A memory fault handling method, processing device, and operation and maintenance platform
By clustering memory fault information, faulty areas can be identified and isolated, resolving the conflict between memory utilization and system downtime, and achieving efficient memory utilization and system stability.
Patent Information
- Application Number
- CN202311385132.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-24
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-10-24
AI Technical Summary
While existing technologies reduce the probability of computer system downtime, they also lead to a decrease in memory utilization.
By analyzing memory fault information, the location and type of faulty storage units are determined, clustering is performed, fault isolation areas are identified, and local isolation is carried out to reduce the waste of available storage units.
It improves memory utilization, avoids wasting storage units, and reduces the risk of system crashes.
Smart Images

Figure CN117608888B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of memory technology, specifically to a memory fault handling method, processing device, and operation and maintenance platform. Background Technology
[0002] Memory is an essential component of computer devices, and there is a certain probability of memory failure during use. Memory failures are divided into corrected errors (CE) and uncorrected errors (UCE). CE can be automatically repaired, while UCE cannot, and UCE is likely to cause the computer device to crash.
[0003] To reduce the probability of computer system crashes, faulty physical regions can be predicted by analyzing fault information in memory logs, allowing for the isolation of these regions. While this reduces the probability of system crashes, it also lowers memory utilization. Therefore, improving memory utilization while reducing the probability of computer system crashes has become a pressing technical problem. Summary of the Invention
[0004] This application provides a memory fault handling method, processing device, and operation and maintenance platform, which can achieve partial isolation of memory and improve memory utilization.
[0005] In a first aspect, embodiments of this application disclose a memory fault handling method, the method comprising: determining fault information of each faulty storage unit among a plurality of faulty storage units based on memory fault information of the memory; if the number of faulty storage units included in a target storage array reaches a preset threshold, determining a target physical region of the target storage array, wherein the target storage array is any one of one or more storage arrays determined based on the location information of the plurality of faulty storage units, the target physical region includes a plurality of first faulty storage units, the plurality of first faulty storage units being at least a portion of the plurality of faulty storage units, the plurality of first faulty storage units being located in the same row or column; performing clustering processing on each of the first faulty storage units in the target physical region based on the fault information of each first faulty storage unit to obtain a fault clustering result, the fault clustering result including a plurality of second faulty storage units, the plurality of second storage units being at least a portion of the plurality of first faulty storage units, the plurality of second faulty storage units satisfying a preset condition; determining a fault isolation region based on the fault clustering result, the fault isolation region being a region formed by the addresses of the plurality of second faulty storage units.
[0006] In this technical solution, each first faulty storage unit in the target physical area is clustered based on the fault information of each first faulty storage unit to obtain fault clustering results; finally, fault isolation areas are determined based on the fault clustering results. In this way, local isolation of faulty storage units can be achieved, reducing the isolation of some available storage units, avoiding the waste of storage units, and thus improving memory utilization.
[0007] In one embodiment, the fault information of each faulty storage unit includes location information, and determining the target physical region of the target storage array includes: determining the memory fault type of each first faulty storage unit based on the location information of the first faulty storage units included in the target storage array, wherein the memory fault type includes row fault type and column fault type; determining the target physical region based on the memory fault type of each first faulty storage unit, wherein the target physical region includes: a target row physical region corresponding to the row fault type, or a target column physical region corresponding to the column fault type, wherein the multiple first faulty storage units included in the target row physical region are located in the same row, and the multiple first faulty storage units included in the target column physical region are located in the same column.
[0008] In this technical solution, the memory fault type is determined by the location information of the faulty storage unit, and then the target physical region is determined according to the memory fault type. The target physical region is a row or a column of the target storage array. Clustering is performed after the target physical region is determined, which can improve the efficiency of determining the fault isolation region.
[0009] In one embodiment, the fault information of each fault storage unit further includes time information. The step of clustering each first fault storage unit in the target physical area based on the fault information of each first fault storage unit to obtain a fault clustering result includes: clustering the first fault storage units in the target physical area based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit to obtain a fault clustering result for the target physical area.
