Congestion control state information processing method and device, chip, network interface card, equipment, storage medium and program product

By aggregating the congestion control status information of queue pairs, the problem of excessive hardware buffer resource consumption in RDMA communication is solved, achieving cost reduction and efficiency improvement.

CN118972329APending Publication Date: 2024-11-15SHENZHEN JAGUAR MICROSYSTEMS CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411128875.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

In existing technologies, the congestion control status information of queue pairs in RDMA communication consumes too much hardware cache resources, resulting in high costs.

Method used

By aggregating congestion control status information, the congestion control status information of each queue pair is aggregated based on the destination address, and only the aggregated information is stored, reducing the occupation of hardware cache resources.

Benefits of technology

This reduces the resource consumption of hardware module cache, lowers costs, and improves the resource utilization efficiency of hardware cache.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118972329A_ABST
    Figure CN118972329A_ABST
Patent Text Reader

Abstract

The invention relates to the field of communication, in particular to a congestion control state information processing method and device, a chip, a network interface card, computer equipment, a storage medium and a computer program product. The method comprises the following steps: determining a destination address of a queue pair; the address of aggregation congestion control state information corresponding to the destination address is inquired, and the aggregation congestion control state information is obtained by aggregating the congestion control state information of each queue pair based on the destination address; and processing the aggregation congestion control state information based on the address of the aggregation congestion control state information. By adopting the method, the occupation of cache resources of the hardware module can be reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a congestion control state information processing method, device, chip, network interface card, computer equipment, storage medium and computer program product. Background Art

[0002] RDMA (Remote Direct Memory Access) is an efficient communication technology widely used in high-performance computing and data center networks. This technology enables servers to directly access each other's memory without the intervention of the operating system, thereby significantly improving data processing speed and network efficiency.

[0003] QP (Queue Pair) is the basic unit of RDMA communication, which is used to describe the communication connection between two nodes. Each QP has an SQ (Send Queue) and an RQ (Receive Queue), which are used to send and receive data respectively. Through QP, RDMA data transmission can be directly performed between nodes without involving the data transmission process of the operating system. A pair of QPs can also be called a flow.

[0004] The congestion control algorithm needs to use the CCC (congestion control context) of the corresponding QP in execution, and change the CCC according to the new event request. Specifically, the hardware may report an event to the algorithm at moments such as when a token is requested for scheduling message sending, when a message is sent, or when a CNP (explicit congestion notification packet) is received. The algorithm runs the corresponding congestion control logic according to the event type, updates the CCC, and feeds back information such as tokens or rates to the hardware.

[0005] Like QPC (qp context), the CCC of each pair of QPs also takes up a certain amount of space, commonly 32B / 48B / 64B. When the hardware module reports an event to the algorithm, it needs to check whether the local cache already contains the CCC of the corresponding QP. If not, it is necessary to first find the CCC table entry address and load the CCC from the host memory to the local cache. The hardware module sends the CCC in the local cache to the algorithm execution module as input. The algorithm execution module runs the corresponding congestion control logic according to the event type, updates the CCC and writes it back to the hardware cache.

[0006] When there are multiple active QPs, a large amount of hardware cache resources will be occupied, which is consuming and costly. Summary of the invention

[0007] Based on this, it is necessary to provide a congestion control state information processing method, device, chip, network interface card, computer equipment, computer-readable storage medium and computer program product that can reduce the occupancy of hardware module cache resources in response to the above technical problems.

[0008] In a first aspect, the present application provides a method for processing congestion control state information, the method comprising:

[0009] Determine the destination address of the queue pair;

[0010] Querying an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address;

[0011] The aggregated congestion control state information is processed based on an address of the aggregated congestion control state information.

[0012] In one of the embodiments, the aggregated congestion control state information includes reference count information; and the processing of the aggregated congestion control state information based on the address of the aggregated congestion control state information includes:

[0013] The reference count in the reference count information of the aggregated congestion control state information is increased.

[0014] In one of the embodiments, the aggregated congestion control state information includes reference count information; and the method further includes:

[0015] In the case where the address of the aggregated congestion control state information corresponding to the destination address is not queried, determining the target storage space from the pre-applied host memory space, generating the aggregated congestion control state information based on the congestion control state information, storing the generated aggregated congestion control state information in the target storage space, and modifying the reference count information of the generated aggregated congestion control state information;

[0016] Loading the aggregated congestion control state information stored in the target storage space into a local cache;

[0017] A mapping relationship between the destination address and the aggregated congestion control state information is added.

