Switching chip, host and endpoint equipment communication system

By introducing a number conversion module in the PCIe switching chip, the local domain number is converted into global domain number, and the problem of high chip complexity and cost in the multi-host interconnection scenario in traditional technology is solved, and efficient cross-domain host and endpoint devices are achieved Communication.

CN118921335APending Publication Date: 2024-11-08SHANDONG YUNHAI GUOCHUANG CLOUD COMPUTING EQUIP IND INNOVATION CENT CO LTD
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202410977127.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In the multi-host interconnection scenario, traditional PCIe switching chips require a large number of non-transparent bridges, which increases chip complexity and cost.

Method used

A switching chip is designed, using a number conversion module to convert local domain numbers into global domain numbers, and using global domain numbers as an intermediary to realize packet transmission between the host and endpoint devices, reducing the difficulty and cost of multi-host interoperability.

Benefits of technology

All local interoperability is achieved through interoperability between each local domain to global domain, significantly reducing the number of non-transparent bridges and reducing chip complexity and cost.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118921335A_ABST
    Figure CN118921335A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of chip design, and discloses a switch chip, a host and an endpoint device communication system, a number conversion module of the switch chip obtains a first request message sent by the host, converts a first requester local domain number in the first request message into a first requester global domain number, and sends the first requester global domain number to the host; sending the converted first request message to a first endpoint device corresponding to the address of the first target endpoint device in the first request message; acquiring a first response message sent by the first endpoint device, if the first response message comprises the first requester global domain number, converting the first requester global domain number in the first response message into a first requester local number, converting the first responder global domain number into a first responder local number, and if the first response message comprises the first requester global domain number, converting the first responder global domain number into a first responder local number; and sending the converted first response message to the corresponding host. The problem that a chip is high in complexity and high in cost due to the fact that a large number of non-transparent bridges are used for achieving cross-domain communication between a host and endpoint equipment is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of chip design, and in particular to a communication system between a switching chip, a host and an endpoint device. Background Art

[0002] According to the PCIe (peripheral component interconnect express, high-speed serial computer expansion bus standard) protocol, the RC (Root Complex) in the host CPU (Central Processing Unit) is the root node of a PCIe tree, which is responsible for managing the entire PCIe tree. The RC can have one or more root ports to connect PCIe devices outside the CPU to expand the PCIe tree. Each root port can connect to an endpoint device, or expand multiple ports by connecting a PCIe switch chip to connect multiple endpoint devices.

[0003] Traditional PCIe switching chips are generally implemented directly in the manner defined by the PCIe protocol, and the internal bridge device is managed by the host RC connected to the corresponding upstream port. As technologies such as AI continue to promote the demand for multi-host interconnection in servers, PCIe switching chips are also facing the demand for multi-PCIe domains and inter-domain interoperability. Since the PCIe standard protocol does not define cross-domain operations, cross-domains generally use custom interconnections between endpoint devices to transmit information. Such a pair of endpoint devices is generally called a "non-transparent bridge" between two hosts. Since the two endpoint devices of the non-transparent bridge belong to the source PCIe domain and the target PCIe domain respectively, each source / target pair requires a dedicated non-transparent bridge. If the number of hosts is further increased and full mutual access is required, the number of non-transparent bridges required will increase further, which will greatly increase the complexity of implementation and significantly increase chip costs.

[0004] Therefore, the related technology has the problem of using a large number of non-transparent bridges to achieve cross-domain communication between the host and the endpoint devices, which leads to high chip complexity and high chip cost. Summary of the invention

[0005] In view of this, the present invention provides a switching chip, a host and endpoint device communication system to solve the problem of high chip complexity and high chip cost caused by using a large number of non-transparent bridges to achieve cross-domain communication between the host and endpoint devices.

[0006] In a first aspect, the present invention provides a switching chip, wherein an uplink port of the switching chip is used to connect to a first preset number of hosts, and a downlink port of the switching chip is used to connect to a second preset number of endpoint devices; the switching chip includes a number conversion module;

[0007] The number conversion module is used to obtain a first request message sent by the host, convert the first requester local domain number in the first request message into the first requester global domain number, and send the converted first request message to the first endpoint device corresponding to the first target endpoint device address in the first request message;

[0008] The number conversion module is used to obtain a first response message sent by a first endpoint device, determine whether the first response message includes a first requester global domain number, and if included, convert the first requester global domain number in the first response message into a first requester local domain number, convert the first responder global domain number in the first response message into a first responder local domain number, and send the converted first response message to a host corresponding to the first requester local domain number.

[0009] The switching chip provided in this embodiment uses the number conversion module in the switching chip to convert the local domain numbers in various messages into global domain numbers, and convert the global domain numbers into local domain numbers. The global domain number is used as an intermediary to achieve smooth transmission of messages between the host and the endpoint device, reducing the difficulty and cost of implementing multi-host interoperability. It is only necessary to implement the interoperability between each local domain and the global domain to achieve all local domain interoperability, which can greatly reduce the combination of local domain interoperability. It solves the problem of using a large number of non-transparent bridges to achieve cross-domain communication between the host and the endpoint device, resulting in high chip complexity and high chip cost.

[0010] In an optional implementation, the number conversion module is further configured to:

[0011] Obtaining a second request message sent by the second endpoint device, and if the second request message contains the first target host address, converting the second requester global domain number in the second request message into the second requester local domain number, and sending the converted second request message to the target host corresponding to the first target host address;

[0012] Obtain a second response message sent by the target host, convert the second requester local domain number in the second response message into a second requester global domain number, convert the second responder local domain number in the second response message into a second responder global domain number, and send the converted second response message to an endpoint device corresponding to the second requester global domain number.

[0013] In this embodiment, a number conversion module is used to convert the local domain number in the second request message and the second response message into a global domain number, and convert the global domain number into a local domain number, so that the endpoint device can actively send a request message to the host and receive a response message from the host, thereby reducing the difficulty and cost of implementing multi-host interoperability.