[0010] In this technical solution, clustering processing is performed on the first fault storage unit in the target physical area by setting clustering parameters, and the fault clustering results in the target physical area can be obtained. The clustering processing is to obtain the local area that needs to be isolated, so as to facilitate the local isolation of the fault isolation area.
[0011] In one embodiment, the step of clustering the first fault storage units within the target physical region based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit to obtain the fault clustering result of the target physical region includes: for any target first fault storage unit in the target physical region, determining the time difference between the target first fault storage unit and other first fault storage units in the target physical region based on the time information of the target first fault storage unit; the target first fault storage unit is any one of the first fault storage units in the target physical region; determining the distance difference between the target first fault storage unit and other first fault storage units in the target physical region based on the location information of the target first fault storage unit; determining the set of fault storage units corresponding to each first fault storage unit included in the target physical region based on the clustering parameters, the time difference, and the distance difference; determining the fault clustering result of the target physical region based on the set of fault storage units corresponding to each first fault storage unit, wherein the fault clustering result includes multiple second fault storage units, and the multiple second fault storage units satisfy preset conditions.
[0012] In this technical solution, the clustering is determined based on the time information, location information, and clustering parameters of the faulty storage unit. The fault information of the faulty storage unit can more accurately determine the fault clustering result.
[0013] In one embodiment, the plurality of second fault storage units satisfying preset conditions include: the number of the plurality of second fault storage units satisfies the clustering density set in the clustering parameters; the time difference between any two second fault storage units in the plurality of second fault storage units is less than the time threshold set in the clustering parameters; and the distance difference between the cluster center of the fault clustering result and any one of the second fault storage units in the plurality of second fault storage units is less than the distance threshold set in the clustering parameters.
[0014] This technical solution mainly describes the conditions that the clustering results must meet. By using these conditions, the fault isolation area can be determined more accurately.
[0015] In one embodiment, determining the fault isolation region based on the fault clustering result includes: if the fault type of the second fault storage unit included in the fault clustering result is the row fault type, then determining the fault isolation region based on the minimum and maximum values of the column addresses of each second fault storage unit; if the fault type of the second fault storage unit included in the fault clustering result is the column fault type, then determining the fault isolation region based on the minimum and maximum values of the row addresses of each second fault storage unit.
[0016] This technical solution mainly describes how the fault isolation area is determined based on row address or column address, which can accurately determine the specific location information of each fault isolation area.
[0017] In one embodiment, if the ratio between the number of storage units included in the fault isolation region and the total number of first storage units included in the target physical region is greater than a ratio threshold, then the target physical region is designated as the fault isolation region, where the target physical region is an entire row or column of the target storage array.
[0018] This technical solution mainly describes how, when performing regulatory isolation, the isolation range can be expanded according to the scope of the isolation area to enhance memory protection and thus improve memory health.
[0019] In one embodiment, the memory fault information is obtained from a central processing unit (CPU); the method further includes sending the fault isolation region to the CPU, the fault isolation region being used to trigger the CPU to isolate the fault isolation region.
[0020] In this technical solution, the memory fault isolation method is mainly applied to the baseboard management controller, that is, the computer device itself, to protect the memory of the computer device.
[0021] In one embodiment, the memory fault information is obtained from the operation and maintenance platform; the method further includes: sending the fault isolation area to the operation and maintenance platform, so that the operation and maintenance platform notifies the central processing unit to isolate the fault isolation area.
[0022] This technical solution mainly applies memory fault isolation methods to the operation and maintenance platform, and uses a third party to analyze memory faults, thereby achieving memory protection.
[0023] Secondly, embodiments of this application disclose a memory fault handling apparatus, the apparatus comprising:
[0024] The determination unit is used to determine the fault information of each faulty storage unit among multiple faulty storage units based on the memory fault information of the memory.