[0018] In one of the embodiments, the aggregated congestion control state information includes reference count information; and the method further includes:

[0019] When creating a queue pair, storing the address of the aggregated congestion control state information corresponding to the queue pair into the queue pair context;

[0020] When deleting the queue pair, reducing the reference count in the reference count information of the aggregated congestion control state information;

[0021] When the reduced reference count is not greater than the target value, releasing the host memory and local cache corresponding to the aggregated congestion control state information;

[0022] The mapping relationship between the aggregate congestion control state information and the destination address is deleted.

[0023] In one of the embodiments, the aggregate congestion control state information further includes message sending constraint information; and the method further includes:

[0024] Receive message sending constraint information, and store the message sending constraint information in the aggregate congestion control state information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, update the message sending constraint information.

[0025] In one of the embodiments, the message sending constraint information is updated based on reference count information in the aggregate congestion control state information.

[0026] In a second aspect, the present application further provides a congestion control state information processing device, the device comprising:

[0027] A processing request acquisition module is used to determine the destination address of the queue pair;

[0028] A query module, configured to query an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address;

[0029] A processing module is used to process the aggregated congestion control state information based on the address of the aggregated congestion control state information.

[0030] In a third aspect, the present application further provides a chip, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0031] In a fourth aspect, the present application further provides a network interface card, comprising the chip as described in the third aspect and a plurality of interfaces, wherein the chip communicates externally through the interfaces.

[0032] In a fifth aspect, the present application also provides a computer device, comprising a central processing unit and a chip as described in the third aspect, wherein the chip is used to dispatch messages to the central processing unit or the chip itself for processing, and the central processing unit is used to process messages dispatched by the chip.

[0033] In a sixth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the steps of the above method when executed by a processor.

[0034] In a seventh aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the above method when executed by a processor.

[0035] In the above-mentioned congestion control state information processing method, device, chip, network interface card, computer equipment, storage medium and computer program product, the aggregated congestion control state information is obtained by aggregating the congestion control state information of each queue pair based on the destination address, so that the address of the aggregated congestion control state information can be queried based on the destination address and controlled, and the congestion control state information of multiple different queue pairs of the same node is aggregated based on the destination address, so that only the aggregated congestion control state information needs to be stored in the hardware cache, which reduces the resource occupation of the hardware module cache and reduces the cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] FIG1 is an application environment diagram of a congestion control state information processing method in one embodiment;

[0038] FIG2 is a block diagram of a sending node in one embodiment;

[0039] FIG3 is a flow chart of a method for processing congestion control status information in one embodiment;

[0040] FIG4 is a schematic diagram of multiple QPs on the same sending node in one embodiment;

[0041] FIG5 is a data structure diagram of aggregated congestion control state information in one embodiment;

[0042] FIG6 is a schematic diagram of a mapping table in one embodiment;

[0043] FIG7 is a flowchart of QP creation in one embodiment;

[0044] FIG8 is a flowchart of QP deletion in one embodiment;

[0045] FIG9 is a structural block diagram of a congestion control state information processing device in one embodiment;

[0046] FIG. 10 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] The congestion control state information processing method provided in the embodiment of the present application can be applied to the application environment shown in FIG1. ​​The sending node 102 communicates with the receiving node 104 through the network. The sending node 102 can communicate with the receiving node 104 through the RDMA communication mode. The QP is the basic unit of the RDMA communication mode. The QP includes a sending queue and a receiving queue, which are used to send data and receive data respectively. A pair of QPs can also be called a flow.

[0049] As shown in FIG2 , the sending node 102 includes an algorithm execution module, a hardware module, and a host memory, wherein the sending node 102 (RDMA network card) includes multiple streams (multiple different QPs). The hardware module can aggregate the congestion control state information corresponding to the multiple streams to obtain aggregated congestion control state information to reduce the total resource consumption. The aggregated congestion control state information obtained by aggregation is stored in the host memory and dynamically loaded into the cache space of the hardware module. The hardware module is used to obtain the message sending constraint information obtained by the algorithm execution module, and store the message sending constraint information in the aggregated congestion control state information for message sending.