[0014] In an optional implementation, the switching chip further includes: a global domain number allocation module, a third preset number of uplink bridge devices, and a fourth preset number of downlink bridge devices;

[0015] The upstream bridge device is used to connect to the host, obtain the configuration space read and write instructions sent by the host, and after receiving the configuration space read and write instructions, send a waiting response to the host;

[0016] The global domain number allocation module is used to obtain the first device information of the downstream bridge device and the second device information of the endpoint device, generate the global domain number of the downstream bridge device according to the first device information, and generate the global domain number of the endpoint device according to the second device information;

[0017] The global domain number allocation module is connected to the upstream bridge device, and is used to obtain the configuration space read and write instructions from the upstream bridge device, obtain the local domain number of the downstream bridge device and the local domain number of the endpoint device from the configuration space read and write instructions, generate a first association relationship according to the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, generate a second association relationship according to the local domain number of the endpoint device and the global domain number of the endpoint device, generate a reply response packet according to the first device information and / or the second device information, and send the reply response packet to the host through the upstream bridge device;

[0018] The global domain number allocation module is connected to the number conversion module and is used to configure the first association relationship and the second association relationship to the number conversion module.

[0019] In this embodiment, the global domain number allocation module generates a global domain number for the downstream bridge device based on the first device information, and generates a global domain number for the endpoint device based on the second device information; generates a first association relationship based on the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, and generates a second association relationship based on the local domain number of the endpoint device and the global domain number of the endpoint device; so that the local domain number and the global domain number of the device can be converted subsequently through the first association relationship and the second association relationship, cross-domain interaction between any two hosts can be realized, and the difficulty and cost of implementing multi-host interoperability are reduced.

[0020] In an optional implementation manner, the global domain number allocation module obtains the configuration space read and write instruction from the upstream bridge device, including:

[0021] Read the internal registers of the upstream bridge device to obtain the configuration space read and write instructions;

[0022] The global domain number allocation module sends a reply response packet to the host through the upstream bridge device, including:

[0023] The reply response packet is configured in the internal register of the upstream bridge device, and the host obtains the reply response packet from the internal register of the upstream bridge device.

[0024] In an optional implementation, the global domain number allocation module is further used to:

[0025] Generate the host global domain number corresponding to the root port in the host;

[0026] Obtain the number of the root port in the host through the upstream bridge device, and use the number of the root port as the host local domain number;

[0027] Generate a third association relationship between the host global domain number and the host local domain number;

[0028] The third association relationship is configured to the number conversion module.

[0029] In this embodiment, the global domain number allocation module generates a host global domain number and obtains a host local domain number, generates a third association relationship between the host global domain number and the host local domain number, and configures it to the number conversion module. The local domain number and the global domain number of the host are subsequently converted through the third association relationship, so that cross-domain interaction between any two hosts can be achieved, reducing the difficulty and cost of implementing multi-host interoperability.

[0030] In an optional implementation manner, a third preset number of downlink bridge devices are used to generate a fourth preset number of downlink bridge device groups;

[0031] The downstream bridge device group includes a target downstream bridge device and other downstream bridge devices, wherein the target downstream bridge device is used to be connected to the target downstream bridge device in other downstream bridge device groups through a physical link, and the other downstream bridge devices are used to be connected to the endpoint device;

[0032] The host sends a first request message to the endpoint devices connected to the remaining downstream bridge devices in the non-corresponding downstream bridge device group through the target downstream bridge device and the physical link.

[0033] In this embodiment, the target downstream bridge device is used to connect with the target downstream bridge devices in other downstream bridge device groups through a physical link, so that the host sends a request message to any endpoint device through the target downstream bridge device and the physical link. The number of interconnected hosts can be greatly increased and cross-domain interaction between any two hosts can be achieved, reducing the difficulty and cost of implementing multi-host interoperability.

[0034] In an optional implementation, the number conversion module is further configured to:

[0035] Obtaining a second request message sent by the endpoint device, and if the second request message does not include the first target host address, no number conversion is performed;

[0036] The endpoint device is used to send the second request message to the endpoint device corresponding to the second request message, and obtain a third response message from the endpoint device corresponding to the second request message.

[0037] In an optional implementation manner, the number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including:

[0038] Obtaining the host's local domain bus number, local domain device number, and local domain function number from the first requester's local domain number;

[0039] According to the third association relationship, determine the global domain bus number corresponding to the local domain bus number, the global domain device number corresponding to the local domain device number, and the global domain function number corresponding to the local domain function number;

[0040] A first requester global domain number is determined, wherein the first requester global domain number includes a global domain bus number, a global domain device number, and a global domain function number.

[0041] In this embodiment, the local domain bus number, local domain device number and local domain function number in the local domain number of the first requester are all converted to ensure that the number conversion is accurate, so that the first request message is accurately sent to the corresponding endpoint device and the host successfully receives the first response message.

[0042] In an optional implementation manner, the number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including:

[0043] Obtaining the host's local domain bus number, local domain device number, and local domain function number from the first requester's local domain number;

[0044] Use the local domain device number as the global domain device number and the local domain function number as the global domain function number;

[0045] Determine the global domain bus number corresponding to the local domain bus number according to the third association relationship;

[0046] A first requester global domain number is determined, wherein the first requester global domain number includes a global domain bus number, a global domain device number, and a global domain function number.

[0047] In this implementation, only the local domain bus number in the local domain number of the first requester is converted to improve the efficiency of the number conversion.

[0048] In a second aspect, the present invention provides a host and endpoint device communication system, the system comprising: a first preset number of hosts, a second preset number of endpoint devices and a switching chip, the switching chip being the switching chip of the first aspect and any corresponding embodiment;

[0049] The uplink port of the switching chip is used to connect to a first preset number of hosts, and the downlink port of the switching chip is used to connect to a second preset number of endpoint devices; the switching chip includes a number conversion module.

[0050] The host and endpoint device communication system provided by this embodiment uses the number conversion module in the switching chip to convert the local domain number in various messages into the global domain number, and convert the global domain number into the local domain number. The global domain number is used as an intermediary to achieve smooth transmission of messages between the host and the endpoint device, reducing the difficulty and cost of implementing multi-host interoperability. It is only necessary to implement the interoperability between each local domain and the global domain to achieve all local domain interoperability, which can greatly reduce the combination of local domain interoperability. It solves the problem of using a large number of non-transparent bridges to achieve cross-domain communication between the host and the endpoint device, resulting in high chip complexity and high chip cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 is a schematic diagram of the structure of a current PCIe switching chip according to an embodiment of the present invention;

[0053] Figure 2 is a schematic diagram of cross-domain interoperability implementation of a current PCIe chip according to an embodiment of the present invention;

[0054] Figure 3 is a schematic diagram of a switching chip according to an embodiment of the present invention;

[0055] Figure 4 is a schematic diagram of number conversion performed by a number conversion module according to an embodiment of the present invention;

[0056] Figure 5 is a flowchart of steps 1 to 3 of generating a number according to an embodiment of the present invention;

[0057] Figure 6 It is a schematic diagram of numbering in the global domain of a switching chip according to an embodiment of the present invention.