[0025] The determining unit is further configured to determine a target physical region of the target storage array if the number of faulty storage units included in the target storage array reaches a preset threshold, wherein the target storage array is any one of one or more storage arrays determined based on the location information of the plurality of faulty storage units, and the target physical region includes a plurality of first faulty storage units, wherein the plurality of first faulty storage units are at least some of the faulty storage units, and the plurality of first faulty storage units are located in the same row or column;
[0026] The processing unit is configured to perform clustering processing on each of the first fault storage units in the target physical area based on the fault information of each first fault storage unit, and obtain a fault clustering result. The fault clustering result includes a plurality of second fault storage units, which are at least some of the first fault storage units among the plurality of first fault storage units, and the plurality of second fault storage units satisfy preset conditions.
[0027] The determining unit is further configured to determine a fault isolation region based on the fault clustering result, wherein the fault isolation region is the region formed by the addresses of the plurality of second fault storage units.
[0028] Thirdly, embodiments of this application disclose a computer device including a processor adapted to implement one or more computer programs; and a computer-readable storage medium storing one or more computer programs adapted to be loaded by the processor and executed in accordance with the memory fault handling method of the first aspect described above.
[0029] Fourthly, embodiments of this application disclose an operation and maintenance platform, which includes a processor adapted to implement one or more computer programs; and a computer-readable storage medium storing one or more computer programs, the one or more computer programs being adapted to be loaded by the processor and executed in accordance with the memory fault handling method of the first aspect described above.
[0030] Fifthly, this application discloses a computer-readable storage medium storing one or more computer programs adapted for loading by a processor and executing the resource management method of the first aspect described above. Attached Figure Description
[0031] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0032] Figure 1 This is a system architecture diagram of a memory fault handling system disclosed in an embodiment of this application;
[0033] Figure 2 This is a system architecture diagram of another memory fault handling system disclosed in the embodiments of this application;
[0034] Figure 3 This is a memory diagram disclosed in an embodiment of this application;
[0035] Figure 4 This is a flowchart illustrating a memory fault handling method disclosed in an embodiment of this application;
[0036] Figure 5 This is a schematic diagram of a fault clustering result disclosed in an embodiment of this application;
[0037] Figure 6 This is a flowchart illustrating another memory fault handling method disclosed in an embodiment of this application;
[0038] Figure 7a This is a flowchart illustrating another memory fault handling method disclosed in the embodiments of this application;
[0039] Figure 7b This is a flowchart illustrating another memory fault handling method disclosed in the embodiments of this application;
[0040] Figure 8 This is a schematic diagram of the structure of a memory fault handling device disclosed in an embodiment of this application;
[0041] Figure 9 This is a schematic diagram of the structure of a processing device disclosed in an embodiment of this application. Detailed Implementation
[0042] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0043] To better understand the technical solutions provided in the embodiments of this application, the scenarios of the embodiments of this application will be introduced first.
[0044] Memory is the storage component of a computer device, used to store data processed by the processor and data exchanged with external storage devices such as hard drives. Memory failures have a certain probability of occurring during use, and these failures are categorized into corrected errors (CE) and uncorrected errors (UCE). CE-level failures can be automatically corrected, while UCE-level failures cannot be automatically corrected and typically lead to system crashes.
[0045] When a memory fault completes (CE) occurs, the system collects information about the location of the fault, including the corresponding memory rank, memory device, storage array group, memory row, and storage column. Based on this information, the fault type can be determined, which is categorized into four types: cell fault, row fault, column fault, and bank fault. A cell fault refers to multiple CEs occurring at a single cell location; a row fault refers to multiple CEs occurring at different locations within a single row; a column fault refers to multiple CEs occurring at different locations within a single column; and a bank fault can also occur across multiple rows and columns. When these faults occur within a specific bank of memory, the probability of a fault completes (CE) occurring within that bank is relatively high. If these fault types can be identified and isolated in advance, subsequent CEs can be prevented, further preventing computer equipment downtime from impacting business operations.
[0046] When multiple memory failures occur, certain physical areas of the memory may exhibit different failure types, such as current failure, column failure, or storage array failure. To reduce the probability of system crashes caused by UCE-level failures, the types of memory failures are analyzed. If a physical area of the memory is determined to have a specific failure type (e.g., column failure, row failure, array failure) based on the current failure, that physical area of the memory is isolated. However, this method of isolating entire rows or columns results in a significant waste of memory space.