[0050] When processing the congestion control state information, first determine the destination address (that is, the address of the receiving node) corresponding to the queue pair (flow), and then query the address of the corresponding aggregated congestion control state information based on the destination address, wherein the aggregated congestion control state information is obtained by aggregating the congestion control state information of each queue pair based on the destination address; based on the address of the aggregated congestion control state information, the aggregated congestion control state information is processed. The congestion control state information of multiple different queue pairs of the same node is aggregated based on the destination address, so that only the aggregated congestion control state information needs to be stored in the hardware cache, which reduces the resource occupation of the hardware module cache and reduces the cost.

[0051] The sending node 102 and the receiving node 104 may be terminals or servers. The terminals may be, but are not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server may be implemented as an independent server or a server cluster consisting of multiple servers.

[0052] In an exemplary embodiment, as shown in FIG3 , a method for processing congestion control state information is provided, which is described by taking the method applied to the hardware module of the sending node in FIG1 as an example, and includes the following steps 302 to 306. Among them:

[0053] S302: Determine the destination address of the queue pair.

[0054] Among them, each queue pair (QP) in this application includes a source address and a destination address, the source address is the address of the sending node, and the destination address is the address of the receiving node. This application supports multiple QPs on a sending node, and the source address of each QP is the same, and the destination address can be the same or different.

[0055] S304: Query the address of the aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating the congestion control state information of each queue pair based on the destination address.

[0056] S306: Process the aggregated congestion control state information based on the address of the aggregated congestion control state information.

[0057] The aggregated congestion control state information is obtained by aggregating the congestion control state information of multiple QPs of the same sending node, that is, when the destination addresses are the same, the congestion control state information of each queue pair is aggregated to obtain the aggregated congestion control state information.

[0058] As shown in Figure 4, for multiple QPs on the same sending node, when the destination address (which can be an IP address) is consistent and the network forwarding device (mainly switches and routers) does not enable ECMP (Equal-Cost Multi-Path) routing, their traffic paths are completely consistent. If the paths are consistent, the congestion states experienced at the same time are also consistent. The sending node can treat the flows corresponding to these QPs as a large flow, that is, adjust them as a single flow. Therefore, the congestion state information of these flows can be aggregated. For multiple flows on the same sending node, if the destination addresses are different, their traffic paths in the network are likely to be different. If the paths are inconsistent, the congestion states experienced at the same time are likely to be different. The sending nodes need to adjust separately according to their actual congestion states, and their congestion state information cannot be aggregated.

[0059] As shown in FIG. 5, FIG. 5 is a data structure diagram of the aggregated congestion control state information. In this embodiment, the aggregated congestion control state information includes metadata and data (i.e., the traditional CCC content with QP as the granularity). The metadata is transparent to the algorithm and is used to implement the flow number counting and cache message sending indication information, serving the hardware module; the data is used to store the real-time status information required by the algorithm and serve the algorithm execution module. Specifically, the metadata part is invisible to the algorithm execution module and is only used for hardware module management and maintenance. The metadata includes two domain segments, reference count information and message sending constraint information. Among them, the reference count information ref_count represents the reference count of the aggregated congestion control state information ACCC, representing the number of QPs using the aggregated congestion control state information ACCC; the message sending constraint information value represents the rate or token value fed back by the algorithm execution module last time, which is used by the hardware module as a constraint for message sending. Optionally, ref_count and value can occupy 4 bytes respectively, and the total metadata overhead is 8 bytes. In other embodiments, the number of bytes occupied by the reference count information and the message sending constraint information can be other values, which are not specifically limited here.

[0060] In this way, multiple flows with the same destination address use the same aggregate congestion control status information ACCC as the congestion control status information. When there are many traffic scenarios with the same destination address, the cache space occupied by the congestion control status information of multiple flows is smaller than before, saving hardware cache space resources and hardware costs. From the perspective of the algorithm execution module, multiple flows with the same destination address can be regarded as a "large flow". There is no need to send events for each "small flow" (different QP) separately. They can be aggregated and reported uniformly, effectively reducing the computing power requirements of the algorithm execution module and saving algorithm core costs. Multiple flows with the same destination address use the same aggregate congestion control status information ACCC as the congestion control status information. Limited hardware cache resources can cache the congestion control status information of multiple flows at the same time. When interacting with the algorithm execution module, the hardware module can greatly reduce the additional memory access overhead and swap-in and swap-out jitter caused by cache misses, reduce algorithm execution latency, and improve algorithm execution throughput. The congestion control status information of the flow with the same destination address can directly use the currently established aggregate congestion control status information ACCC, which is equivalent to being able to directly obtain the congestion information of the path as a known condition. The initial sending rate can directly use the value field value recorded in the metadata of the aggregate congestion control status information ACCC, starting in a "slow start" manner, without the need for line speed bursts to cause unnecessary congestion.