[0058] Figure 7 is a flowchart of steps 4 to 6 of generating numbers according to an embodiment of the present invention;

[0059] Figure 8 is a schematic diagram of numbering under a host local domain according to an embodiment of the present invention;

[0060] Fig. 9 is a flowchart of step 7 and step 8 of generating a number according to an embodiment of the present invention;

[0061] Fig.10 4 is a schematic diagram of the structure of a host and endpoint device communication system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0063] Figure 1 It is a schematic diagram of the structure of the current PCIe switching chip, such as Figure 1 As shown in FIG. 1 , a typical PCIe switching chip includes at least one upstream port and multiple downstream ports. There are multiple PCIe bridge devices between the upstream port and the downstream port. Each upstream port and downstream port is connected to a bridge device. The bridge device connected to the upstream port is an upstream bridge device, and the bridge device connected to the downstream port is a downstream bridge device. Each upstream bridge device can be connected to multiple downstream bridge devices for expansion. Each downstream bridge device can be connected to an endpoint device, or further expanded by cascading an upstream bridge device to expand multiple downstream bridge devices.

[0064] RC's management of the PCIe tree is reflected in the fact that it can configure these devices by sending configuration space read and write requests to each device on the tree, including upstream bridge devices, downstream bridge devices and endpoint devices, and assign a bus number to each device. It should be noted that sending configuration space read and write requests and assigning bus numbers are unique functions of RC. There are two main routing methods of PCIe, namely address routing and ID (Identity document, number) routing. RC allocates an address space for each endpoint device and each bridge device through configuration space reading and writing, where the address space of each bridge device covers the address space of all downstream devices. For data packets using address routing, it is determined in turn whether the target address in the packet hits the address segment in the above device, and the path of the data packet and the final target device are found. The ID routing method is similar to address routing. ID routing uses three parts: bus number, device number and function number for matching. These three parts are referred to as BDF. The bus number is assigned by RC when initializing the device, and the device number and function number are determined by the device itself. All devices on the PCIe tree must have a unique ID number, and all endpoint devices must have non-overlapping address ranges to ensure that the two main routing methods mentioned above can find a unique target. The PCIe tree managed by the same RC needs to be allocated in an address space and bus space, which is also called a PCIe domain.

[0065] Traditional PCIe switching chips are generally implemented directly in accordance with the definition of the PCIe protocol, and the internal bridge device is managed by the host RC connected to the corresponding uplink port. For the traditional single PCIe domain environment, this method does not have any problems. However, as technologies such as AI continue to promote the demand for multi-host interconnection in servers, PCIe switching chips are also facing the demand for multiple PCIe domains and inter-domain interoperability. In this scenario, the traditional implementation of PCIe switching chips will face a series of difficulties. For example, Figure 2 As shown, since the PCIe standard protocol does not define cross-domain operations, cross-domain generally uses custom interconnection between endpoint devices to transmit information. Such a pair of endpoint devices is generally called a "non-transparent bridge" between two hosts. The implementation of the non-transparent bridge is not within the scope of the present invention, but the present invention will logically discuss the implementation difficulties of traditional chips: Since the two endpoint devices of the non-transparent bridge belong to the source PCIe domain and the target PCIe domain respectively, each pair of source PCIe domains and target PCIe domains requires a dedicated non-transparent bridge, for example, Figure 2In the example, there is a non-transparent bridge between host 0 and host 1, and between host 0 and host 2. However, the communication between host 1 and host 2 cannot be achieved by any of the above non-transparent bridges. Instead, an additional pair of endpoint devices is required to form a non-transparent bridge. If the number of hosts is further increased and full mutual access is required, the number of transparent bridges required will increase further, which will greatly increase the complexity of implementation and significantly increase chip costs.

[0066] Based on the above content, an embodiment of the present invention provides a switching chip, in which the downstream bridge device and the connected endpoint device inside the switching chip are managed by the CPU inside the switching chip, and the CPU is the actual RC of the above devices. The bus numbers of these bridge devices and endpoint devices are allocated by the CPU instead of the host. The host RC actually only directly manages the upstream bridge device and a small number of other exclusive endpoint devices. The data communication between the host and the bridge device and the endpoint device needs to be converted through the BDF conversion logic. Through the switching chip of the present application, the difficulty and cost of implementing multi-host interoperability can be reduced. The PCIe domain (hereinafter referred to as the global domain) managed by the CPU inside the switching chip acts as an intermediary, allowing multiple hosts to directly interoperate. In order to achieve the interoperability between all local domains, it is only necessary to implement the interoperability between each local domain and the global domain, which can greatly reduce the combined effect of local domain interoperability.

[0067] In this embodiment, a switching chip is provided, such as Figure 3 As shown, the uplink port of the switching chip is used to connect to a first preset number of hosts, and the downlink port of the switching chip is used to connect to a second preset number of endpoint devices; the switching chip includes a number conversion module;

[0068] The number conversion module is used to obtain a first request message sent by the host, convert the first requester local domain number in the first request message into the first requester global domain number, and send the converted first request message to the first endpoint device corresponding to the first target endpoint device address in the first request message;

[0069] The number conversion module is used to obtain a first response message sent by a first endpoint device, determine whether the first response message includes a first requester global domain number, and if included, convert the first requester global domain number in the first response message into a first requester local domain number, convert the first responder global domain number in the first response message into a first responder local domain number, and send the converted first response message to a host corresponding to the first requester local domain number.

[0070] Specifically, Figure 3As shown, the uplink port of the switching chip is connected to a first preset number of hosts, and the downlink port of the switching chip is used to connect to a second preset number of endpoint devices. The first preset number and the second preset number both represent multiple, and no specific number limit is made here. The switching chip includes a number conversion module (BDF conversion module).

