial

    Integrates application checkpoints, RTE interfaces, communication protection status, and diagnostic reporting with the underlying platform.

  • E/E architecture Partial

    Assigns control authority and ensures safety reactions can reach other controllers without depending on the failed application processor.

  • Software update & OTA Partial

    Coordinate application-processor and safety-island releases when updates change monitored behavior, timing or interfaces. Define compatible version combinations and verify that update-related restarts preserve the required monitoring and safe reactions.

No direct stake: Cybersecurity · Hardware & silicon · Networks & communication · Diagnostics & lifecycle

Where it sits in the software stack

Generic layered software of a central HPC (application processor with hypervisor, plus a safety island) and of a classic MCU-based ECU (AUTOSAR Classic). Highlighted modules matter for this topic. Click any module to open its page.

HPC · application processor (MPU) and safety island (MCU)
  • Safety monitors — Runs bounded control, validates processor outputs and coordinates safety reactions within allocated detection and reaction times.
MCU-based ECU · AUTOSAR Classic

Not directly involved on this node type.