Abstract: A host with a network interface that offloads a TCP connection is disclosed in which the notification that relatively small data transmit commands have completed is batched whereas the notification that relatively large data transmit commands have completed are not batched. The notification that data transmit commands have completed may be intrinsically modulated by the size and frequency of the commands and the processing of the data transfer by the TCP connection. One embodiment involves a method comprising: running an application on a computer having a network interface; running, on the network interface, a TCP connection for the application; providing, by the computer to the network interface, a command to send data from the application; updating, by the network interface, a SndUna value for the TCP connection; and providing, by the network interface to the computer, the SndUna value, thereby indicating to the computer that the command has been completed.

Abstract: An intelligent network interface card (INIC) or communication processing device (CPD) works with a host computer for data communication. The device provides a fast-path that avoids protocol processing for most messages, greatly accelerating data transfer and offloading time-intensive processing tasks from the host CPU. The host retains a fallback processing capability for messages that do not fit fast-path criteria, with the device providing assistance such as validation even for slow-path messages, and messages being selected for either fast-path or slow-path processing. A context for a connection is defined that allows the device to move data, free of headers, directly to or from a destination or source in the host. The context can be passed back to the host for message processing by the host. The device contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: A network host such as a caching device is disclosed that greatly increases the speed with which a server reads and writes data for clients. The host may include a specialized network interface that not only processes TCP but also parses received network file system headers, identifying those headers within the TCP data stream, separating them from any corresponding network file system data, and providing them separately from that data to the network file system of the host for processing as a batch, all without any interrupt to the host. Similarly, the network file system of the host may communicate directly with the network interface by writing network file system headers directly to a register of the network interface to transmit data.

Abstract: A system for protocol processing in a computer network has an intelligent network interface card (INIC) or communication processing device (CPD) associated with a host computer. The INIC provides a fast-path that avoids protocol processing for most large multipacket messages, greatly accelerating data communication. The INIC also assists the host for those message packets that are chosen for processing by host software layers. A communication control block for a message is defined that allows DMA controllers of the INIC to move data, free of headers, directly to or from a destination or source in the host. The context is stored in the INIC as a communication control block (CCB) that can be passed back to the host for message processing by the host. The INIC contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: A 10 Gb/s network interface device offloads TCP/IP datapath functions. Frames without IP datagrams are processed as with a non-offload NIC. Receive frames are filtered, then transferred to preallocated receive buffers within host memory. Outbound frames are retrieved from host memory, then transmitted. Frames with IP datagrams without TCP segments are transmitted without any protocol offload, but received frames are parsed and checked for protocol errors, including checksum accumulation for UDP segments. Receive frames without datagram errors are passed to the host and error frames are dumped. Frames with Tcp segments are parsed and error-checked. Hardware checking is performed for ownership of the socket state. TCP/IP frames which fail the ownership test are passed to the host system with a parsing summary. TCP/IP frames which pass the ownership test are processed by a finite state machine implemented by the CPU. TCP/IP frames for non-owned sockets are supported with checksum accumulation/insertion.