[0071] Due to the difference between the local bus number and the global bus number, the local domain of the host and the global domain of the switching chip cannot directly interact with each other in data packets. Instead, BDF conversion is required when transmitting across the two domains. Therefore, a BDF conversion module is required to convert the local domain numbers in various messages into global domain numbers, and convert the global domain numbers into local domain numbers.

[0072] When the host sends a first request message to the endpoint device, the first request message needs to pass through the BDF conversion module to convert the first requester local domain number therein into the first requester global domain number. Figure 4 As shown, host 0 sends a first request message to the BDF conversion module, the first request message includes a first requester local domain number of {15, 0, 0}, the BDF conversion module converts the first requester local domain number in the first request message into a first requester global domain number of {255, 0, 0}, and the BDF conversion module sends the converted first request message to the first endpoint device corresponding to the first target endpoint device address in the first request message, the first target endpoint device address is configured by the host for each endpoint device before the host sends the first request message to the endpoint device.

[0073] When the endpoint device sends the first response message to the host, it will first determine whether the first response message is sent to the host according to the requester BDF. If it is, the first response message also needs to pass through the BDF conversion module to restore the first requester global domain number to the first requester local domain number, and look up the table in the BDF conversion module to convert the first responder global domain number to the corresponding first responder local domain number; if it is not sent to the host, the BDF does not need to be converted. For example: Figure 4 As shown, the breakpoint device sends a first response message to the BDF conversion module, and the BDF conversion module determines whether the first response message includes the first requester global domain number. If it does, it can be determined that the first response message is sent to the host. Figure 4The first response message includes a first requester global domain number of {255, 0, 0} and a first responder global domain number of {2, 0, 0}. Therefore, it can be determined that the first response message is sent to the host. The BDF conversion module converts the first requester global domain number in the first response message into a first requester local domain number of {15, 0, 0}, and converts the first responder global domain number in the first response message into a first responder local domain number of {18, 0, 0}. The BDF conversion module sends the converted first response message to the host corresponding to the first requester local domain number.

[0074] It should be noted that the BDF conversion module may be located at the upstream bridge device or distributed at the downstream bridge device, and this application does not limit the specific location.

[0075] The switching chip provided in this embodiment uses the number conversion module in the switching chip to convert the local domain numbers in various messages into global domain numbers, and convert the global domain numbers into local domain numbers. The global domain number is used as an intermediary to achieve smooth transmission of messages between the host and the endpoint device, reducing the difficulty and cost of implementing multi-host interoperability. It is only necessary to implement the interoperability between each local domain and the global domain to achieve all local domain interoperability, which can greatly reduce the combination of local domain interoperability. It solves the problem of using a large number of non-transparent bridges to achieve cross-domain communication between the host and the endpoint device, resulting in high chip complexity and high chip cost.

[0076] In some optional implementations, the number conversion module is further used to:

[0077] Obtaining a second request message sent by the second endpoint device, and if the second request message contains the first target host address, converting the second requester global domain number in the second request message into the second requester local domain number, and sending the converted second request message to the target host corresponding to the first target host address;

[0078] Obtain a second response message sent by the target host, convert the second requester local domain number in the second response message into a second requester global domain number, convert the second responder local domain number in the second response message into a second responder global domain number, and send the converted second response message to an endpoint device corresponding to the second requester global domain number.

[0079] Specifically, the method in which the endpoint device sends a request message to the host and the host returns a response message to the device is similar to the method in which the host sends a request message to the endpoint device and the endpoint device returns a response message to the host, and specifically includes:

[0080] When the second endpoint device sends a second request message to the BDF conversion module, the BDF conversion module obtains the second request message sent by the second endpoint device. If the second request message contains the first target host address, it is determined that the second endpoint device is the second request message sent to the host. The second request message needs to pass through the BDF conversion module to convert the second requester global domain number into the second requester local domain number. The BDF conversion module sends the converted second request message to the target host corresponding to the first target host address.

[0081] When the target host sends the second response message to the BDF conversion module, the second response message needs to pass through the BDF conversion module to convert the second responder local domain number into the second responder global domain number. The BDF conversion module sends the converted second response message to the endpoint device corresponding to the second requester global domain number.

[0082] In this embodiment, a number conversion module is used to convert the local domain number in the second request message and the second response message into a global domain number, and convert the global domain number into a local domain number, so that the endpoint device can actively send a request message to the host and receive a response message from the host, thereby reducing the difficulty and cost of implementing multi-host interoperability.

[0083] In some optional implementations, the switching chip further includes: a global domain number allocation module, a third preset number of uplink bridge devices, and a fourth preset number of downlink bridge devices;

[0084] The upstream bridge device is used to connect to the host, obtain the configuration space read and write instructions sent by the host, and after receiving the configuration space read and write instructions, send a waiting response to the host;

[0085] The global domain number allocation module is used to obtain the first device information of the downstream bridge device and the second device information of the endpoint device, generate the global domain number of the downstream bridge device according to the first device information, and generate the global domain number of the endpoint device according to the second device information;

[0086] The global domain number allocation module is connected to the upstream bridge device, and is used to obtain the configuration space read and write instructions from the upstream bridge device, obtain the local domain number of the downstream bridge device and the local domain number of the endpoint device from the configuration space read and write instructions, generate a first association relationship according to the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, generate a second association relationship according to the local domain number of the endpoint device and the global domain number of the endpoint device, generate a reply response packet according to the first device information and / or the second device information, and send the reply response packet to the host through the upstream bridge device;

[0087] The global domain number allocation module is connected to the number conversion module and is used to configure the first association relationship and the second association relationship to the number conversion module.