[0047] Therefore, to avoid wasting internal space, this application discloses a memory fault handling method, which includes: ① determining the fault information of each faulty storage unit among multiple faulty storage units based on memory fault information; ② if the number of faulty storage units included in the target storage array reaches a preset threshold, then determining the target physical region of the target storage array, wherein the target storage array is any one of one or more storage arrays determined based on the location information of the multiple faulty storage units, the target physical region includes multiple first faulty storage units, the multiple first faulty storage units are at least some of the multiple faulty storage units, and the multiple first faulty storage units are located in the same row or column; ③ based on the fault information of each first faulty storage unit, performing clustering processing on each first faulty storage unit in the target physical region to obtain a fault clustering result, the fault clustering result including multiple second faulty storage units, the multiple second storage units are at least some of the multiple first faulty storage units, and the multiple second faulty storage units satisfy preset conditions; ④ determining a fault isolation region based on the fault clustering result, the fault isolation region being the region formed by the addresses of the multiple second faulty storage units. This method first determines the location information of the faulty storage unit, then clusters the faulty storage units in the corresponding target physical area based on the location information, and determines the fault isolation area based on the obtained fault clustering results. This approach can achieve local isolation of faulty storage units, reduce the isolation of some available storage units, avoid the waste of storage units, and thus improve memory utilization and achieve a healthy balance between memory and capacity.
[0048] like Figure 1 The diagram shown illustrates the architecture of a memory fault handling system provided in this embodiment. The system includes a central processing unit (CPU) 101, a baseboard management controller (BMDC) 102, and memory 103. The CPU primarily collects memory fault information and sends it to the BMDC when a memory fault occurs, and isolates the faulty region based on a fault isolation request received from the BMDC. The memory fault handling system provided in this embodiment can be a computer device. This computer device can be, for example, a server, computer, PC, mobile phone, or other computing device with memory. The server can be a general-purpose server or a GPU server.
[0049] The baseboard management controller, also known as a processing device, analyzes faulty memory cells based on received memory fault information sent by the CPU to determine the location information of the faulty memory cells, clusters faulty memory cells within the physical area to obtain fault clustering results, determines fault isolation areas based on the fault clustering results, and sends fault isolation requests to the CPU. Optionally, the memory fault handling system also includes a memory chip (…). Figure 1 (Not shown in the image), this memory chip is used to store the processor firmware.
[0050] In one possible implementation, the CPU runs processing firmware and an operating system (OS). When a memory failure occurs, the processing firmware collects memory fault information and sends it to the baseboard management controller. The baseboard management controller then determines the fault isolation region using a clustering method and sends a fault isolation request and the fault isolation region to the processing firmware. The processing firmware then reports the fault isolation request to the OS, which then performs local isolation of the memory fault isolation region.
[0051] In optional embodiments, the fault isolation request and the fault isolation region can be directly sent to the processing firmware, which then performs partial isolation of the fault isolation region in the memory; or the fault isolation request and the fault isolation region can be directly sent to the OS, which then performs partial isolation of the fault isolation region in the memory based on the fault isolation request.
[0052] For example, processor firmware (also called processor firmware program) can be firmware, basic input output system (BIOS), management engine (ME), microcode, or intelligent management unit (IMU), etc. It should be noted that the specific form of the processor firmware in this application embodiment is not limited; the above is merely illustrative. In the following embodiments, BIOS is used as an example for explanation.
[0053] In one possible implementation, the baseboard management controller can maintain and manage the memory.
[0054] For example, a baseboard management controller may include a memory operation state management unit, a management system in a management chip outside the processor, a baseboard management controller (BMC), or a system management mode (SMM), etc. It should be noted that the specific form of the baseboard management controller in this application is not limited; the above is merely illustrative. In the following embodiments, a baseboard management controller comprising a BMC is used as an example for explanation.
[0055] It should be noted that different computer devices may use different names for BMC. For example, some computer devices call it BMC, some call it iLO, and others call it iDRAC. Regardless of whether it is called BMC, iLO, or iDRAC, it can be understood as the BMC in the embodiments of this application.
