BCM53XX/BCM56XX Stack & Software Development Kit Guide (Release 5.4.0)

3星 · 超过75%的资源需积分: 35 201 浏览量更新于2024-07-23 1 收藏 1.25MB PDF 举报

本文档是关于BCM53XX/BCM56XX系列芯片的"Stacking Guide/Software Development Kit (SDK)"，由Broadcom Corporation发布，日期为2006年8月24日，版本为5.4.0。这份指南详细介绍了该芯片组的编程开发者手册，包含了软件开发工具和技术支持信息。文档的核心内容涵盖了以下几点： 1. 芯片型号与系列： BCM53XX和BCM56XX是主要的芯片型号，它们可能用于不同的应用环境，如网络连接、无线通信等领域。这些芯片可能是Wi-Fi、蓝牙或其它无线技术的处理核心。 2. 文档更新历史： - 56XX-PG703-R（2006年8月24日）：支持5.4.0软件版本，表明此版本的SDK包含了对最新固件和功能的全面支持。 - 56XX-PG702-R（2006年6月1日）：支持5.3.1软件版本，展示了版本迭代的持续性。 - 56XX-PG701-R（2005年6月20日）：针对5.3.0软件版本提供了指导，早期版本可能包含基础功能和API的介绍。 - 56XX-PG700-R（2004年11月5日）：首次发布的版本，标志着SDK的起步阶段。 3. 版权和法律声明： Broadcom Corporation对其商标和标志享有权利，并强调其他提及的品牌和名称属于其各自所有者。文档声明所有权利保留，并注明是在美国印刷。 4. 目录：文档结构通常包括详细的目录，可能涵盖了章节标题、功能概述、API参考、示例代码、故障排除、安装指南等内容，帮助开发者高效地理解和使用BCM53XX/BCM56XX芯片。 5. 文档性质：这是一份程序员指南，旨在提供给开发人员深入了解芯片工作原理、接口规范、驱动程序开发以及软件集成的详细资料，以便于构建基于BCM芯片的应用程序。这份BCM53XX/BCM56XX SDK文档是开发人员在设计和实现与这些芯片相关的硬件及软件系统时的重要参考资料，提供了从基础到高级功能的完整开发工具包和支持。

8/24/2006 ZF05G



BCM53XX/BCM56XX Programmer’s Guide

08/24/06

Broadcom Corporation

Page 2 Overview Document 56XX-PG703-D2

WHAT IS STACKING?

Stacking refers to a method of connecting two or more physical switch chips to build a larger system that behaves as a single

logical entity. For Strata

devices like the BCM5645, this means connecting devices together using Gigabit ports configured

in SL stacking mode. For XGS

devices, stacking is done over the HiGig™ link (HiGig stacking) or over a Gigabit link (SL

stacking). Starting with XGSIII, SL stacking on Gigabit links is no longer supported.

For both HiGig and SL stacking, the hardware attaches information to each packet traversing a stack link. This allows chip

functions to work properly across multiple devices. For example, link aggregation requires special information be passed

with the packet to provide load balancing across the aggregated ports. More detailed information about stacking and stack

tags can be found in the relevant chip documentation. However, the goal of the stacking software module is to eliminate the

need for applications to be aware of these details.

STACKING IMPLICATIONS FOR DEVICE PROGRAMMING

When multiple devices exist in a system, configuration changes such as creating a VLAN may require the programming of

all of these devices. For example, consider two BCM5690 devices, or any XGS devices with HiGig links, connected back-

to-back using the HiGig link. Usually, only one CPU is used to manage this configuration. Using the BCM API to control these

devices is relatively straight-forward, as two devices are recognized on the PCI bus and programmed directly by the CPU.

Figure 1 shows two devices connected back-to-back with a CPU. In such a configuration, there are programming require-

ments to ensure the two chips act as one. For example, to create a VLAN that includes all 24 Gigabit ports, each device

must be programmed to create the VLAN, and add all 12 GE ports and the HiGig links to the VLAN. This is required to ensure

traffic flows properly between the two devices.