[0061] Among them, in order to realize the management of the aggregated congestion control state information, the mapping relationship between the address of each aggregated congestion control state information and the destination address is pre-established. For example, a mapping table can be used to record the mapping relationship between the address of each aggregated congestion control state information and the destination address. The essence of the mapping table is a hash table with the destination address as the key and the aggregated congestion control state information as the value. The structure of the mapping table is shown in Figure 6. The hash algorithm can be unconstrained. The table can be stored in the host memory, and the software can process the establishment, insertion, search, and deletion of each table entry in the mapping table. This is because the mapping table access only occurs when the QP is created and deleted, which is a non-critical data path. Therefore, when the QP is created and deleted, the destination address corresponding to the QP to be created or deleted can be determined, and then the mapping table can be queried through the destination address to obtain the address of the corresponding aggregated congestion control state information. The aggregated congestion control state information mapping table implemented by the hash table manages the mapping of the destination address to the address of the aggregated congestion control state information, and realizes the rapid positioning of the first aggregated congestion control state information when the QP is created.

[0062] After the address of the aggregated congestion control state information is obtained through query, the aggregated congestion control state information may be determined based on the address of the aggregated congestion control state information, and then the aggregated congestion control state information may be processed.

[0063] In one of the optional embodiments, the aggregated congestion control state information includes reference count information; based on the address of the aggregated congestion control state information, the aggregated congestion control state information is processed, including: increasing a reference count in the reference count information of the aggregated congestion control state information.

[0064] After the address of the aggregated congestion control state information is queried, it means that the congestion control state information corresponding to the QP has been stored and no space needs to be occupied for storage. It is only necessary to mark the state of the stored aggregated congestion control state information, such as modifying the reference count information of the aggregated congestion control state information, such as increasing the reference count corresponding to the reference count information, such as adding one.

[0065] In addition, in the process of creating QP, the address of the aggregated congestion control state information needs to be stored in the queue pair context, so that when processing QP, the address of the corresponding aggregated congestion control state information can be determined based on the queue pair context QPC to query the aggregated congestion control state information. After creating QP, the address accc_addr of the aggregated congestion control state information in the queue pair context QPC can be directly used to quickly access the subsequent aggregated congestion control state information ACCC.

[0066] In the above-mentioned congestion control status information processing method, the aggregated congestion control status information is obtained by aggregating the congestion control status information of each queue pair based on the destination address, so that the address of the aggregated congestion control status information can be queried based on the destination address and controlled, and the congestion control status information of multiple different queue pairs of the same node is aggregated based on the destination address, so that only the aggregated congestion control status information needs to be stored in the hardware cache, which reduces the resource occupation of the hardware module cache and reduces the cost.

[0067] In one of the optional embodiments, the aggregated congestion control state information includes reference count information; the method also includes: when the address of the aggregated congestion control state information corresponding to the destination address is not queried, determining the target storage space from the pre-applied host memory space, generating the aggregated congestion control state information based on the congestion control state information, and storing the generated aggregated congestion control state information in the target storage space, modifying the reference count information of the generated aggregated congestion control state information; loading the aggregated congestion control state information stored in the target storage space into a local cache; storing the address of the aggregated congestion control state information in the queue pair context; and adding a mapping relationship between the destination address and the aggregated congestion control state information.

[0068] When the address of the aggregated congestion control status information corresponding to the destination address is not queried, that is, when establishing a QP corresponding to the destination address, the target storage space is determined from the pre-applied host memory space, the aggregated congestion control status information is generated based on the congestion control status information, and the generated aggregated congestion control status information is stored in the target storage space. The target storage space may be pre-applied by the driver and used to store the aggregated congestion control state information. When a new QP is created and there is no existing QP with the same destination address as the newly created QP, it is necessary to generate the aggregated congestion control state information based on the congestion control state information of the QP, that is, to add a metadata field to the congestion control state information of the QP to obtain the aggregated congestion control state information, and to modify the reference count information of the generated aggregated congestion control state information, for example, to increase the reference count in the reference count information of the aggregated congestion control state information. Since it is the newly added aggregated congestion control state information, the aggregated congestion control state information stored in the target storage space is also loaded into the local cache; the address of the aggregated congestion control state information is stored in the queue pair context; and a mapping relationship between the destination address and the aggregated congestion control state information is added to facilitate subsequent queries.