[0088] Specifically, Figure 3 As shown, a CPU is set in the switching chip as a global domain number allocation module, and a PCIe RC is included in the CPU. The root port of the CPU is connected to all downstream bridge devices so that the RC is connected to all downstream bridge devices through the above-mentioned upstream bridge device. These internal connections are used to transmit packets of the PCIe protocol standard and follow the routing rules of the upstream / downstream bridge devices defined by the PCIe protocol. In addition, the switching chip also includes a third preset number of upstream bridge devices and a fourth preset number of downstream bridge devices. The third preset number is an integer, and no specific number limit is made here. The fourth preset number represents multiple, and no specific number limit is made here. The upstream bridge device connected to the host root port is managed by the host RC and belongs to the host PCIe domain. It is also connected to all downstream bridge devices. The connection can transmit some PCIe packets, such as storage read and write access requests, but does not transmit configuration space read and write requests. And, this connection is not directly routed according to the upstream / downstream rules defined by PCIe, because the upstream bridge device connected to the host will not be directly regarded as an "upstream" device; and the access will perform ID conversion.

[0089] Combination Figure 5 The process of generating a global domain number is explained. When the switching chip is in an uninitialized state, the host attempts to enumerate the PCIe tree by sending configuration space read and write instructions to the upstream bridge device. The upstream bridge device is used to connect to the host, obtain the configuration space read and write instructions sent by the host, and after receiving the configuration space read and write instructions, send a waiting response to the host. The waiting response is, for example, a CRS (Configuration Request Retry Status) packet, which is used to notify the host that the device is not ready and the host needs to retry. The CPU obtains the first device information of the downstream bridge device and the second device information of the endpoint device, and generates the global domain number of the downstream bridge device based on the first device information, and generates the global domain number of the endpoint device based on the second device information. After the CPU completes the device enumeration, it notifies the upstream bridge device connected to the host, 4.1. The path between the upstream bridge device connected to the host and the internal CPU is opened. As Figure 6 As shown, the global domain number of the downstream bridge device is, for example, BDF = {1, 0, 0}, BDF = {1, 1, 0}, ..., BDF = {1, 31, 0}, and the global domain number of the endpoint device is, for example, BDF = {2, 0, 0}, BDF = {3, 0, 0}, ..., BDF = {33, 0, 0}. For details, see Figure 6 , I will not go into details here. Figure 5The PCIe switch chip from a real perspective is shown in the figure below. The bus number of the actual bridge device and endpoint device is assigned by the internal CPU. The bus number is the global bus number. The figure below is an assignment example. The bridge device and endpoint device use the global bus number to send data packets. The upstream bridge device connected to the host does not belong to the PCIe tree managed by the internal CPU. The CPU manages it using internal connections that are not defined by the PCIe protocol.

[0090] Combination Figure 7 The process of generating a global domain number is described. The CPU is connected to an upstream bridge device, and the path between the upstream bridge device connected to the host and the CPU is opened. At this time, the CPU obtains the configuration space read and write instructions sent by the host from the upstream bridge device, and obtains the local domain number of the downstream bridge device and the local domain number of the endpoint device from the configuration space read and write instructions. The CPU calculates the target device of the configuration space read and write request packet currently sent by the host according to the topology structure presented to the host as needed, and then associates and records the bus number assigned by the host with the device, specifically including: the CPU generates a first association relationship according to the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, and generates a second association relationship according to the local domain number of the endpoint device and the global domain number of the endpoint device. Then, the CPU generates a reply response packet according to the first device information and / or the second device information, finds the corresponding content of the device corresponding to the configuration space read and write instructions, and sends the reply response packet to the host through the upstream bridge device.

[0091] The CPU configures the local domain number assigned by the host, the global domain number assigned by the CPU, the first association relationship and the second association relationship to the BDF conversion module. The above process is repeated until all devices are enumerated by the host and the local domain number and the global domain number are configured.

[0092] Figure 8 A schematic diagram of the numbering of the host local domain, such as Figure 8 As shown, the local domain number of the upstream bridge device is BDF = {16, 0, 0}, the local domain number of the downstream bridge device is BDF = {17, 0, 0}, BDF = {17, 1, 0}, ..., BDF = {17, 31, 0}, and the local domain number of the endpoint device is BDF = {18, 0, 0}, BDF = {19, 0, 0}, ..., BDF = {49, 0, 0}. For details, see Figure 8, which will not be described here. The PCIe domain managed by the host RC (hereinafter referred to as the local domain) and the PCIe domain managed by the internal CPU (hereinafter referred to as the global domain) are isolated from each other and do not interact directly. However, the host cannot perceive the existence of this domain isolation. From the perspective of the host, the downstream bridge device and the downstream endpoint device are still managed by it. This implementation will be described in detail later. From the perspective of the host, the internal bridge device of the switching chip and the connected endpoint device are still managed by the host RC, but the host's configuration space read and write requests for these devices and the bus number allocation process do not directly act on these bridge devices and endpoint devices, but are intercepted and recorded by the CPU inside the switching chip. Therefore, the bus number configured by the host for each device is not the same as the bus number actually used in the data packet sent by the device, which is also an obvious detectable feature of the present application. The switching chip implementing the present invention cannot perceive the two PCIe domains of the local domain and the global domain and the domain isolation between them on the host side. The topology seen by the host is shown in the figure above. Under this topology, all downstream bridge devices are connected to the upstream bridge device connected to the upstream port, and the host cannot see the CPU and the upstream bridge device connected to the CPU. Therefore, the host still believes that the bus numbers of all bridge devices and endpoint devices are assigned by it. The figure above shows an example of assignment. These bus numbers assigned by the host are referred to as local bus numbers below. These bus numbers are not actually assigned to bridge devices and endpoint devices to produce actual effects, but are only "considered" by the host. In this way, the host indirectly assigns bus numbers to bridge devices and endpoint devices by sending configuration space read and write request packets. These configuration space read and write request packets are not sent to bridge devices and endpoint devices, but are routed to the internal CPU. The internal CPU records these assigned bus numbers and responds to these configuration space read and write request packets, so that the host "believes" that these bridge devices and endpoint devices are assigned specified bus numbers. The actual bus numbers of bridge devices and endpoint devices are assigned by the internal CPU, but when these devices send data packets to the upstream bridge device connected to the host, the bus numbers need to be converted to the corresponding local bus numbers.

[0093] Fig. 9 is a flowchart of steps 7 and 8 for generating numbers, such as Fig. 9 As shown, after initialization is completed, that is, after the first association relationship and the second association relationship are configured, the path from the host to the downstream bridge device is opened, the host can send a request message to the endpoint device, and the endpoint device can also send a response message to the host. The specific process refers to the above embodiment and will not be repeated here.