Figure 1: Two Devices Connected Back-to-Back with a CPU

To enable rapid development on configurations with multiple devices, the BCMX API is provided. BCMX provides a logical

view of all the switch chips in the system. Moreover, BCMX handles any extra programming required to configure multiple

8/24/2006 ZF05G



Programmer’s Guide BCM53XX/BCM56XX

08/24/06

Broadcom Corporation

Document 56XX-PG703-D2 Overview Page 3

device systems. For example in Figure1onpage2, it handles the proper configuration of the HiGig stacking port connecting

the two devices when VLANs are created, or deleted.

The BCMX layer provides a uniform approach to managing simple configurations like that discussed above as well as more

complicated systems involving many devices and multiple CPUs.

MULTIPLE CPUS

A more complex form of stacking involves multiple CPUs. It is common to allow multiple independent switches to be con-

nected together to act as one logical switch. For example, a 24-port GE switch with its own CPU subsystem and local con-

figuration can be connected to another such unit.

Figure 2 shows an example of such connection with three boxes (A, B, and C) stacked together. Note that each box can

include multiple switch devices itself. In such a model, each device in the system is controlled by a CPU that must correctly

program the local devices, as well as coordinate with the CPUs in the other boxes. This type of environment introduces a

new set of issues over those discussed previously.

Figure 2: Three Boxes Stacked Together

In SDK 5, a new concept called the Dispatch Layer is introduced. The Dispatch Layer allows devices to be attached either

locally or remotely. For local devices, BCM calls are processed on the local CPU. For devices that are attached to remote

CPUs, BCM API calls are processed using an RPC mechanism. This effectively makes a remote switch chip look like a local

switch chip to applications using the BCM API. Since the BCMX API is implemented on top of the BCM layer, it manages

systems involving multiple CPUs as easily as single CPU systems controlling multiple devices.

Two new problems are introduced when dealing with multiple CPUs. The first is system discovery. This includes determining

the devices in the system and their interconnections, referred to as the topology of the system. Once this information is ob-

tained, the devices must be properly programmed to allow data and control plane traffic to flow correctly.

The second problem is providing a control plane communications protocol. An in-band protocol uses the existing stack links

to send control information between the CPUs, that is, alongside the normal data packets. An out-of-band protocol uses a

separate communications path to send control information between the CPUs. The SDK stacking software provides the in-

band protocol; support for out-of-band protocols is also provided.

We now examine some of the tasks involved in distributed systems development and the services provided by the software

distribution to address those tasks.

8/24/2006 ZF05G



BCM53XX/BCM56XX Programmer’s Guide

08/24/06

Broadcom Corporation

Page 4 Overview Document 56XX-PG703-D2

TASK OVERVIEW

The following paragraphs briefly describe the tasks performed by the stacking software.

Managing Information

About the System

The CPU database is used to organize information about the system. CPUs are identified

by a unique “key” (which by default is the MAC address associated with the CPU). Each

entry in the database maintains information such as the number of devices controlled by

that CPU, the module ID requirements for each device and the list of ports that can provide

external connections – stack ports.

Accumulating Information

About the System

The Discovery module provides a reference implementation for determining the system

configuration. This process exchanges CPU database information using the Next Hop

communications protocol. Discovery supports timeouts and detects completion of the

process. It does not do any configuration of devices. It can run at any time, either before

or after configuration for normal switching and does not interfere with normal switching.

This allows responding to system events without disrupting normal packet flow.

Analyzing the System

Configuration

The Topology module provides a reference implementation for determining an operational

configuration for the system. It examines the connections detected by discovery,

determines the shortest path between subsystems and indicates which stack ports should

be enabled for which connections. It does not do any system configuration; it only updates

the database object it is given.

Configuring Devices Based

on the Topology

Once a configuration is determined that allows the system to operate normally, the

information is used to program the physical devices accordingly.

