# US 2020/0403940 A1 · Hardware assisted virtual switch

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0403940 A1
- Document ID: US20200403940A1
- Status: Published application
- Kind: A1
- Application number: 16/643,626
- Filed: 2017-10-24
- Published: 2020-12-24
- Earliest claimed: 2017-10-24
- Assignees: Intel Corporation
- Applicants: Intel Corporation
- Inventors: Daniel P. DALY; Cunming LIANG; Jian WANG; Martin ROBERTS; Shih-Wei CHIEN; Gerald Alan ROGERS
- Web page: https://patentgrep.com/patent/US20200403940A1

## CPC

- H04L 49/70: {Virtual switches}
- G06F 12/0868: Data transfer between cache memory and other subsystems, e.g. storage devices or host systems
- G06F 12/10: Address translation
- G06F 12/12: Replacement control
- H04L 45/64: using an overlay routing layer
- H04L 47/762: triggered by the network
- H04L 47/783: Distributed allocation of resources, e.g. bandwidth brokers
- G06F 2212/152: Virtualized environment, e.g. logically partitioned system
- G06F 2212/154: Networked environment
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

There is disclosed an example of a computing apparatus for providing a hardware-assisted virtual switch on a host, including: a hardware virtual switch (vSwitch) circuit; and a hardware virtual host (vHost) circuit, the vHost circuit having an interface driver specific to the hardware vSwitch and configured to provide a vHost data plane to: provide a plurality of hardware queues to communicatively couple the hardware vSwitch to a guest virtual function (VF); and present to a virtual network driver of the guest VF an interface that is backward compatible with a software network interface.

## Claims

### Claim 1

A computing apparatus for providing a hardware-assisted virtual switch on a host, comprising:
a hardware virtual switch (vSwitch) circuit; and
a hardware virtual host (vHost) circuit, the vHost circuit having an interface driver specific to the hardware vSwitch and configured to provide a vHost data plane to:
provide a plurality of hardware queues to communicatively couple the hardware vSwitch to a guest virtual function (VF); and
present to a virtual network driver of the guest VF an interface that is backward compatible with a software network interface.

### Claim 2 (depends on 1)

The computing apparatus of claim 1, wherein the virtual network driver is to provide vHost control plane functionality.

### Claim 3 (depends on 1)

The computing apparatus of claim 1, wherein the vHost control plane functionality comprises capability negotiation.

### Claim 4 (depends on 1)

The computing apparatus of claim 1, wherein the vHost control plane functionality comprises live migration.

### Claim 5 (depends on 1)

The computing apparatus of claim 1, wherein the vHost circuit is to receive an eventfd kick and translate the kick into a hardware write on the host.

### Claim 6 (depends on 1)

The computing apparatus of claim 1, wherein the vHost circuit is to provide an interrupt request (IRQ) relay to receive a host-side hardware IRQ, and translate the host-side IRQ into a software IRQ on the guest.

### Claim 7 (depends on 1)

The computing apparatus of claim 1, wherein the vHost circuit is to provide a host-side interrupt request (IRQ) directly to the VM, comprising translating the host-side IRQ into a guest-specific interrupt via an input/output memory management unit (IOMMU).

### Claim 8 (depends on 1)

The computing apparatus of claim 1 further comprising an input/output memory management unit (IOMMU), wherein the vHost circuit is to map a guest physical address (GPA) space into a host physical address (HPA) space in the IOMMU, wherein host hardware is to locate HPAs via GPAs.

### Claim 9 (depends on 8)

The computing apparatus of claim 8, wherein the vHost circuit is provide hardware separation of input/output operations for a plurality of guest virtual machines.

### Claim 10 (depends on 8)

The computing apparatus of claim 8, wherein the vHost circuit is to receive a kick, read an “avail” index, and read from a host physical memory address according to a GPA in a descriptor queue of the avail index.

### Claim 11 (depends on 8)

The computing apparatus of claim 8, wherein the vHost circuit is to pin GPA pages into host memory.

### Claim 12 (depends on 11)

The computing apparatus of claim 11, wherein pinning GPA pages into host memory comprises pinning only memory for the virtual network driver.

### Claim 13 (depends on 1)

The computing apparatus of claim 1, wherein the vHost circuit is to receive a live migration event notification, and responsive to the live migration event notification, dirty active memory pages of the guest VF and swap active memory pages out of cache.

### Claim 14 (depends on 1)

The computing apparatus of claim 1, wherein the vHost circuit is to perform pure host-side updates of guest-to-host address translation and interrupts, wherein VF passthrough for the virtual network driver is not provided to the guest.

### Claim 15 (depends on 1)

The computing apparatus of any of claim 1, wherein
the hardware vSwitch comprises a hardware IP block.

### Claim 16

One or more tangible, non-transitory computer-operable mediums having encoded thereon instructions to provide a hardware virtual host (vHost) to:
communicatively couple to a hardware virtual switch (vSwitch) circuit;
provide an interface driver specific to the hardware vSwitch and configured to provide a vHost data plane to:
provide a plurality of hardware queues to communicatively couple the hardware vSwitch to a guest virtual function (VF); and
present to a virtual network driver of the guest VF an interface that is backward compatible with a software network interface.

### Claim 17

(canceled)

### Claim 18

(canceled)

### Claim 19

(canceled)

### Claim 20 (depends on 16)

The one or more tangible, non-transitory computer-operable mediums of claim 16, wherein the hardware vHost is to receive an eventfd kick and translate the kick into a hardware write on the host.

### Claim 21

(canceled)

### Claim 22 (depends on 16)

The one or more tangible, non-transitory computer-operable mediums of claim 16, wherein the hardware vHost is to communicatively couple to an input/output memory management unit (IOMMU), wherein the hardware vHost is to map a guest physical address (GPA) space into a host physical address (HPA) space in the IOMMU, wherein host hardware is enabled to locate HPAs via GPAs.

### Claim 23

(canceled)

### Claim 24

(canceled)

### Claim 25

(canceled)

### Claim 26

A computer-implemented method of providing hardware-assisted virtual host (vHost) services to a hardware virtual switch (vSwitch), comprising:
communicatively coupling to the hardware vSwitch;
providing an interface driver specific to the hardware vSwitch and configured to provide a vHost data plane:
providing a plurality of hardware queues to communicatively couple the hardware vSwitch to a guest virtual function (VF); and
presenting to a virtual network driver of the guest VF an interface that is backward compatible with a software network interface.

### Claim 27 (depends on 26)

The method of claim 26, wherein the virtual network driver is to provide vHost control plane functionality.