[0056] Optionally, in addition to clustering the memory faults using the baseboard management controller to determine the fault isolation area, the memory fault handling method provided in this application embodiment can also send the memory fault information to the operation and maintenance platform, and then the operation and maintenance platform processes the memory fault information to determine the fault isolation area; please refer to Figure 2 This is an architecture diagram of another memory fault handling system disclosed in an embodiment of this application. The memory fault handling system includes a server 201, an operation and maintenance platform 202, and a processing device 203. The processing device 203 may include a baseboard management controller or other controllers or processors with computing capabilities. The operation and maintenance platform 202 and the processing device 203 may be two separate devices or the same device, i.e., the processing device 203 is deployed within the operation and maintenance platform.
[0057] In one possible implementation, if the operation and maintenance platform 202 and the processing device 203 are the same device, then the corresponding memory fault handling method includes: when the memory of server 201 fails, it collects memory fault information and sends it to the operation and maintenance platform 202. Then, after the operation and maintenance platform 202 calls its internal processing device to determine the fault isolation region through a clustering method, it sends the fault isolation request and the fault isolation region to server 201. Server 201 then notifies the OS to perform local isolation of the memory fault isolation region.
[0058] In another possible implementation, if the operation and maintenance platform 202 and the processing device 203 are two separate devices, then the corresponding memory fault handling method includes: when the memory of server 201 fails, it collects memory fault information and sends it to the operation and maintenance platform 202. The operation and maintenance platform 202 sends a memory diagnostic request to the processing device 203. The processing device 203 processes the memory fault information using a clustering method to obtain a fault isolation region. The fault isolation region is returned to the operation and maintenance platform 202, which then sends a fault isolation request and the fault isolation region to server 201 so that server 201 can perform isolation processing.
[0059] In some alternative embodiments, in addition to sending memory fault information to a processing device (e.g., ... Figure 1 The baseboard management controller or Figure 2 In addition to determining the fault isolation area using the processing device 203, the OS or BIOS itself can also be used to process memory fault information to determine the isolation area.
[0060] For the memory structure, please refer to Figure 3 , Figure 3 This is an exemplary memory diagram provided in an embodiment of this application. Memory, also known as internal memory or main memory, is installed in memory slots on the motherboard of a computer device. Memory communicates with the CPU through memory channels.
[0061] Memory has at least one memory rank, such as Figure 3 The diagram illustrates a dual in-line memory (DIMM) module. A DIMM includes memory column 0 and memory column 1. Each memory column is located on one face of the memory module, and each column includes multiple memory chips (devices). Figure 3 The image shows 16 memory chips, namely [memory chip 00, memory chip 01, ..., memory chip 15].
[0062] Each memory chip comprises multiple bankgroups, and each bankgroup comprises multiple banks, such as... Figure 3 The storage arrays shown are 0, 1, etc. Each storage array consists of multiple storage cells, each storage cell having a row address and a column address. In other words, a storage array is divided into multiple rows and multiple columns.
[0063] To describe a storage unit, consider one of its storage units, such as... Figure 3As shown, storage array 5 exemplarily illustrates its included storage cells. The storage array 5 is divided into 8 storage rows and 8 storage columns, with each cell representing a storage cell. This application embodiment does not limit the number of storage rows or columns of the storage array; in practical applications, a storage array will have more storage cells.
[0064] In one possible implementation, memory can be divided into memory particles, memory array groups, memory arrays, memory rows / columns, and memory cells from top to bottom. The addresses of memory particles, memory array groups, memory arrays, memory rows / columns, and memory cells are called physical addresses. In this partitioning method, the granularity of memory particles, memory array groups, memory arrays, memory rows / columns, and memory cells decreases sequentially. The granularity of memory rows is the same as that of memory rows, while the granularity of memory arrays is larger than that of memory rows and memory columns.
