System and method for input/output acceleration device having storage virtual appliance (SVA) using root of PCI-E endpoint

1. A method for accelerating data transfer operations, the method comprising:

prior to booting a hypervisor on a processor subsystem having a root complex, configuring a first Peripheral Component Interconnect Express (PCI-E) endpoint and a second PCI-E endpoint associated with an accelerator device, wherein the accelerator device is a physical PCI-E device exclusively accessible to the processor subsystem;

booting the hypervisor using the processor subsystem;

loading a storage virtual appliance (SVA) as a virtual machine on the hypervisor, wherein the hypervisor assigns the second PCI-E endpoint for exclusive access by the SVA using a second root port of the root complex;

activating the first PCI-E endpoint by the SVA using a first root port of the root complex via the second PCI-E endpoint;

responsive to activating the first PCI-E endpoint, loading a hypervisor device driver for the first PCI-E endpoint, wherein the first PCI-E endpoint appears to the hypervisor as a logical hardware adapter accessible via the hypervisor device driver; and

initiating, by the SVA, a data transfer operation between the first PCI-E endpoint and the second PCI-E endpoint.

2. The method of claim 1 , wherein the data transfer operation is between a first virtual machine and a second virtual machine executing on the hypervisor.

3. The method of claim 2 , wherein the first PCI-E endpoint is associated with a first address space for the first virtual machine and the second PCI-E endpoint is associated with a second address space for the second virtual machine, wherein the accelerator device performs address translation transactions between the first address space and a physical memory address space and between the second address space and the physical memory address space.

4. The method of claim 3 , wherein the accelerator device performs address translation caching for the address translation transactions, wherein the address translation caching is performed before the data transfer operation is initiated.

5. The method of claim 1 , wherein the data transfer operation includes a direct memory access (DMA) operation performed by the accelerator device, and wherein initiating the data transfer operation includes:

causing, by the SVA, DMA parameters for the DMA operation to be sent the accelerator device.

6. The method of claim 1 , wherein the data transfer operation includes a programmed input/output (PIO) operation performed by the accelerator device.

7. The method of claim 1 , further comprising, by the SVA via the second PCI-E endpoint:

when the data transfer operation is in progress, terminating the data transfer operation;

deactivating the first PCI-E endpoint;

programming, on the accelerator device, a first personality profile for the first PCI-E endpoint and a second personality profile for the second PCI-E endpoint, wherein a personality profile includes configuration information for a PCI-E endpoint;

restarting the second PCI-E endpoint; and

responsive to restarting the second PCI-E endpoint, restarting the first PCI-E endpoint.

8. The method of claim 1 , wherein the data transfer operation includes data processing operations performed by the accelerator device, wherein the data processing operations are selected from:

encryption;

compression;

checksum; and

malicious code detection.

9. The method of claim 1 , wherein the SVA uses the second PCI-E endpoint to access a private device on the accelerator device, the private device selected from:

a memory device;

a network interface adapter;

a storage adapter; and

a storage device.

10. The method of claim 2 , wherein the SVA programs the accelerator device to generate interrupts associated with the data transfer operation.

11. An information handling system, comprising:

an accelerator device that is a physical Peripheral Component Interconnect Express (PCI-E) device;

a processor subsystem having access to a memory subsystem and having exclusive access to the accelerator device, wherein the memory subsystem stores instructions executable by the processor subsystem for accelerating data transfer operations on a hypervisor using the accelerator device, the instructions, when executed by the processor subsystem, cause the processor subsystem to:

prior to executing instructions to boot the hypervisor, configure a first PCI-E endpoint and a second PCI-E endpoint associated with the accelerator device;

boot the hypervisor using the processor subsystem;

load a storage virtual appliance (SVA) as a virtual machine on the hypervisor, wherein the hypervisor assigns the second PCI-E endpoint for exclusive access by the SVA;

activate the first PCI-E endpoint by the SVA via the second PCI-E endpoint;

responsive to the instructions to activate the first PCI-E endpoint, load a hypervisor device driver for the first PCI-E endpoint, wherein the first PCI-E endpoint appears to the hypervisor as a logical hardware adapter accessible via the hypervisor device driver; and

initiate, by the SVA, a data transfer operation between the first PCI-E endpoint and the second PCI-E endpoint.

12. The information handling system of claim 11 , wherein the data transfer operation is between a first virtual machine and a second virtual machine executing on the hypervisor.

13. The information handling system of claim 12 , wherein the first PCI-E endpoint is associated with a first address space for the first virtual machine and the second PCI-E endpoint is associated with a second address space for the second virtual machine, wherein the accelerator device performs address translation transactions between the first address space and a physical memory address space and between the second address space and the physical memory address space.

14. The information handling system of claim 13 , wherein the accelerator device performs address translation caching for the address translation transactions, wherein the address translation caching is performed before the data transfer operation is initiated.

15. The information handling system of claim 11 , wherein the data transfer operation includes a direct memory access (DMA) operation performed by the accelerator device, and wherein the instructions to initiate the data transfer operation include instructions to:

cause, by the SVA, DMA parameters for the DMA operation to be sent the accelerator device.

16. The information handling system of claim 11 , wherein the data transfer operation includes a programmed input/output (PIO) operation performed by the accelerator device.

17. The information handling system of claim 11 , further comprising instructions executed by the SVA via the second PCI-E endpoint to:

when the data transfer operation is in progress, terminate the data transfer operation;

deactivate the first PCI-E endpoint;

program, on the accelerator device, a first personality profile for the first PCI-E endpoint and a second personality profile for the second PCI-E endpoint, wherein a personality profile includes configuration information for a PCI-E endpoint;

restart the second PCI-E endpoint; and

responsive to the instructions to restart the second PCI-E endpoint, restart the first PCI-E endpoint.

18. The information handling system of claim 11 , wherein the data transfer operation includes data processing operations performed by the accelerator device, wherein the data processing operations are selected from:

encryption;

compression;

checksum; and

malicious code detection.

19. The information handling system of claim 11 , wherein the SVA uses the second PCI-E endpoint to access a private device on the accelerator device, the private device selected from:

a memory device;

a network interface adapter;

a storage adapter; and

a storage device.

20. The information handling system of claim 12 , wherein the SVA programs the accelerator device to generate interrupts associated with the data transfer operation.