KVM PV DEVICES.

Transcription

1 K DEVICES 1

2 Agenda Introduction & brief history VirtIO Enhanced VirtIO with K support Further implementation 2

3 General & history Fully virtualized devices performs bad 55 Mbps for RTL Lots of io-exits per packet Decided to implement a modern e1000 Advantage: Only coding no guest tools involved Irq coalescing Only 2-3 io-exits per packet Can be the base of user-space But then came Ingo 3

4 architecture H guest H guest R I H frontend N guest R I G H frontend guest R N G I frontend R N I G N G frontend Linux kernel & K K 4

5 General & history V0 - leveraging Ingo Molnar s code Make loadable module Add H support Add NAPI Add memory barriers and improved ring Keep running after performance & stability Merge to the kernel? Alternatives Xen Polished s Xen specific VirtIO was just published. 5

6 requirements Close to native performance Merge with the kernel Leverage existing code Have usermode implementation Block device for qcow, vmdk formats Ability to function without K (-no-kvm) Ability to run H guests Robust code Easily add s 6

7 VirtIO An API for virtual I/O Implements network & block logic Written by Rusty Russell Motivation Many hypervisors of all types Hard to tune and maintain each one Code reuse The K way ;) Implementations Lguest K Possible (Xen, UML,, ware?..) 7

8 VirtIO Guest kernel VIRTIO BlK VIRTIO NET Add_buf Get_buf Sync Restart Detach Hypervisor Specifics code Kick Linux kernel & K K scatterlist sg[] for skb/blk_req data 8

9 VirtIO Hypervisor specifics The front end logic is implemented by VirtIO The needs Probing & Bus services Enumeration Irq Parameters (mac,..) Shared memory with remote side Hypercalls Host /userspace Guest kernel VIRTIO BlK Linux kernel & K K VIRTIO NET Hypervisor Specifics code 9

10 Enhanced VirtIO Motivation Increase re-use Allow operation with various bus types Make new devices code smallest Components Shared memory code With per hypervisor I/O hypercalls Bus (pci, virtual bus) Host Guest kernel VIRTIO BlK Linux kernel & K K VIRTIO NET Hypervisor Specifics code 10

11 Enhanced VirtIO Blk Net VIRTIO Net VIRTIO BlK Guest kernel Enhanced VIRTIO VIRTIO Virtual Bus Hypercalls Shared mem Linux kernel & K K 11

12 Enhanced VirtIO Status: Interface needs polishing K support PCI bus support Result: Makes tiny 620 Mbps throughput for network H Linux guest Before optimization Userspace 162 LOC 195 LOC Guest kernel Blk Enhanced VIRTIO Virtual Bus Linux kernel & K K Net Hypercalls VIRTIO Net VIRTIO Shared mem VIRTIO BlK 12

13 Enhanced VirtIO shared memory VirtIO Implements VirtIO interface Callbacks to hypervisor and Code consists of add_buf, get_buf, restart, detach_buf Only shared memory logic needed sync Ring logic IO pending hypercall be_virtqueue_interrupt handler Guest kernel Blk Enhanced VIRTIO Virtual Bus Linux kernel & K K Net Hypercalls VIRTIO Net VIRTIO Shared mem VIRTIO BlK 13

14 Enhanced VirtIO Day in a life of packet Guest kernel Blk Enhanced VIRTIO Virtual Bus Net Hypercalls 3.start_xmit VIRTIO Net 4.add_buf VIRTIO 6.sync 11.irq 8.Notify hcall Shared mem VIRTIO BlK 5.add_buf 7.sync 2.Kernel Tcp/ip 1.User write 12. skb_recv_done 13. get_buf Linux kernel & K 10.Trigger irq K 9.The pkts travels to tap/ host device 14

15 Enhanced VirtIO shared memory details Based on lguest 1-1 shared memory Data structure Page of descriptors for rx, tx. Available pointers page controlled by guest Used pointers page controlled by host SG list is currently internal to descriptors Descriptors are chained by next pointer 15

16 Enhanced VirtIO network be Implements kvm_virtnet_probe for pci bus Creates tx,rx be_new_virtqueue Probes virtnet 195 LOC Guest kernel Blk Enhanced VIRTIO Virtual Bus Net Hypercalls VIRTIO Net VIRTIO Shared mem VIRTIO BlK Request_irq Irq# taken from bus Linux kernel & K K Register hypercall - shared memory pfn Device key for enumeration taken from bus 16

17 Further work Basic Add readv/writev handlers to Complete the user-space block device Update with VirtIO gso. Complete migration support Blk Net VIRTIO Net VIRTIO BlK Advanced (also simple) Publish the enhanced interface Optimize and stabilize Add host back end s Add virtual bus H improvements Test with kernel Guest kernel Enhanced VIRTIO Virtual Bus Linux kernel & K K Hypercalls VIRTIO Shared mem 17

18 Q&A Thank you ;) 18