[0065] In another possible implementation, the CPU can divide the memory chips into multiple memory pages based on a paging mechanism. Each memory page corresponds to multiple storage units in the memory, and these storage units are composed of multiple storage units in multiple memory arrays. The address of the memory page is called the system address (a virtual address), and it needs to be translated into a real physical address. Taking a 32GB, x4 DDR4 memory chip as an example, one page corresponds to 4KB, 32 pages per row correspond to 128KB of memory, and 131,072 pages per column correspond to 512MB of memory.
[0066] It should be specifically noted that the data involved in the memory fault handling process in this application embodiment, such as memory fault information of the storage array, is relevant. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use, and processing of related data must comply with relevant laws, regulations, and standards, and conform to the principles of legality, legitimacy, and necessity, without involving the acquisition of data types prohibited or restricted by laws and regulations. In some optional embodiments, the relevant data involved in this application embodiment is obtained after separate authorization from the object. In addition, when obtaining separate authorization from the object, the purpose of the relevant data is indicated to the object.
[0067] The memory fault handling method proposed in the embodiments of this application will be described in detail below.
[0068] Please see Figure 4 This is a flowchart illustrating a memory fault handling method disclosed in an embodiment of this application. This memory fault handling method can be executed by a processing device, which can be... Figure 1 The baseboard management controller 102 in the middle can also Figure 2The memory fault handling method may include, but is not limited to, the following steps: (The steps are provided in the original text, but are not limited to, the operation and maintenance platform 202 or the processing device 203 independent of the operation and maintenance platform 202.)
[0069] S401: Determine the fault information of each faulty storage unit among multiple faulty storage units based on the memory fault information of the memory.
[0070] In one possible implementation, such as Figure 1 The computer equipment shown or such Figure 2 During the operation of the server, the computer device or the server's CPU records the memory usage, i.e. memory fault information. The memory fault information can include fault information for each faulty storage unit, including the location information of the faulty storage unit and the time information of the occurrence of each faulty storage unit.
[0071] Based on memory fault information, the fault information of each faulty storage unit can be determined, and based on the location information of each faulty storage unit, the number of faulty storage units that have failed in each storage array can also be determined.
[0072] For example, the location information for each failed storage cell includes first location information and second location information. The first location information includes the memory identifier (Dimm ID), memory column identifier (Rank ID), memory chip identifier (Device ID), storage array group identifier (BankGruop ID), and storage array identifier (Bank ID). The second location information includes the storage row identifier (row) and storage column identifier (column ID). The first location information can be used to determine which storage arrays have failed and the number of failed storage cells in each storage array.
[0073] After determining the fault information of each faulty storage unit, we can use the first location information of each faulty storage unit to determine which faulty storage units belong to the same storage array. If the first location information of two faulty storage units is exactly the same, it means that these two faulty storage units occurred in the same storage array. Furthermore, if the faulty storage units occurred at different times, it can be understood that two faults occurred in the storage array. Additionally, if the storage array identifier in the first location information of two faulty storage units is the same, but any other identifier is different, then it can be determined that these two faulty storage units do not belong to the same storage array. Assuming there are a total of N faulty storage units, based on the first location information, we can determine which faulty storage units belong to the same storage array. If M faulty storage units out of the N faulty storage units belong to the same storage array, where M is less than or equal to N, then we determine the fault time of these M faulty storage units, grouping faulty storage units with the same time as one, thus achieving deduplication. Finally, we can determine the number of faulty storage units included in each storage array.
[0074] Since memory can fail repeatedly during use, we need to process memory fault information and isolate or repair it to prevent computer equipment or server system crashes. It's important to note that the fault storage units mentioned in this embodiment are repairable fault storage units. Unrepairable fault storage units can easily cause computer equipment system crashes. Therefore, we need to use repairable fault storage units to predict in advance which areas may experience unrepairable failures, and then isolate or repair them.
[0075] In one possible implementation, memory fault information can be reported proactively by the memory or sent by the memory after receiving a request to obtain it. For example, after the baseboard management controller sends a request to the central processing unit to obtain memory fault information, the memory sends the memory fault information to the central processing unit, and the central processing unit then reports the memory fault information to the baseboard management controller.