[0094] In this embodiment, the global domain number allocation module generates a global domain number for the downstream bridge device based on the first device information, and generates a global domain number for the endpoint device based on the second device information; generates a first association relationship based on the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, and generates a second association relationship based on the local domain number of the endpoint device and the global domain number of the endpoint device; so that the local domain number and the global domain number of the device can be converted subsequently through the first association relationship and the second association relationship, cross-domain interaction between any two hosts can be realized, and the difficulty and cost of implementing multi-host interoperability are reduced.

[0095] In some optional implementations, the process in which the global domain number allocation module generates the global domain number of the downstream bridge device according to the first device information and generates the global domain number of the endpoint device according to the second device information includes step A1 and step A2.

[0096] Step A1, allocating a maximum configuration space address to each downstream bridge device and each endpoint device;

[0097] Specifically, the maximum configuration space address is determined according to the number of bus numbers, device numbers, and function numbers. For example, the total number of downstream bridge devices and endpoint devices in the system is 256, that is, the total number of host devices is 256, and 256 bus numbers are required. There are 32 device numbers (Device number) on each bus, and each device number has 8 function numbers (Function number). Each function number is associated with a device. Assuming that the size of the configuration space of each device is 4KB, the maximum size of the configuration space address occupied by a device is 256*32*8*4KB=256MB.

[0098] Therefore, when it is impossible to determine how many PCIe devices each host has, the embodiment of the present application allocates a configuration space address of 256MB to each host to ensure that each device corresponds to a bus number, device number, and function number.

[0099] It can be understood that the number of bus numbers, device numbers, and function numbers in the embodiments of the present application can be set according to specific needs, and the size of the maximum configuration space address is determined according to the number of bus numbers, device numbers, and function numbers.

[0100] Step A2, establishing a correspondence between the maximum configuration space address and the bus number, device number, and function number.

[0101] Specifically, the maximum configuration space address includes multiple configuration space addresses. The bus number, device number, and function number of each device are not exactly the same. Therefore, it is necessary to establish a correspondence between the maximum configuration space address and multiple devices, that is, to establish a one-to-one correspondence between each configuration space address in the maximum configuration space address and a bus number, device number, and function number.

[0102] Assuming that the size of the configuration space of each device is 4KB and the configuration space address of each host is 256M, the configuration space address of each PCIe downstream port is [starting address, starting address + 256M]. At this time, [starting address, starting address + 4KB] can be corresponded to a combination of a bus number, device number, and function number of an AXI (Advanced eXtensible Interface) downstream port, and the next configuration space address [starting address + 4KB, starting address + 8KB] corresponds to the next combination of a bus number, device number, and function number, and so on, thereby establishing a correspondence between the bus number, device number, function number of each device and a configuration space address in the maximum configuration space address.

[0103] In some optional implementations, the global domain number allocation module obtains the configuration space read and write instructions from the upstream bridge device, including:

[0104] Read the internal registers of the upstream bridge device to obtain the configuration space read and write instructions;

[0105] The global domain number allocation module sends a reply response packet to the host through the upstream bridge device, including:

[0106] The reply response packet is configured in the internal register of the upstream bridge device, and the host obtains the reply response packet from the internal register of the upstream bridge device.

[0107] Specifically, the global domain number allocation module of the switching chip, namely the CPU, can also access the upstream bridge device connected to the host through the internal connection. This access mainly includes: receiving the configuration space read and write request sent by the host; sending a reply response packet of the configuration space read and write request to the host; reading the internal register of the bridge device to obtain information; and configuring the register of the bridge device.

[0108] Based on the above content, the CPU obtains the configuration space read and write instructions by reading the internal register of the upstream bridge device; the CPU configures the internal register of the upstream bridge device with the reply response packet, and the host obtains the reply response packet from the internal register of the upstream bridge device.

[0109] In some optional implementations, the global domain number allocation module is further configured to:

[0110] Generate the host global domain number corresponding to the root port in the host;

[0111] Obtain the number of the root port in the host through the upstream bridge device, and use the number of the root port as the host local domain number;

[0112] Generate a third association relationship between the host global domain number and the host local domain number;

[0113] The third association relationship is configured to the number conversion module.

[0114] Specifically, the global domain number allocation module, ie, the CPU, needs to generate a bus number for each root port of the host RC, and the bus number needs to be recorded in the BDF conversion module as the host global domain number.

[0115] The host's root port number also needs to be recorded in the BDF conversion module, which is called the host local domain number. When the host assigns bus numbers to downstream bridge devices and endpoint devices, the host's root port number needs to be recorded by the CPU and configured to the BDF conversion module, using the root port number as the host local domain number. Specifically, it includes:

[0116] The CPU obtains the number of the root port in the host through the upstream bridge device, and uses the number of the root port as the host local domain number; generates a third association relationship between the host global domain number and the host local domain number; and configures the third association relationship to the number conversion module.

[0117] In this embodiment, the global domain number allocation module generates a host global domain number and obtains a host local domain number, generates a third association relationship between the host global domain number and the host local domain number, and configures it to the number conversion module. The local domain number and the global domain number of the host are subsequently converted through the third association relationship, so that cross-domain interaction between any two hosts can be achieved, reducing the difficulty and cost of implementing multi-host interoperability.

[0118] In some optional implementations, a third preset number of downlink bridge devices are used to generate a fourth preset number of downlink bridge device groups;

[0119] The downstream bridge device group includes a target downstream bridge device and other downstream bridge devices, wherein the target downstream bridge device is used to be connected to the target downstream bridge device in other downstream bridge device groups through a physical link, and the other downstream bridge devices are used to be connected to the endpoint device;

[0120] The host sends a first request message to the endpoint devices connected to the remaining downstream bridge devices in the non-corresponding downstream bridge device group through the target downstream bridge device and the physical link.

[0121] Specifically, Fig.10As shown, the switching chip includes a third preset number of downlink bridge devices, and the third preset number represents a plurality.

[0122] The third preset number of downstream bridge devices is used to generate a fourth preset number of downstream bridge device groups, and each switching chip includes a downstream bridge device group. Fig.10 As shown, Fig.10 There are two switching chips in the host and endpoint device communication system shown, for example: the switching chip connected to host 0 and host 1 is switching chip 1, and the switching chip connected to host 2 is switching chip 2. Switching chip 1 includes downstream bridge device group 1, and switching chip 2 includes downstream bridge device group 2.

