Marvell Israel (misl) Ltd.

Pmc-sierra, Inc.

Marvell Semiconductor Israel Ltd.

Yaniv Kopelman

Nafea Bishara

Carmi Arad

Yariv Anafi

Aviran Kadosh

Eitan Medina

Lior Khermosh

Ruven Torok

Zachy Haramaty

Dan Aharoni

Alon Meirson

A network device includes a first port configured to receive an incoming data packet. A memory stores the incoming data packet. A second port is configured to transmit outgoing packets. A packet processor is configured to generate a data structure, corresponding to the incoming data packet, that includes information based on a header portion of the incoming data packet and, in each of a plurality of processing operations, perform at least one processing task on the data packet using the data structure. The processing operations include adding to and/or subtracting from the information stored in the data structure, and preparing the data structure to be further modified in a subsequent processing operation. The packet processor is further configured to modify the header portion according to the data structure as modified at the plurality of processing operations and provide the stored data packet with the modified header portion to the second port.

Packet processor for altering a header portion of a data packet

A packet processor for a network device includes an incoming port that receives a first packet. The first packet includes a data portion, a control portion and a first outgoing port. A control data processing device receives the control portion from the incoming port while the data portion is stored in memory, and transmits the control portion to the first outgoing port. The first outgoing port transmits a first request for the data portion based on the control portion. A header altering device retrieves the data portion from the memory and strips, modifies, and encapsulates the data portion based on the first request.

Packet header altering device

A flow classifier for a network device that processes packets including packet headers includes a hash generator that generates hash index values from search keys derived from the packet headers. A hash table receives the hash index values and outputs pointers. A flow table includes flow keys and corresponding actions. A variable length (VL) trie data structure uses the pointers to locate the flow keys for the search keys. The VL trie data structure selects different flow keys for the search keys that share a common hash index value. The pointers include node, NIL and leaf pointers. The flow classifier performs a default action for the NIL pointers. A pointer calculator accesses a VL trie table using the pointers.

Apparatus and method for high speed flow classification

A packet processor for processing a data unit received from a network includes a header analyzer unit configured to obtain indications of locations in a header of the data unit of one or more fields to be parsed from the data unit to perform a packet processing operation on the data unit. The header analyzer unit comprises a ternary content addressable memory (TCAM), and a memory separate from the TCAM and coupled to an output of the TCAM, wherein a content of the TCAM and a content of the memory are programmable. The header analyzer unit is configured to obtain, responsive to a lookup of at least one portion of the data unit in the TCAM, indications of locations in a header of the data unit of one or more fields to be parsed from the data unit to perform a packet processing operation on the data unit. The packet processor further comprises a parser configured to parse the header using the indications of locations of one or more fields in the header to obtain data from the one or more fields.

Method and apparatus for determining locations of fields in a data unit

A system for redundancy in Ethernet passive optical networks (EPONs) facilitates fast recovery from failure (less than 50 msec), path redundancy of the fiber optic network, and location redundancy of the OLTs. An optical networking unit (ONU) in a normal state monitors input communications, and when the input communications are quiet for a predetermined minimum length of time, the ONU transitions to a lenient state in which: the ONU accepts old and new security keys; upon receiving a packet: the ONU updates an ONU timestamp based on the packet's timestamp; and the ONU transitions to the normal state of operation. While the ONU is in the lenient state if a packet is not received for a predetermined given length of time the ONU transitions to a deregistered state. In this system, main and standby OLTs do not require synchronization of security parameters or synchronization for differences in fiber lengths.

System for redundancy in Ethernet passive optical networks (EPONs)

The disclosure provides a method for reducing an amount of simultaneous switching outputs (SSO) of a device. The method of reducing the amount of simultaneous switching outputs can include driving outputs of the device to a first set of values, scrambling a second set of values to reduce an amount of simultaneous switching outputs resulting from the switching of the first to the second set of values, and driving the outputs of the device to the scrambled second set of values. Further, the method can include descrambling the scrambled second set of values back to the second set of values.