[0076] After obtaining memory fault information, it can be analyzed based on time windows. For example, a first time window can be used to analyze memory fault information occurring within that window. The first time window refers to the total duration from a past point in time to the current point in time. Analyzing memory fault information based on time windows can reduce the possibility of system downtime for computer equipment or servers. For example, the first time window could be 24 hours, 48 hours, etc. The first time window should not be too long or too short. If it is too long, faults cannot be detected and isolated in time; if it is too short, it is difficult to pinpoint the fault area, leading to repeated executions and wasted resources. Optionally, memory fault information can also be analyzed by storage array, that is, memory fault information for different storage arrays can be obtained based on their identifiers.
[0077] For example, if the current time is 2022 / 10 / 2 / 12:12:12, and the first time window is 24 hours, then the determined memory fault information includes memory faults that occurred between 2022 / 10 / 1 / 12:12:12 and 2022 / 10 / 2 / 12:12:12. As shown in Table 1, each memory fault information includes time information, memory column identifier, memory chip identifier, storage array group identifier, storage array identifier, storage row identifier, and storage column identifier.
[0078] As shown in Table 1, faults 1-3 and faults 5-7 belong to the same memory fault information in storage array 5. This leads to the conclusion that storage array 5 contains 6 faulty memory cells, and storage array 2 contains 1 faulty memory cell. For any one of these faulty memory cells, the corresponding fault information can be determined. For example, in Table 1, faulty memory cell 1 has the following first location information: memory column 0, memory chip 5, storage array group 2, storage array 5; and the second location information: row 10005, column 512. The time of the fault is 2022 / 10 / 1 / 12:12:18.
[0079] Table 1
[0080]
[0081]
[0082] In another possible implementation, when analyzing memory fault information by storage array, memory fault information for each storage array can be obtained according to a preset number. Specifically, a counter is set in the CPU. If the number of fault messages for the same storage array reaches the preset number, these fault messages are reported to the baseboard management controller. The preset number can be 10, 20, etc., and can be determined based on the frequency of memory faults in the past. This application embodiment does not limit the value of the preset number. For example, if the preset number is 10 and the current time is 2022 / 10 / 2 12:12:12, then the 10 memory fault messages of the specified storage array before 2022 / 10 / 2 12:12:12 are obtained.
[0083] S402: If the number of faulty storage units included in the target storage array reaches a preset threshold, then the target physical area of the target storage array is determined, wherein the target storage array is any one of one or more storage arrays determined based on the location information of multiple faulty storage units, and the target physical area includes multiple first faulty storage units, wherein the multiple first faulty storage units are at least some of the multiple faulty storage units, and the multiple first faulty storage units are located in the same row or column.
[0084] In one possible implementation, after determining the number of faulty storage units included in multiple storage arrays, any one of the multiple storage arrays is taken as the target storage array. It is then determined whether the number of faulty storage units in the target storage array reaches a preset threshold. If the number of faulty storage units in the target storage array is greater than or equal to the preset threshold, it is determined that the target storage array needs to be processed. The target physical region of the faulty storage units in the target storage array is determined. This target physical region is a row or column determined based on multiple faulty storage units. The target storage array may include multiple target physical regions, that is, multiple rows or multiple columns of the target storage array.
[0085] In one possible implementation, the target physical region of the faulty storage units within the target storage array is determined based on the memory fault type of each faulty storage unit. The specific process is as follows: determine the second location information of each faulty storage unit included in the target storage array; determine the memory fault type corresponding to each faulty storage unit based on the second location information; and then determine the target physical region based on the memory fault type. The target physical region includes multiple first faulty storage units, which are at least some or all of the multiple faulty storage units (determined based on memory fault information).
[0086] In this embodiment, memory fault types include row fault types and column fault types. If multiple faulty storage cells appear in the same row but different columns in the target storage array, these faulty storage cells are determined to belong to the row fault type; if multiple faulty storage cells appear in the same column but different rows in the same storage array, these faulty storage cells are determined to belong to the column fault type.