[0123] The downstream bridge device group includes a target downstream bridge device and remaining downstream bridge devices, wherein the target downstream bridge device is used to be connected to the target downstream bridge device in other downstream bridge device groups through a physical link, and the remaining downstream bridge devices are used to be connected to endpoint devices. For example: the target downstream bridge device in downstream bridge device group 1 is connected to the target downstream bridge device in downstream bridge device group 2 through a PCIe link. When the host 0 needs to send a first request message to the endpoint device connected to the remaining downstream bridge devices in the downstream bridge device group 2, the host 0 sends the first request message to the endpoint device connected to the non-corresponding downstream bridge device group, i.e., the remaining downstream bridge devices in the downstream bridge device group 2, through the target downstream bridge device in the downstream bridge device group 1 and the PCIe link.

[0124] In this embodiment, the target downstream bridge device is used to connect with the target downstream bridge devices in other downstream bridge device groups through a physical link, so that the host sends a request message to any endpoint device through the target downstream bridge device and the physical link. The number of interconnected hosts can be greatly increased and cross-domain interaction between any two hosts can be achieved, reducing the difficulty and cost of implementing multi-host interoperability.

[0125] In some optional implementations, the number conversion module is further used to:

[0126] Obtaining a second request message sent by the endpoint device, and if the second request message does not include the first target host address, no number conversion is performed;

[0127] The endpoint device is used to send the second request message to the endpoint device corresponding to the second request message, and obtain a third response message from the endpoint device corresponding to the second request message.

[0128] Specifically, the number conversion module of the present application, i.e., the BDF conversion module, only needs to convert the local domain number and the global domain number in the scenario of cross-domain transmission of the message, such as data communication between the host and the endpoint device. However, when PCIe point-to-point communication occurs between endpoint devices, since it does not involve cross-domain, the BDF conversion module is not required to perform BDF conversion on the number.

[0129] The BDF conversion module obtains the second request message sent by the endpoint device. If the second request message does not contain the first target host address, it can be determined that the second request message is not sent to the host, but is a communication between endpoint devices and does not involve cross-domain transmission, so no number conversion is performed.

[0130] The endpoint device is used to send the second request message to the endpoint device corresponding to the second request message, and obtain the third response message from the endpoint device corresponding to the second request message, so as to realize PCIe point-to-point communication between the endpoint devices.

[0131] In some optional implementations, the number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including:

[0132] Obtaining the host's local domain bus number, local domain device number, and local domain function number from the first requester's local domain number;

[0133] According to the third association relationship, determine the global domain bus number corresponding to the local domain bus number, the global domain device number corresponding to the local domain device number, and the global domain function number corresponding to the local domain function number;

[0134] A first requester global domain number is determined, wherein the first requester global domain number includes a global domain bus number, a global domain device number, and a global domain function number.

[0135] Specifically, the first requester's local domain number includes a local domain bus number, a local domain device number, and a local domain function number. If the local domain device number and the local domain function number are configured by the host, the device number and the function number may be different between the local domain and the global domain, so it is necessary to convert the local domain bus number, the local domain device number, and the local domain function number.

[0136] According to the third association relationship, determine the global domain bus number corresponding to the local domain bus number, the global domain device number corresponding to the local domain device number, and the global domain function number corresponding to the local domain function number, summarize the global domain bus number, global domain device number, and global domain function number, determine the global domain number of the first requester, and complete the BDF number conversion.

[0137] In this implementation, the local domain bus number, the local domain device number and the local domain function number in the local domain number of the first requester are all converted to ensure that the number conversion is accurate so that the first request message is accurately sent to the corresponding endpoint device.

[0138] In some optional implementations, the number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including:

[0139] Obtaining the host's local domain bus number, local domain device number, and local domain function number from the first requester's local domain number;

[0140] Use the local domain device number as the global domain device number and the local domain function number as the global domain function number;

[0141] Determine the global domain bus number corresponding to the local domain bus number according to the third association relationship;

[0142] A first requester global domain number is determined, wherein the first requester global domain number includes a global domain bus number, a global domain device number, and a global domain function number.

[0143] Specifically, the first requester's local domain number includes a local domain bus number, a local domain device number, and a local domain function number. If the local domain device number and the local domain function number are determined by the device itself, there is no difference between the local domain and the global domain in the device number and the function number. Therefore, there is no need to convert the local domain device number and the local domain function number, and only the local domain bus number needs to be converted. The local domain device number is directly used as the global domain device number, and the local domain function number is used as the global domain function number.

[0144] According to the third association relationship, the global domain bus number corresponding to the local domain bus number is determined. The global domain bus number, the global domain device number and the global domain function number are summarized to determine the first requester global domain number, and the BDF number conversion is completed.

[0145] In this implementation, only the local domain bus number in the local domain number of the first requester is converted to improve the efficiency of the number conversion.

[0146] In this embodiment, a host and endpoint device communication system is provided, such as Fig.10 As shown, the system includes: a first preset number of hosts, a second preset number of endpoint devices, and a switching chip, and the switching chip is the switching chip in any of the above embodiments;

[0147] The uplink port of the switching chip is used to connect to a first preset number of hosts, and the downlink port of the switching chip is used to connect to a second preset number of endpoint devices; the switching chip includes a number conversion module.

[0148] Specifically, the first preset number represents a plurality, for example, 3 or other integers, such as Fig.10 As shown, the host and endpoint device communication system includes three hosts, namely host 0, host 1 and host 2, and the second preset number represents a plurality of hosts. Fig.10 As shown, the host and endpoint device communication system includes multiple endpoint devices. Fig.10 The host and endpoint device communication system shown includes two switch chips, and the two switch chips are connected through two target downstream bridge devices and a PCIe link between the two target downstream bridge devices.

[0149] Please refer to the above embodiments for details, which will not be described in detail here.