[0069] For ease of understanding, FIG. 7 is a flowchart of QP creation in an embodiment. In this embodiment, when creating a QP, the hardware module checks whether the destination address has established aggregate congestion control state information by searching the IP-ACCC mapping table. If yes, the reference count ref_count of the metadata of the aggregate congestion control state information is directly atomically increased by 1, and then the address of the aggregate congestion control state information is added to the QPC (queue pair context) to support fast search. If no, the hardware module needs to perform the following operations: (1) select a space from the host memory pre-applied by the driver to store the aggregate congestion control state information, and load it into the local cache; (2) set the metadata reference count (ref_count) to 1; (3) atomically add an entry (destination address ip, aggregate congestion control state information address accc_addr) to the mapping table; (4) add the address of the aggregate congestion control state information (accc_addr) to the QPC (queue pair context) to support fast search.

[0070] In one of the optional embodiments, the aggregated congestion control status information includes reference count information; the method also includes: when creating a queue pair, storing the address of the aggregated congestion control status information corresponding to the queue pair in the queue pair context; when deleting the queue pair, reducing the reference count in the reference count information of the aggregated congestion control status information; when the reduced reference count is not greater than the target value, releasing the host memory and local cache corresponding to the aggregated congestion control status information; and deleting the mapping relationship between the aggregated congestion control status information and the destination address.

[0071] As shown in FIG8 , when deleting a QP, the hardware module finds the aggregated congestion control state information through the address of the aggregated congestion control state information recorded in the QPC (queue pair context), and atomically decrements the metadata reference count by 1 by ref_count. If ref_count is still greater than 0, it indicates that there are still other QPs using and maintaining the aggregated congestion control state information, and the current QP resources can be destroyed normally. Otherwise, this means that the current QP is the only maintainer of the aggregated congestion control state information, and the hardware module needs to perform the following operations: (1) Release the aggregated congestion control state information resources; (2) Atomically delete the corresponding entry from the mapping table; (3) Destroy the current QP resources.

[0072] In the above embodiment, the reference count ref_count field of the aggregated congestion control state information ACCC metadata area can count the number of flows associated with the same destination address and dynamically update it with the life cycle of the flow, thereby supporting the safe creation and destruction of the aggregated congestion control state information ACCC, and avoiding the problem of duplicate aggregated congestion control state information ACCC and free aggregated congestion control state information ACCC (the QP associated with the aggregated congestion control state information ACCC has been destroyed but the aggregated congestion control state information ACCC space has not been safely reclaimed).

[0073] In one of the optional embodiments, the aggregated congestion control status information also includes message sending constraint information; the method also includes: receiving the message sending constraint information, and storing the message sending constraint information to the aggregated congestion control status information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, the message sending constraint information is updated.

[0074] The message sending constraint information includes rate or token. Therefore, the aggregated congestion control state information can support the following two hardware message sending mechanisms: rate-based message sending and token-based message sending.

[0075] For the hardware module that sends messages based on rate, after obtaining the rate returned by the algorithm execution module, the hardware stores the rate information in the value field of the message sending constraint information in the aggregated congestion control status information metadata area. Multiple flows associated with the ACCC automatically perform rate averaging conversion when executing message sending: rate_per_flow=value / ref_count.

[0076] In this way, for the hardware module that sends messages based on rate, after obtaining the rate returned by the algorithm execution module, the hardware stores the rate information in the value field of the aggregate congestion control state information ACCC metadata area. Multiple flows associated with the aggregate congestion control state information ACCC can automatically perform fair rate sharing when executing message sending. The algorithm is not aware of the aggregate flow and there is no need to modify the algorithm logic.