The Topology module, in cooperation with the diagnostic shell, provides device

configuration for many Broadcom reference designs. To provide flexibility, the analysis

and configuration operations have been decoupled from each other and from standard

monitoring and control operations.

Detecting and Responding

to System Configuration

Events

The stack task module registers with appropriate API calls to detect events such as link

change and discovery packet arrival. It manages discovery and topology analysis and

provides the necessary hooks for notification of system reconfiguration events. This is

referred to as attaching and detaching devices to the BCM, BCMX and application layers.

The BCMX layer makes calls to the application for each port detected during attach.

CPU Transport Services

The ATP module provides client-based, reliable and best effort transport services

between CPU subsystems. This module is normally used by the BCM dispatch layer for

its RPC calls. The ATP module is fully configurable and supports local loopback and per-

packet override of many transport characteristics.

Uniform System Control

The BCM API allows seamless control of both local and remote devices. Nearly all BCM

API calls use a “unit” number which is a reference to a specific switch or fabric device. The

dispatch layer provides redirection of the calls to support multiple architectures as well as

remote operation requests.

The BCMX API provides a uniform, port-based view of the entire system. Most BCMX

calls use a “logical port” number, which is a means of identifying an arbitrary port in the

system. Operations, which require updating multiple devices, are automatically carried out

by the BCMX layer through calls to the BCM layer.

8/24/2006 ZF05G



Programmer’s Guide BCM53XX/BCM56XX

08/24/06

Broadcom Corporation

Document 56XX-PG703-D2 About This Document Page 5

ABOUT THIS DOCUMENT

This document describes the BCM56XX Network Switching Software Stacking APIs in the sections briefly described in

Table 1.

REVISION LEVELS

This document is part of the following guides that cover SDK releases through 5.4.0. The other two documents are:

•

Network Switching SDK for BCM API Version 5.4 Programmer's Guide

, 56XX-PG607-R

•

Network Switching SDK for Software Release 5.4.0 Programmer’s Guide

, 56XX-PG801-R

•

Network Switching Simulation Environment (BCMSIM) Programmer’s

Guide, 56XX-PG900-R

•

Suggested Use of Broadcom APIs for Network Security Enforcement on BCM5651X Devices Application Note

, 5651X-

AN100-R

These documents replace the 56XX-DD02-R Device Driver Writer’s Guide, see “Revision History” on page ii in the

Network

Switching SDK for the BCM API Version 5.4 Programmer's Guide

, 56XX-PG607-R.

NOTATIONAL CONVENTIONS

API Reference Page

A sketch of an API reference page follows. The API Call and Syntax sections are always present. All others are present if

necessary.

Table 1: Sections Overview

Section Brief Description

Section 1: “Introduction” on page 1 Stacking software overview; about this document.

Section 2: “CPU Database” on page 10 CPU database objects and their management.

Section 3: “CPU Transport Services” on page 33 Low level communication services.

Section 4: “Next Hop Transport” on page 55 Simple distributed communication protocol.

Section 5: “ATP: Acknowledged Transport

Protocol” on page 84

Client-oriented, reliable communication service.

Section 6: “ATP Transport over Sockets” on page 113 Reference module for ATP using socket interface.

Section 7: “Discovery” on page 123 Reference module for module discovery.

Section 8: “Topology Analysis” on page 140 Reference module which analyzes system interconnections.

Section 9: “Stack Task” on page 150 Reference system management module.

剩余205页未读，继续阅读

tongzy2009

粉丝: 0
资源: 2

BCM53XX/BCM56XX Stack & Software Development Kit Guide (Release ...

最新资源

BCM53XX/BCM56XX Stack & Software Development Kit Guide (Release ...

博通56160芯片datasheet

树莓派4B BCM2711 ARM Peripherals.rar

BCM2837- 树莓派3B 芯片手册.pdf

raspberry pi bcm2711文档

bcm芯片的acl限速的代码配置资料

bcm2837芯片手册

bcm6332.pdf

bcm4360 pdf

bcm8955x 手册

bcm5464s中文技术文档

最新资源