[0150] The host and endpoint device communication system provided by this embodiment uses the number conversion module in the switching chip to convert the local domain number in various messages into the global domain number, and convert the global domain number into the local domain number. The global domain number is used as an intermediary to achieve smooth transmission of messages between the host and the endpoint device, reducing the difficulty and cost of implementing multi-host interoperability. It is only necessary to implement the interoperability between each local domain and the global domain to achieve all local domain interoperability, which can greatly reduce the combination of local domain interoperability. It solves the problem of using a large number of non-transparent bridges to achieve cross-domain communication between the host and the endpoint device, resulting in high chip complexity and high chip cost.

[0151] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A switching chip, characterized in that: The uplink port of the switching chip is used to connect to a first preset number of hosts, and the downlink port of the switching chip is used to connect to a second preset number of endpoint devices; the switching chip includes a number conversion module; The number conversion module is used to obtain the first request message sent by the host, convert the first requester local domain number in the first request message into the first requester global domain number, and send the converted first request message to the first endpoint device corresponding to the first target endpoint device address in the first request message; The number conversion module is used to obtain a first response message sent by the first endpoint device, determine whether the first requester global domain number is included in the first response message, and if included, convert the first requester global domain number in the first response message into the first requester local domain number, convert the first responder global domain number in the first response message into the first responder local domain number, and send the converted first response message to a host corresponding to the first requester local domain number.

2. The switching chip according to claim 1, characterized in that: The number conversion module is also used for: Obtain a second request message sent by the second endpoint device, and if the second request message contains the first target host address, convert the second requester global domain number in the second request message into the second requester local domain number, and send the converted second request message to the target host corresponding to the first target host address; Obtain a second response message sent by the target host, convert the second requester local domain number in the second response message into the second requester global domain number, convert the second responder local domain number in the second response message into the second responder global domain number, and send the converted second response message to the endpoint device corresponding to the second requester global domain number.

3. The switching chip according to claim 1, characterized in that: The switching chip further includes: a global domain number allocation module, a third preset number of uplink bridge devices, and a fourth preset number of downlink bridge devices; The upstream bridge device is used to connect to the host, obtain the configuration space read and write instructions sent by the host, and after receiving the configuration space read and write instructions, send a waiting response to the host; The global domain number allocation module is used to obtain the first device information of the downstream bridge device and the second device information of the endpoint device, generate the global domain number of the downstream bridge device according to the first device information, and generate the global domain number of the endpoint device according to the second device information; The global domain number allocation module is connected to the upstream bridge device, and is used to obtain the configuration space read and write instructions from the upstream bridge device, obtain the local domain number of the downstream bridge device and the local domain number of the endpoint device from the configuration space read and write instructions, generate a first association relationship according to the local domain number of the downstream bridge device and the global domain number of the downstream bridge device, generate a second association relationship according to the local domain number of the endpoint device and the global domain number of the endpoint device, generate a reply response packet according to the first device information and / or the second device information, and send the reply response packet to the host through the upstream bridge device; The global domain number allocation module is connected to the number conversion module, and is used to configure the first association relationship and the second association relationship to the number conversion module.

4. The switching chip according to claim 3, characterized in that: The global domain number allocation module obtains the configuration space read and write instruction from the upstream bridge device, including: Reading the internal register of the upstream bridge device to obtain the configuration space read and write instructions; The global domain number allocation module sends the reply response packet to the host through the uplink bridge device, including: The reply response packet is configured in an internal register of the upstream bridge device, and the host obtains the reply response packet from the internal register of the upstream bridge device.

5. The switching chip according to claim 3, characterized in that: The global domain number allocation module is also used for: Generate a host global domain number corresponding to the root port in the host; Obtaining the number of the root port in the host through the uplink bridge device, and using the number of the root port as the host local domain number; Generating a third association relationship between the host global domain number and the host local domain number; The third association relationship is configured to the number conversion module.

6. The switching chip according to claim 3, characterized in that: The third preset number of the downlink bridge devices is used to generate a fourth preset number of downlink bridge device groups; The downlink bridge device group includes a target downlink bridge device and other downlink bridge devices, wherein the target downlink bridge device is used to be connected to the target downlink bridge device in other downlink bridge device groups through a physical link, and the other downlink bridge devices are used to be connected to the endpoint device; The host sends the first request message to the endpoint devices connected to the remaining downstream bridge devices in the non-corresponding downstream bridge device group through the target downstream bridge device and the physical link.

7. The switching chip according to claim 2, characterized in that: The number conversion module is also used for: Obtaining a second request message sent by the endpoint device, and if the second request message does not include the first target host address, not performing number conversion; The endpoint device is used to send the second request message to the endpoint device corresponding to the second request message, and obtain a third response message from the endpoint device corresponding to the second request message.

8. The switching chip according to claim 5, characterized in that: The number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including: Acquire the local domain bus number, local domain device number and local domain function number of the host from the first requester local domain number; According to the third association relationship, determine the global domain bus number corresponding to the local domain bus number, the global domain device number corresponding to the local domain device number, and the global domain function number corresponding to the local domain function number; The first requester global domain number is determined, wherein the first requester global domain number includes the global domain bus number, the global domain device number, and the global domain function number.

9. The switching chip according to claim 5, characterized in that: The number conversion module converts the first requester local domain number in the first request message into the first requester global domain number, including: Acquire the local domain bus number, local domain device number and local domain function number of the host from the first requester local domain number; Using the local domain device number as the global domain device number and using the local domain function number as the global domain function number; Determine the global domain bus number corresponding to the local domain bus number according to the third association relationship; The first requester global domain number is determined, wherein the first requester global domain number includes the global domain bus number, the global domain device number, and the global domain function number.

10. A host and endpoint device communication system, characterized in that: The system comprises: a first preset number of hosts, a second preset number of endpoint devices and a switching chip, wherein the switching chip is the switching chip according to any one of claims 1 to 9; The uplink port of the switching chip is used to connect to the first preset number of hosts, and the downlink port of the switching chip is used to connect to the second preset number of endpoint devices; the switching chip includes a number conversion module.

Citation Information

Cited By

  • Bus number distribution method, device, system and equipment of switch chip and medium

    CN120238521A

  • Bus number allocation method, device, system, equipment and medium of exchange chip

    CN120238521B

  • Firmware recovery method and device in switching equipment interconnection scene, product and medium

    CN120743319A

  • Switching device, independent switching chip, structural switching chip and message transmission method

    CN121750584A