[0077] For the hardware module that sends messages based on tokens, after obtaining the token returned by the algorithm execution module, the hardware stores the token information in the value field of the aggregate congestion control state information metadata area. The token-based sending mechanism requires two value subfields: token_total is the total number of tokens returned by the algorithm, and token_single is the number of tokens that any single flow in the aggregate flow can obtain at one time, and token_single = token_total / ref_count is a static value for a period of time. Multiple flows associated with the aggregate congestion control state information automatically perform fair token acquisition when executing message sending, and the token_total value will be updated each time a token is obtained: token_total = token_total-token_single. When all tokens are used up by the aggregate flow, the algorithm can respond to new token requests and return new token values.

[0078] In the above embodiment, for the hardware module that sends messages based on tokens, after obtaining the token returned by the algorithm execution module, the hardware stores the token information in the value field of the aggregate congestion control state information ACCC metadata area. Multiple flows associated with the aggregate congestion control state information ACCC automatically perform fair token acquisition when executing message sending. The algorithm is not aware of the aggregate flow, and there is no need to modify the algorithm logic.

[0079] In one of the optional embodiments, the message sending constraint information is updated based on the reference count information in the aggregated congestion control state information.

[0080] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0081] Based on the same inventive concept, the embodiment of the present application also provides a congestion control state information processing device for implementing the congestion control state information processing method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more congestion control state information processing device embodiments provided below can refer to the limitations of the congestion control state information processing method above, and will not be repeated here.

[0082] In an exemplary embodiment, as shown in FIG9 , a congestion control state information processing device is provided, including: a processing request acquisition module 901, a query module 902 and a processing module 903, wherein:

[0083] A processing request acquisition module 901 is used to determine the destination address of the queue pair;

[0084] A query module 902 is used to query the address of the aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating the congestion control state information of each queue pair based on the destination address;

[0085] The processing module 903 is used to process the aggregated congestion control state information based on the address of the aggregated congestion control state information.

[0086] In one of the optional embodiments, the aggregated congestion control state information includes reference count information; the processing module 903 is specifically used to increase the reference count in the reference count information of the aggregated congestion control state information.

[0087] In one of the optional embodiments, the aggregated congestion control state information includes reference count information; the device also includes: a creation module, which is used to determine the target storage space from the pre-applied host memory space when the address of the aggregated congestion control state information corresponding to the destination address is not queried, generate the aggregated congestion control state information based on the congestion control state information, and store the generated aggregated congestion control state information in the target storage space, modify the reference count information of the generated aggregated congestion control state information; load the aggregated congestion control state information stored in the target storage space into the local cache; and increase the mapping relationship between the destination address and the aggregated congestion control state information.

[0088] In one of the optional embodiments, the aggregated congestion control status information includes reference count information; the above-mentioned device also includes a deletion module, which is used to store the address of the aggregated congestion control status information corresponding to the queue pair to the queue pair context when creating a queue pair; when deleting the queue pair, reduce the reference count in the reference count information of the aggregated congestion control status information; when the reduced reference count is not greater than the target value, release the host memory and local cache corresponding to the aggregated congestion control status information; delete the mapping relationship between the aggregated congestion control status information and the destination address.

[0089] In one of the optional embodiments, the aggregated congestion control status information also includes message sending constraint information; the device also includes: an update module, used to receive the message sending constraint information, and store the message sending constraint information to the aggregated congestion control status information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, the message sending constraint information is updated.

[0090] In one of the optional embodiments, the message sending constraint information is updated based on the reference count information in the aggregated congestion control state information.

[0091] Each module in the above-mentioned congestion control state information processing device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module above.

[0092] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG10. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a congestion control state information processing method is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0093] Those skilled in the art will understand that the structure shown in FIG. 10 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0094] In an exemplary embodiment, a chip is provided, such as a data processing unit (DPU) chip, the chip including a memory and a processor, the memory storing a computer program, and the processor implementing the following steps when executing the computer program: determining a destination address of a queue pair; querying an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address; and processing the aggregated congestion control state information based on the address of the aggregated congestion control state information.

[0095] In one embodiment, the aggregate congestion control state information includes reference count information; the processor executes the computer program to process the aggregate congestion control state information based on the address of the aggregate congestion control state information, including: increasing the reference count in the reference count information of the aggregate congestion control state information.

[0096] In one embodiment, the aggregated congestion control state information includes reference count information; when the processor executes the computer program, the following steps are also implemented: if the address of the aggregated congestion control state information corresponding to the destination address is not queried, the target storage space is determined from the pre-applied host memory space, the aggregated congestion control state information is generated based on the congestion control state information, and the generated aggregated congestion control state information is stored in the target storage space, and the reference count information of the generated aggregated congestion control state information is modified; the aggregated congestion control state information stored in the target storage space is loaded into the local cache; and a mapping relationship between the destination address and the aggregated congestion control state information is added.

