International Business Machines Corporation

Motorola Corporation

Chih-Jen Chang

Jean Louis Calvignac

Claude Basso

Fabrice Jean Verplanken

Colin Beaton Verrilli

Natarajan Vaidhyanathan

Philippe Damon

Gordon Taylor Davis

William Joseph Armstrong

Cyriel Johan Minkenberg

Mircea R Gusat

Ronald Edward Fuhs

Scott Michael Willenborg

Renato John Recio

Joseph Franklin Logan

Hubertus Franke

Harm Peter Hofstee

Kenneth Michael Valk

William Todd Boyd

Douglas James Joseph

Fredy Daniel Neeser

Christoph Raisch

Francois Abel

Daniel Wind

Jeffrey James Lynch

Kenneth M Walk

Ravi Kumar Arimilli

Daniel Gene Eisenhauer

Harvey Gene Kiel

Josep Cors

Kyle Alan Lucke

Satya Prakash Sharma

Hans-Werner Tast

Jens Leenstra

Whay Chiou Lee

Henry Joseph May

Herman Dietrich Dierks

Jan-Bernd Themann

Terry L Nelms, Ii

Chetan S Yaliwal

Damon Philippe

Alan Kyle Lucke

IEEE 802.1Q and Enhanced Transmission Selection provide only eight different traffic classes that may be used to control bandwidth in a particular physical connection (or link). Instead of relying only on these eight traffic classes to manage bandwidth, the embodiments discussed herein disclose using an Enhanced Transmission Selection scheduler that permits a network device to set the bandwidth for an individual virtual LAN. Allocating bandwidth in a port based on a virtual LAN ID permits a network device to allocate bandwidth to, e.g., millions of unique virtual LANs. Thus, this technique may increase the granular control of the network fabric and its performance.

Flexible and scalable enhanced transmission selection method for network fabrics

A network fabric may divide a physical connection into a plurality of VLANs as defined by IEEE 802.1Q. Moreover, many network fabrics use Priority Flow Control to identify and segregate network traffic based on different traffic classes or priorities. Current routing protocols define only eight traffic classes. In contrast, a network fabric may contain thousands of unique VLANs. When network congestion occurs, network devices (e.g., switches, bridges, routers, servers, etc.) can negotiate to pause the network traffic associated with one of the different traffic classes. Pausing the data packets associated with a single traffic class may also stop the data packets associated with thousands of VLANs. The embodiments disclosed herein permit a network fabric to individually pause VLANs rather than entire traffic classes.

Flexible and scalable data link layer flow control for network fabrics

A network processor includes first communication protocol ports that each support 'M' minimum size packet data path traffic on 'N' lanes at &#x2018;S&#x2019; Gigabits per second (Gbps) and traffic with different communication protocol units on &#x2018;n&#x2019; additional lanes at &#x2018;s&#x2019; Gbps. The first communication protocol ports support access to an external coprocessor using parsing logic located in each of the first communication protocol ports. The parsing logic, during a parsing period, is configured to send a request to the external coprocessor at reception of a &#x2018;M&#x2019; size packet and to receive a response from the external coprocessor. The parsing logic sends a request maximum &#x2018;m&#x2019; size byte word to the external coprocessor on one of the additional lanes and receives a response maximum &#x2018;m&#x2019; size byte word from the external coprocessor on the one of the additional lanes while complying with the equation N&#xd7;S/M=&#x3c;n&#xd7;s/m.

Techniques for connecting an external network coprocessor to a network processor packet parser

A mechanism is provided for sharing a communication used by a parser (parser path) in a network adapter of a network processor for sending requests for a process to be executed by an external coprocessor. The parser path is shared by processors of the network processor (software path) to send requests to the external processor. The mechanism uses for the software path a request mailbox comprising a control address and a data field accessed by MMIO for sending two types of messages, one message type to read or write resources and one message type to trigger an external process in the coprocessor and a response mailbox for receiving response from the external coprocessor comprising a data field and a flag field. The other processors of the network poll the flag until set and get the coprocessor result in the data field.