Abstract: A system for protocol processing in a computer network has a TCP/IP Offload Network Interface Device (TONID) associated with a host computer. The TONID provides a fast-path that avoids protocol processing for most large multi-packet messages, greatly accelerating data communication. The TONID also assists the host for those message packets that are chosen for processing by host software layers. A communication control block for a message is defined that allows DMA controllers of the TONID to move data, free of headers, directly to or from a destination or source in the host. The context is stored in the TONID as a communication control block (CCB) that can be passed back to the host for message processing by the host. The TONID contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: An intelligent network interface card (INIC) or communication processing device (CPD) works with a host computer for data communication. The device provides a fast-path that avoids protocol processing for most messages, greatly accelerating data transfer and offloading time-intensive processing tasks from the host CPU. The host retains a fallback processing capability for messages that do not fit fast-path criteria, with the device providing assistance such as validation even for slow-path messages, and messages being selected for either fast-path or slow-path processing. A context for a connection is defined that allows the device to move data, free of headers, directly to or from a destination or source in the host. The context can be passed back to the host for message processing by the host. The device contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: A network interface device has a fast-path ACK generating and transmitting mechanism. ACKs are generated using a finite state machine (FSM). The FSM retrieves a template header and fills in TCP and IP fields in the template. The FSM is not a stack, but rather fills in the TCP and IP fields without performing transport layer processing and network layer processing sequentially as separate tasks. The filled-in template is placed into a buffer and a pointer to the buffer is pushed onto a high-priority transmit queue. Pointers for ordinary data packets are pushed onto a low-priority transmit queue. A transmit sequencer outputs a packet by popping a transmit queue, obtaining a pointer, and causing information pointed to by the pointer to be output from the network interface device as a packet. The sequencer pops the high-priority queue in preference to the low-priority queue, thereby accelerating ACK generation and transmission.

Abstract: A network interface device connected to a host provides hardware and processing mechanisms for accelerating data transfers between the host and a network. Some data transfers are processed using a dedicated fast-path whereby the protocol stack of the host performs no network layer or transport layer processing. Other data transfers are, however, handled in a slow-path by the host protocol stack. In one embodiment, the host protocol stack has an ISCSI layer, but a response to a solicited ISCSI read request command is nevertheless processed by the network interface device in fast-path. In another embodiment, an initial portion of a response to a solicited command is handled using the dedicated fast-path and then after an error condidtion occurs a subsequent portion of the response is handled using the the slow-path. The interface device uses a command status message to communicate status to the host.

Abstract: A TCP Offload Engine (TOE) device includes a state machine that performs TCP/IP protocol processing operations in parallel. In a first aspect, the state machine includes a first memory, a second memory, and combinatorial logic. The first memory stores and simultaneously outputs multiple TCP state variables. The second memory stores and simultaneously outputs multiple header values. In contrast to a sequential processor technique, the combinatorial logic generates a flush detect signal from the TCP state variables and header values without performing sequential processor instructions or sequential memory accesses. In a second aspect, a TOE includes a state machine that performs an update of multiple TCP state variables in a TCB buffer all simultaneously, thereby avoiding multiple sequential writes to the TCB buffer memory. In a third aspect, a TOE involves a state machine that sets up a DMA move in a single state machine clock cycle.

Abstract: A host CPU runs a network protocol processing stack that provides instructions not only to process network messages but also to allocate processing of certain network messages to a specialized network communication device, offloading some of the most time consuming protocol processing from the host CPU to the network communication device. By allocating common and time consuming network processes to the device, while retaining the ability to handle less time intensive and more varied processing on the host stack, the network communication device can be relatively simple and cost effective. The host CPU, operating according to instructions from the stack, and the network communication device together determine whether and to what extent a given message is processed by the host CPU or by the network communication device.

Abstract: At least one intelligent network interface card (INIC) is coupled to a host computer to offload protocol processing for multiple network connections, reducing the protocol processing of the host. Plural network connections can maintain, via plural INIC ports and a port aggregation switch, an aggregate connection with a network node, increasing bandwidth and reliability for that aggregate connection. Mechanisms are provided for managing this aggregate connection, including determining which port to employ for each individual network connection, and migrating control of an individual network connection from a first INIC to a second INIC.