Method and apparatus for reducing simultaneous switching outputs

A packet processor includes a forwarding table configured to store address prefixes and a search engine configured to receive a packet and to search the forwarding table. The search engine includes a set bit counter configured to receive an address of the packet, to count set bits of a first bit vector associated with the packet, and to output a number of the set bits. A next hop table is configured to store a next hop pointer. A next bucket pointer receives the number of set bits and outputs one of the next hop pointers and the next bucket pointer based on the number of set bits.

Apparatus and method for efficient longest prefix match lookup

A network system may include a number of switches connected to a network processor that handles the bulk of the switching and/or routing in the system. The switches may provide per port flow control status information, e.g., the flow control status of a number of their ports, over a link to the network processor. The network processor may use this information to make traffic management decisions.

Flow control for multiport PHY

A network device includes a multi-port media access controller (MAC) device that comprises a plurality of MAC devices. Some of the MAC devices output respective data streams at different speeds. A plurality of speed translators translates the speeds of the respective data streams to be greater than or equal to a highest output speed of the plurality of MAC devices and generates parallel speed translated data streams. A multiplexer multiplexes the parallel speed translated data streams to generate a multiplexed data stream corresponding to one of the plurality of MAC devices with the highest output speed defined by the parallel speed translated data streams. A first serializer and deserializer receives the multiplexed data stream that is encoded at a physical coding sublayer and serially transmits the multiplexed data stream to a multi-port physical layer device.

Network device with multiple MAC/PHY ports

A header analyzer unit generates attribute information regarding headers of a data unit. The header analyzer unit includes a programmable memory unit having a content addressable memory (CAM) with an input to receive a first portion of the data unit and a second portion of the data unit. The programmable memory unit also includes a memory separate from the CAM and coupled to an output of the CAM. The CAM stores indications of locations within the memory separate from the CAM, and the memory separate from the CAM programmably stores header attribute information regarding a plurality of different types of headers for data units having different formats.

Apparatus for determining locations of fields in a data unit

Resources allocated to a group of ports include a plurality of storage regions. Each storage region includes a committed area and a shared area. A destination storage region is identified for a packet. A packet queuing engine stores the packet in the committed area of the determined destination storage region if it has a first drop precedence value, and if available storage space in the committed area exceeds a first threshold. The packet queuing engine stores the packet in the shared area of the determined destination storage region if the packet is not stored in the committed area, and if available storage space exceeds a second threshold defined by the packet's drop precedence value. If the packet is not stored either in the committed or shared area, it may be dropped.

Efficient management of queueing resources for switches

A hash function is applied to a set of data to generate a hash. A first subset of the hash is used to lookup an entry in a lookup table for a forwarding database. A second subset of the hash is used to identify, within the entry, data comprising an offset. The offset is applied to a location identified in the forwarding database by the first subset of the hash to determine an entry in the forwarding database. Optionally, the lookup mechanism is used in conjunction with one or more other forwarding databases. A method of updating the forwarding database within the double hash lookup framework is also described.

Double-hash lookup mechanism for searching addresses in a network device

A network device comprises a multi-port MAC device that includes N MAC devices and that outputs a port multiplexed and encoded parallel data stream. A first serializer/deserializer (SERDES) serializes the port multiplexed and encoded parallel data stream from the multi-port MAC device.

A network switch includes a matching device that employs a first data structure to select one of N address groups including a first partial address of an IP packet. An action table selects one of a next hop location corresponding to a longest prefix match and a first matching operation for the IP packet based on the selected one of the N groups. When the first matching operation is selected, the matching device employs a second data structure to select one of M address groups including a second partial address of the IP packet. The action table selects one of a next hop location corresponding to a longest prefix match and a second matching operation for the IP packet based on the selected one of the M groups. A routing engine routes the IP packet based on the first and second next hop locations.

Growing community of inventors

Holon, Israel

Yaniv Kopelman