A Virtio Message Based Transport
summary
Proposal defines a new message-based transport for virtio that replaces trap-and-emulate MMIO/PCI with explicit messages to cut context switches and avoid grant/DMA assumptions. It uses FF-A for synchronous and indirect messaging, memory share/unshare and UUID device discovery, and exposes a bus abstraction layer so implementations can run over FF-A, Xen-specific channels or mailboxes. Draft spec exists, Google/Linaro proof-of-concept between TrustZone/Trusty is underway, and patches aim to keep existing virtio drivers unchanged.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker background
- 1:11What is Virtio and its purpose
- 3:20Current Virtio transports overview
- 7:13New use cases for Virtio
- 10:08Problems with existing transports
- 13:38Why message-based transport
- 15:24Design of message-based transport
- 18:56Implementation over FFA
- 21:27Xen-specific bus alternative
- 22:29Current status and next steps
- 24:15Q&A session
related talks
42mDisplay Sharing in Heterogeneous Systems Using Linux as High...
29mIntroduction to OpenAMP
26mFreeFlow: Software-based Virtual RDMA Networking for Containerized Clouds
45mUSB Support in Zephyr OS
25mEnabling Real-Time Secure Connectivity to the Industrial Edge with Single-Pair Ether...
14mUnderstanding RDMA Microarchitecture Resources for Performance Isolation
23mMind the gaps why we need new, truly open, multi domain heterogeneous solutions
31mOn Paravirualizing TCP: Congestion Control on Xen VMs