Abstract: An intelligent network interface card (INIC) or communication processing device (CPD) works with a host computer for data communication. The device provides a fast-path that avoids protocol processing for most messages, greatly accelerating data transfer and offloading time-intensive processing tasks from the host CPU. The host retains a fallback processing capability for messages that do not fit fast-path criteria, with the device providing assistance such as validation even for slow-path messages, and messages being selected for either fast-path or slow-path processing. A context for a connection is defined that allows the device to move data, free of headers, directly to or from a destination or source in the host. The context can be passed back to the host for message processing by the host. The device contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: An intelligent network interface card (INIC) or communication processing device (CPD) works with a host computer for data communication. The device provides a fast-path that avoids protocol processing for most messages, greatly accelerating data transfer and offloading time-intensive processing tasks from the host CPU. The host retains a fallback processing capability for messages that do not fit fast-path criteria, with the device providing assistance such as validation even for slow-path messages, and messages being selected for either fast-path or slow-path processing. A context for a connection is defined that allows the device to move data, free of headers, directly to or from a destination or source in the host. The context can be passed back to the host for message processing by the host. The device contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: A host CPU runs a network protocol processing stack that provides instructions not only to process network messages but also to allocate processing of certain network messages to a specialized network communication device, offloading some of the most time consuming protocol processing from the host CPU to the network communication device. By allocating common and time consuming network processes to the device, while retaining the ability to handle less time intensive and more varied processing on the host stack, the network communication device can be relatively simple and cost effective. The host CPU, operating according to instructions from the stack, and the network communication device together determine whether and to what extent a given message is processed by the host CPU or by the network communication device.

Abstract: A host CPU runs a network protocol processing stack that provides instructions not only to process network messages but also to allocate processing of certain network messages to a specialized network communication device, offloading some of the most time consuming protocol processing from the host CPU to the network communication device. By allocating common and time consuming network processes to the device, while retaining the ability to handle less time intensive and more varied processing on the host stack, the network communication device can be relatively simple and cost effective. The host CPU, operating according to instructions from the stack, and the network communication device together determine whether and to what extent a given message is processed by the host CPU or by the network communication device.

Abstract: An intelligent network interface card (INIC) or communication processing device (CPD) works with a host computer for data communication. The device provides a fast-path that avoids protocol processing for most messages, greatly accelerating data transfer and offloading time-intensive processing tasks from the host CPU. The host retains a fallback processing capability for messages that do not fit fast-path criteria, with the device providing assistance such as validation even for slow-path messages, and messages being selected for either fast-path or slow-path processing. A context for a connection is defined that allows the device to move data, free of headers, directly to or from a destination or source in the host. The context can be passed back to the host for message processing by the host. The device contains specialized hardware circuits that are much faster at their specific tasks than a general purpose CPU.

Abstract: A host CPU runs a network protocol processing stack that provides instructions not only to process network messages but also to allocate processing of certain network messages to a specialized network communication device, offloading some of the most time consuming protocol processing from the host CPU to the network communication device. By allocating common and time consuming network processes to the device, while retaining the ability to handle less time intensive and more varied processing on the host stack, the network communication device can be relatively simple and cost effective. The host CPU, operating according to instructions from the stack, and the network communication device together determine whether and to what extent a given message is processed by the host CPU or by the network communication device.

Abstract: A host CPU runs a network protocol processing stack that provides instructions not only to process network messages but also to allocate processing of certain network messages to a specialized network communication device, offloading some of the most time consuming protocol processing from the host CPU to the network communication device. By allocating common and time consuming network processes to the device, while retaining the ability to handle less time intensive and more varied processing on the host stack, the network communication device can be relatively simple and cost effective. The host CPU, operating according to instructions from the stack, and the network communication device together determine whether and to what extent a given message is processed by the host CPU or by the network communication device.

Abstract: A TCP Offload Engine (TOE) device includes a state machine that performs TCP/IP protocol processing operations in parallel. In a first aspect, the state machine includes a first memory, a second memory, and combinatorial logic. The first memory stores and simultaneously outputs multiple TCP state variables. The second memory stores and simultaneously outputs multiple header values. In contrast to a sequential processor technique, the combinatorial logic generates a flush detect signal from the TCP state variables and header values without performing sequential processor instructions or sequential memory accesses. In a second aspect, a TOE includes a state machine that performs an update of multiple TCP state variables in a TCB buffer all simultaneously, thereby avoiding multiple sequential writes to the TCB buffer memory. In a third aspect, a TOE involves a state machine that sets up a DMA move in a single state machine clock cycle.