Partitioning ARM Systems with the Jailhouse Hypervisor
summary
An ARM-focused minimal hypervisor that boots underneath Linux to carve out fixed partitions, assign CPUs and devices one-to-one, and avoid scheduling; implementation is intentionally tiny at <10k lines of code and runs on ARM64 and x86 in QEMU and real boards. Communication uses configurable shared memory plus a simple event channel; IOMMU/SMMU work is partial, pinmux and clock handoff remain tricky, and recovery if the root Linux fails requires planned reboots or extra design.
AI-written summary · about
chapters (7)
- 0:00Introduction and speaker background
- 2:57Jailhouse hypervisor design goals
- 11:57Current ARM support and features
- 21:14Demo images and future plans
- 31:10Hardware partitioning challenges
- 41:25Safety certification and hardware requirements
- 50:11Q&A and community discussion
related talks
50mProtected KVM on Arm64: A Technical Deep Dive
18mZVM: Zephyr-based Virtual Machine
21mEmbedded solution for multiprocess application
21mOak stage0: a minimal firmware for (confidential) virtual machines
38mSecure Containers in Embedded Deployments
27mvTZ: Virtualizing ARM TrustZone
39mCloudLinux runtime kernel update for ARM growth difficulties
43mAddress Space Isolation for Enhanced Safety of the Linux Kernel