[0097] In one embodiment, the aggregated congestion control state information includes reference count information; when the processor executes the computer program, the following steps are also implemented: when creating a queue pair, the address of the aggregated congestion control state information corresponding to the queue pair is stored in the queue pair context; when deleting the queue pair, the reference count in the reference count information of the aggregated congestion control state information is reduced; when the reduced reference count is not greater than the target value, the host memory and local cache corresponding to the aggregated congestion control state information are released; and the mapping relationship between the aggregated congestion control state information and the destination address is deleted.

[0098] In one embodiment, the aggregated congestion control status information also includes message sending constraint information; when the processor executes the computer program, the following steps are also implemented: receiving the message sending constraint information, and storing the message sending constraint information in the aggregated congestion control status information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, the message sending constraint information is updated.

[0099] In one embodiment, the message sending constraint information involved in the execution of the computer program by the processor is updated based on the reference count information in the aggregated congestion control state information.

[0100] In one embodiment, a network interface card (NIC) is provided. The network interface card includes the chip described in the above embodiment and multiple interfaces, such as a PCI / PCIE interface, a network communication interface, etc. The chip communicates externally through the interface.

[0101] In one embodiment, a computer device is provided, characterized in that it includes a central processing unit (CPU) and a chip as described in the above embodiment, wherein the chip is used to dispatch messages to the central processing unit or the chip itself for processing, and the central processing unit is used to process the messages dispatched by the chip.

[0102] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: determining a destination address of a queue pair; querying an address of aggregated congestion control status information corresponding to the destination address, wherein the aggregated congestion control status information is obtained by aggregating congestion control status information of each queue pair based on the destination address; and processing the aggregated congestion control status information based on the address of the aggregated congestion control status information.

[0103] In one embodiment, the aggregate congestion control state information includes reference count information; the computer program implemented when executed by the processor processes the aggregate congestion control state information based on the address of the aggregate congestion control state information, including: increasing the reference count in the reference count information of the aggregate congestion control state information.

[0104] In one embodiment, the aggregated congestion control state information includes reference count information; when the computer program is executed by the processor, the following steps are also implemented: if the address of the aggregated congestion control state information corresponding to the destination address is not queried, the target storage space is determined from the pre-applied host memory space, the aggregated congestion control state information is generated based on the congestion control state information, and the generated aggregated congestion control state information is stored in the target storage space, and the reference count information of the generated aggregated congestion control state information is modified; the aggregated congestion control state information stored in the target storage space is loaded into the local cache; and a mapping relationship between the destination address and the aggregated congestion control state information is added.

[0105] In one embodiment, the aggregated congestion control state information includes reference count information; when the computer program is executed by the processor, the following steps are also implemented: when creating a queue pair, the address of the aggregated congestion control state information corresponding to the queue pair is stored in the queue pair context; when deleting the queue pair, the reference count in the reference count information of the aggregated congestion control state information is reduced; when the reduced reference count is not greater than the target value, the host memory and local cache corresponding to the aggregated congestion control state information are released; and the mapping relationship between the aggregated congestion control state information and the destination address is deleted.

[0106] In one embodiment, the aggregated congestion control status information also includes message sending constraint information; when the computer program is executed by the processor, the following steps are also implemented: receiving the message sending constraint information, and storing the message sending constraint information in the aggregated congestion control status information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, the message sending constraint information is updated.

[0107] In one embodiment, the message sending constraint information involved in the execution of the computer program by the processor is updated based on the reference count information in the aggregated congestion control state information.

[0108] In one embodiment, a computer program product is provided, including a computer program, which implements the following steps when executed by a processor: determining a destination address of a queue pair; querying an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address; and processing the aggregated congestion control state information based on the address of the aggregated congestion control state information.

[0109] In one embodiment, the aggregate congestion control state information includes reference count information; the computer program implemented when executed by the processor processes the aggregate congestion control state information based on the address of the aggregate congestion control state information, including: increasing the reference count in the reference count information of the aggregate congestion control state information.