Providing to a parser and processors in a network processor access to an external coprocessor

A mechanism is provided for merging in a network processor results from a parser and results from an external coprocessor providing processing support requested by said parser. The mechanism enqueues in a result queue both parser results needing to be merged with a coprocessor result and parser results which have no need to be merged with a coprocessor result. An additional queue is used to enqueue the addresses of the result queue where the parser results are stored. The result from the coprocessor is received in a simple response register. The coprocessor result is read by the result queue management logic from the response register and merged to the corresponding incomplete parser result read in the result queue at the address enqueued in the additional queue.

Merging result from a parser in a network processor with result from an external coprocessor

Mechanisms are provided for a network processor comprising a parser, the parser being operable to work in normal operation mode or in repeat operation mode, the parser in normal operation mode loading and executing at least one rule in a first and a second working cycle respectively, the parser in repeat operation mode being operable to repeatedly execute a repeat-instruction, the execution of each repeat corresponding to one working cycle.

Accelerating data packet parsing

A network node that forwards traffic of a converged network received from a source end node receives a second message addressed to the network node, but intended for the source end node. The second message includes at least a portion of a first message originated by the source end node and previously forwarded by the network node. The network node extracts from the first message a source identifier of the source end node in a first communication protocol and determines by reference to a data structure a destination address of the second message in a second communication protocol. The network node modifies the second message to include the destination address and forwards the second message toward the source end node in accordance with the destination address.

Message forwarding toward a source end node in a converged network environment

A host Ethernet adapter (HEA) and method of managing network communications is provided. The HEA includes a host interface configured for communication with a multi-core processor over a processor bus. The host interface comprises a receive processing element including a receive processor, a receive buffer and a scheduler for dispatching packets from the receive buffer to the receive processor; a send processing element including a send processor and a send buffer; and a completion queue scheduler (CQS) for dispatching completion queue elements (CQE) from the head of the completion queue (CQ) to threads of the multi-core processor in a network node mode. The method comprises operatively coupling an Ethernet adapter to a multi-core processor system via a processor bus, selectively assigning a first plurality of packets to a first queue pair for servicing in an endpoint mode, running a device driver on the multi-core processing system, the device driver controlling the servicing of the first queue pair by dispatching the first plurality of packets to only one processor core of the multi-core processor system, selectively assigning a second plurality of packets to a second queue pair for servicing in a network node mode; and the Ethernet adapter controlling the servicing of the second queue pair by dispatching the second plurality of packets to multiple processor threads.

Host ethernet adapter for handling both endpoint and network node communications

Assigning work, such as data packets, from a plurality of sources, such as data queues in a network processing device, to a plurality of sinks, such as processor threads in the network processing device is provided. In a given processing period, sinks that are available to receive work are identified and sources qualified to send work to the available sinks are determined taking into account any assignment constraints. A single source is selected from an overlap of the qualified sources and sources having work available. This selection may be made using a hierarchical source scheduler for processing subsets of supported sources simultaneously in parallel. A sink to which work from the selected source may be assigned is selected from available sinks qualified to receive work from the selected source.

Assigning work from multiple sources to multiple sinks given assignment constraints

An assignment constraint matrix is used in assigning work, such as data packets, from a plurality of sources, such as data queues in a network processing device, to a plurality of sinks, such as processor threads in the network processing device. The assignment constraint matrix is implemented as a plurality of qualifier matrixes adapted to operate simultaneously in parallel. Each of the plurality of qualifier matrixes is adapted to determine sources in a subset of supported sources that are qualified to provide work to a set of sinks based on assignment constraints. The determination of qualified sources may be based sink availability information that may be provided for a set of sinks on a single chip or distributed on multiple chips.

Assignment constraint matrix for assigning work from multiple sources to multiple sinks

Growing community of inventors

Apex, NC, United States of America

Chih-Jen Chang