Bare-Metal KVM Hypervisor: Interview Quick-Reference Index

A structured navigation hub for explaining the Bare-Metal KVM Hypervisor & Virtualization Engine (CS 695: Topics in Virtualization and Cloud Computing | Prof. Purushottam Kulkarni).

📂 Quick Links & Speeches

Page	Type	Duration	Primary Focus
Intro Pitch	Speech	60–90 sec	High-level overview: KVM API, guest memory mapping, I/O traps, and multi-VM scheduling.
Detailed Architecture	Speech	3.5–5 min	End-to-end 4-stage walkthrough: VM creation, vCPU configuration, VM exit handling, and inter-VM coordination.
Topic 01: Core KVM API & Memory	Deep Dive	2–3 min	Three-tier file descriptor hierarchy (/dev/kvm, vm_fd, vcpu_fd), mmap with MAP_NORESERVE, and KVM_SET_USER_MEMORY_REGION.
Topic 02: VM Exits & I/O Traps	Deep Dive	2–3 min	Hardware VM exit mechanics, KVM_EXIT_HLT, and implementing trap-based hypercalls via in/out assembly instructions.
Topic 03: GVA→HVA Translation	Deep Dive	2–3 min	Guest Virtual Address to Host Virtual Address mapping: GVA → GPA → HVA (HVA = vm->mem + GVA).
Topic 04: Dual-VM Circular Buffer	Deep Dive	2–3 min	Producer/Consumer VM architecture, isolated memory spaces, and hypervisor-mediated 20-element ring buffer sync.
Topic 05: vCPU Scheduling	Deep Dive	2–3 min	Userspace time-sliced scheduling trace, hardware state management in KVM, and cooperative context switches.

🎯 Key Systems Takeaways

Hardware-Assisted Virtualization: Operating system turns into a Type-2 hypervisor using Intel VMX / AMD-V via /dev/kvm ioctls.
Memory Virtualization: Userspace memory allocated via mmap() is committed on-demand and assigned as Guest Physical RAM via EPT/NPT tables.
Hypercall Trap Interface: Freestanding guest code communicates across the VM boundary by deliberately triggering KVM_EXIT_IO via x86 I/O port instructions.
Inter-VM Memory Isolation: Multiple virtual machines communicate securely without shared physical memory by routing updates through hypervisor shadow buffers.