Browse Prior Art Database

Network Processing with Ethernet Frame Information and Firmware Processing Encapsulated in Search Keys

IP.com Disclosure Number: IPCOM000099083D
Published in the IP.com Journal: Volume 5 Issue 4 (2005-04-16)
Included in the Prior Art Database: 2005-Apr-16
Document File: 5 page(s) / 67K

Publishing Venue

Siemens

Related People

Juergen Carstens: CONTACT

Abstract

Computer networks may be implemented using a variety of protocol stack architectures. The protocol stack is processed inside a firmware integrated in various networking devices, e.g. routers, bridges etc. One is interested in reducing similar processing of code and network protocol packets in firmware since it would save the memory and multiple firmware instruction counts which are highly critical in limited resources systems. Further, one is interested in finding a simple and easy method to remove and add new network protocol stacks according to varying customer requirements. Till now, there are two common ways to handle with these problems: a) sharing a firmware code and b) separating firmware modules for different protocol processing. It is proposed to use an alternative method of firmware processing. The basic concept of the method can be summarized as follows: 1. The proprietary network protocols are detected by hardware and/or firmware. a) Hardware detection may be realized as follows: Every byte is streamed into hardware. Hence, hardware can process each byte to detect the network protocol. b) Alternatively, detection may be done by an internal firmware processing module. The firmware internal module will process the frame packets passed in. For example, the frame packet structure can be compared and the protocol can be detected.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 32% of the total text.

Page 1 of 5

S

Network Processing with Ethernet Frame Information and Firmware Processing Encapsulated in Search Keys

Idea: Jin Sze Sow, SG-Singapore; Taro Kamiko, SG-Singapore; Ganesha Nayak, SG-Singapore;

Yao Chye Lee, SG-Singapore;

Computer networks may be implemented using a variety of protocol stack architectures. The protocol stack is processed inside a firmware integrated in various networking devices, e.g. routers, bridges etc. One is interested in reducing similar processing of code and network protocol packets in firmware since it would save the memory and multiple firmware instruction counts which are highly critical in limited resources systems. Further, one is interested in finding a simple and easy method to remove and add new network protocol stacks according to varying customer requirements.

Till now, there are two common ways to handle with these problems: a) sharing a firmware code and
b) separating firmware modules for different protocol processing.

It is proposed to use an alternative method of firmware processing. The basic concept of the method can be summarized as follows:

1. The proprietary network protocols are detected by hardware and/or firmware.
a) Hardware detection may be realized as follows: Every byte is streamed into hardware. Hence, hardware can process each byte to detect the network protocol.
b) Alternatively, detection may be done by an internal firmware processing module. The firmware internal module will process the frame packets passed in. For example, the frame packet structure can be compared and the protocol can be detected.

2. Detected protocol frame data forms the search key in the firmware to preserve frame value and information.

Search Key 1: a) Byte Offset Position, b) Payload Length c) Mask Bit Search Key 2: a) Byte Offset Position, b) Payload Length c) Mask Bit Search Key 3: a) Byte Offset Position, b) Payload Length c) Mask Bit Search Key contains IP Datagram's Destination Address, Source Address, Destination Port, Source Port, Type of Service etc. Refer to Fig 2 & Fig 3.

3. The search key return values for
a) the next firmware processing for the Ethernet Frame or routed packets and bridged packets in the wide area and local area networks
b) the next search key.

Return Search Key values can be Destination Port, Queue Number of PDU, Priority Value for PDU Queue, Bit-wise indicator for lookup, security check result etc.

4. Editing frame format: The 2 Return Search Key values [Insert field] and [Remove field] indicate data to be added to or removed from Ethernet Frame in firmware stages.

LLC Header insertion is an example of Insert field. Removal of encapsulation data to receive payload is an example of data field removal.

5. The end of firmware processing is indicated: Firmware processing ends when Bit-wise indicator for lookup = 0. Frame is sent to hardware egress port or external host.

A possible embodiment of the proposed method is described in the following.

1. A detection of the proprietary...