Kernel bypass for ISCSI and NVME/TCP applications

A method performed by a host device to offload iSCSI and NVMe/TCP data plane traffic processing to a network interface card (NIC) connected to the host device, the method comprising:

• running an iSCSI or NVMe/TCP application in a user space of the host device;
• receiving, at a kernel space of an operating system a system call to register transmit and receive virtual memory buffers with the NIC;
• receiving, at a kernel space of an operating system, a system call to set up an offloaded iSCSI or NVMe/TCP server or to set up an iSCSI or NVMe/TCP connection between the host device and destination device with which the application desires to communicate;
• receiving, at a kernel space of an operating system, a system call to create a transmit queue and a system call to map transmit queue memory in application's virtual address space;
• receiving, at a kernel space of an operating, a system call to create a receive queue and to map receive queue memory in application's virtual address space;
• establishing, by the kernel space, the iSCSI or NVMe/TCP connection with the destination device;
• writing, by the application, iSCSI or NVMe/TCP PDU transmit work request from the user space directly into a transmit queue memory mapped in application's virtual address space, wherein the PDU is transmitted over a hardware interface and via a data path that bypasses the kernel space;
• and
• retrieving, by the application, iSCSI or NVMe/TCP PDU receive completion from a receive queue memory mapped in application's virtual address space, the PDU being retrieved over the hardware interface and via the data path that bypasses the kernel space.

A network interface card (NIC) comprising:

• one or more processors;
• a communications interface configured to communicatively couple the NIC to a host device facilitate establishment, via the NIC, of an iSCSI or NVMe/TCP connection between the host device and a destination device;
• a transmit queue configured to receive, via the communications interface, iSCSI or NVMe/TCP PDU transmit work request from the host device, the PDU transmit work request being written into the transmit queue memory mapped in application's virtual address space directly from an application running in a user space of the host device;
• one or more hardware chips encoded with a iSCSI and NVMe/TCP stack configured to process the PDU to generate a packet to be transmitted to the destination device;
• a network interface configured to transmit the packet over a network to the destination device and receive a packet from the destination device;
• and
• a receive queue configured to receive PDU transmit and receive completion and provide the application running in the user space virtual address of PDU header and buffer.