Hypervisor Fundamentals
Hypervisor Fundamentals in SDV Context Popek/Goldberg Theorem A well-designed hypervisor is expected to satisfy the following three key criteria: itemize Equivalence Software running inside a Virtual Machine (VM) should behave identically to how it would run on physical hardware. The
Hypervisor Fundamentals in SDV Context
Popek/Goldberg Theorem
A well-designed hypervisor is expected to satisfy the following three key criteria:
itemize
Equivalence
Software running inside a Virtual Machine (VM) should behave identically to how it would run on physical hardware.
The guest operating system should not require awareness of virtualization to function correctly.
Safety (Resource Control)
The hypervisor must retain full control over system resources at all times. This includes:
itemize
Controlling access to hardware devices
Restricting memory access to allocated regions
Managing CPU scheduling
Handling interrupt routing
Isolating faults and terminating misbehaving VMs
itemize
Performance (Efficiency)
Programs running inside virtual machines should experience minimal performance overhead compared to execution on native hardware.
itemize
Relevance in SDV (Software-Defined Vehicle)
In modern SDV architectures, a hypervisor enables multiple domains to coexist on a single hardware platform:
itemize
Infotainment Domain (e.g., Android Automotive OS)
Vehicle Control Domain (e.g., RTOS for safety-critical functions)
Connectivity / Telematics Domain
itemize
Key implications:
itemize
Equivalence ensures legacy automotive software can run without modification.
Safety guarantees strict isolation between safety-critical and non-critical domains (e.g., ADAS vs. IVI).
Performance is essential for real-time responsiveness (e.g., instrument cluster, ADAS processing).
itemize
Types of Virtualization
Full Virtualization
In full virtualization, the guest operating system is unaware of virtualization.
itemize
No modification to guest OS (e.g., Android, Linux)
Hardware is fully emulated or virtualized
Strong isolation between domains
Slightly higher overhead
itemize
SDV Usage:
itemize
Running Android Automotive OS without modification
Supporting third-party OS or legacy systems
itemize
Para-Virtualization
In para-virtualization, the guest OS is aware of virtualization and interacts with the hypervisor via hypercalls.
itemize
Requires OS modification or specialized drivers
Reduced overhead and improved performance
Efficient resource sharing
itemize
SDV Usage:
itemize
Real-time domains (e.g., AUTOSAR, RTOS)
High-performance communication between domains
Optimized I/O (e.g., shared memory, virtio)
itemize
Comparison
center
tabular |c|c|c|
Aspect & Full Virtualization & Para-Virtualization
Guest OS Modification & Not required & Required
Awareness of Virtualization & No & Yes
Performance & Moderate & High
Compatibility & High & Limited
Typical SDV Usage & Android / Linux & RTOS / AUTOSAR
tabular
center
Practical Architecture in SDV
itemize
A single SoC hosts multiple VMs
Hypervisor enforces strict isolation between domains
Mixed-criticality workloads coexist:
itemize
Safety-critical (e.g., ADAS, vehicle control)
Non-critical (e.g., infotainment, apps)
itemize
itemize
Key Benefit:
itemize
Hardware consolidation (fewer ECUs)
Improved maintainability and OTA capability
Strong safety and security isolation
itemize
Summary
An effective hypervisor in SDV must balance:
itemize
Transparency (Equivalence) for software portability
Isolation (Safety) for functional safety (ISO 26262)
Efficiency (Performance) for real-time automotive workloads
itemize
Different virtualization techniques (full vs. para) are selected based on domain requirements, balancing compatibility and performance.