[0110] In one embodiment, the aggregated congestion control state information includes reference count information; when the computer program is executed by the processor, the following steps are also implemented: if the address of the aggregated congestion control state information corresponding to the destination address is not queried, the target storage space is determined from the pre-applied host memory space, the aggregated congestion control state information is generated based on the congestion control state information, and the generated aggregated congestion control state information is stored in the target storage space, and the reference count information of the generated aggregated congestion control state information is modified; the aggregated congestion control state information stored in the target storage space is loaded into the local cache; and a mapping relationship between the destination address and the aggregated congestion control state information is added.

[0111] In one embodiment, the aggregated congestion control state information includes reference count information; when the computer program is executed by the processor, the following steps are also implemented: when creating a queue pair, the address of the aggregated congestion control state information corresponding to the queue pair is stored in the queue pair context; when deleting the queue pair, the reference count in the reference count information of the aggregated congestion control state information is reduced; when the reduced reference count is not greater than the target value, the host memory and local cache corresponding to the aggregated congestion control state information are released; and the mapping relationship between the aggregated congestion control state information and the destination address is deleted.

[0112] In one embodiment, the aggregated congestion control status information also includes message sending constraint information; when the computer program is executed by the processor, the following steps are also implemented: receiving the message sending constraint information, and storing the message sending constraint information in the aggregated congestion control status information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, the message sending constraint information is updated.

[0113] In one embodiment, the message sending constraint information involved in the execution of the computer program by the processor is updated based on the reference count information in the aggregated congestion control state information.

[0114] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0115] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0116] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for processing congestion control state information, characterized in that: The method comprises: Determine the destination address of the queue pair; Querying an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address; The aggregated congestion control state information is processed based on an address of the aggregated congestion control state information.

2. The method according to claim 1, characterized in that: The aggregate congestion control state information includes reference count information; and the processing of the aggregate congestion control state information based on the address of the aggregate congestion control state information includes: The reference count in the reference count information of the aggregated congestion control state information is increased.

3. The method according to claim 1, characterized in that The aggregated congestion control state information includes reference count information; the method further includes: In the case where the address of the aggregated congestion control state information corresponding to the destination address is not queried, determining the target storage space from the pre-applied host memory space, generating the aggregated congestion control state information based on the congestion control state information, storing the generated aggregated congestion control state information in the target storage space, and modifying the reference count information of the generated aggregated congestion control state information; Loading the aggregated congestion control state information stored in the target storage space into a local cache; A mapping relationship between the destination address and the aggregated congestion control state information is added.

4. The method according to claim 1, characterized in that: The aggregated congestion control state information includes reference count information; the method further includes: When creating a queue pair, storing the address of the aggregated congestion control state information corresponding to the queue pair into the queue pair context; When deleting the queue pair, obtaining aggregated congestion control state information based on a queue pair context query of the queue pair, and reducing a reference count in reference count information of the aggregated congestion control state information; When the reduced reference count is not greater than the target value, releasing the host memory and local cache corresponding to the aggregated congestion control state information; The mapping relationship between the aggregate congestion control state information and the destination address is deleted.

5. The method according to any one of claims 1 to 4, characterized in that: The aggregate congestion control state information also includes message sending constraint information; the method also includes: Receive message sending constraint information, and store the message sending constraint information in the aggregate congestion control state information, the message sending constraint information is used for the sending queue in the queue pair to send messages based on the message sending constraint information, and after the message is sent, update the message sending constraint information.

6. The method according to claim 5, characterized in that The message sending constraint information is updated based on the reference count information in the aggregate congestion control state information.

7. A congestion control state information processing device, characterized in that: The device comprises: A processing request acquisition module is used to determine the destination address of the queue pair; A query module, configured to query an address of aggregated congestion control state information corresponding to the destination address, wherein the aggregated congestion control state information is obtained by aggregating congestion control state information of each queue pair based on the destination address; A processing module is used to process the aggregated congestion control state information based on the address of the aggregated congestion control state information.

8. A chip comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A network interface card, characterized in that: The method comprises the chip as claimed in claim 8 and a plurality of interfaces, wherein the chip communicates externally through the interfaces.

10. A computer device, characterized in that: It comprises a central processing unit and a chip as claimed in claim 8, wherein the chip is used to dispatch messages to the central processing unit or the chip itself for processing, and the central processing unit is used to process the messages dispatched by the chip.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Citation Information

Cited By

  • Congestion control system and method

    CN119420700A

  • Congestion Control Systems and Methods

    CN119420700B