[0087] Then, the target physical region is determined based on the memory fault type of the faulty storage unit. The target physical region includes: the target row physical region corresponding to the row fault type, or the target column physical region corresponding to the column fault type. If the target physical region is a target row physical region, the multiple first faulty storage units included in the target row physical region are located in the same row; if the target physical region is a target column physical region, the multiple first faulty storage units included in the target column physical region are located in the same column.
[0088] Based on the above explanation, it can be understood that if at least two faulty storage cells appear in a row, and there are no other faulty storage cells in the column corresponding to the faulty storage cells, then these at least two faulty storage cells can be identified as a row fault type, and this row is then used as the target physical region. For example, if the third row of the target storage array has 5 faulty storage cells located in columns 2, 3, 5, 8, and 10, and there are no other faulty storage cells in these 5 columns, then these 5 faulty storage cells are identified as a row fault type, and the third row is used as a target physical region. Here, "at least two" means two or more.
[0089] Optionally, if a row contains a faulty memory cell and the corresponding column contains other faulty memory cells, then the faulty memory cell is classified as a column fault. If a row contains a faulty memory cell and the corresponding column does not contain other faulty memory cells, then the faulty memory cell can be classified as either a column fault or a row fault. In this case, the faulty memory cell can be left unprocessed, indicating that the situation is not yet sufficient to threaten memory usage, so it does not need to be addressed for the time being.
[0090] If at least two faulty memory cells appear in a column, and there are no other faulty memory cells in the row corresponding to the faulty memory cells, then these two faulty memory cells can be identified as a column fault type, and this column can be used as the target physical region. For example, if there are 6 faulty memory cells in the eighth column of the target storage array, and these 6 faulty memory cells are located in rows 2, 3, 5, 8, 10, and 12 respectively, and there are no other faulty memory cells in other positions in these 6 rows, then these 6 faulty memory cells can be directly identified as a column fault type, and the eighth column can be used as a target physical region.
[0091] Optionally, if a faulty memory cell appears in a column, and other faulty memory cells exist in the row corresponding to that faulty memory cell, then that faulty memory cell is classified as a row fault. If a faulty memory cell appears in a column, and no other faulty memory cells exist in the row corresponding to that faulty memory cell, then that faulty memory cell can be classified as either a column fault or a row fault. In this case, the faulty memory cell can be left unprocessed, indicating that the situation is not yet sufficient to threaten memory usage, so it does not need to be processed for the time being.
[0092] In some optional embodiments, a faulty storage cell may belong to both row fault type and column fault type. For example, there are 5 faulty storage cells in the third row of the target storage array, which are located in columns 2, 3, 5, 8 and 10 respectively. At the same time, there are two more faulty storage cells in other positions in column 8. In this case, the target physical area has a third row and an eighth column.
[0093] In another possible implementation, if the number of faulty storage cells in the target storage array is less than a preset threshold, then the target storage array does not need to be processed. This indicates that the situation is not serious enough to threaten memory usage, so no action is taken for the time being. The preset threshold is a custom value, such as 10 or 20.
[0094] In one possible implementation, to improve memory health, the isolation region can be expanded to isolate a larger area. When the number of faulty storage units in the target physical region reaches a certain threshold, the isolation region can be increased. Specifically, it can be defined as follows: if the ratio between the number of faulty storage units in the target physical region and the total number of storage units in the target physical region is greater than a ratio threshold, then clustering is not required, and the target physical region can be directly used as a fault isolation region. The target physical region is an entire row or column in the storage array. If the ratio between the number of faulty storage units in the target physical region and the total number of storage units in the target physical region is less than or equal to the ratio threshold, then step S403 is executed. This ratio threshold is a percentage, which can be set to 0.5 or other values. This value can be customized, and this embodiment does not limit it.
[0095] For example, if the target physical region is column 13, which has a total of 100 storage units, and 60 of them are faulty storage units, the ratio is 0.6. If the ratio threshold is 0.5, then this column can be directly used as a fault isolation region, reducing the clustering steps and improving the speed.
[0096] S403: Based on the fault information of each first fault storage unit, perform clustering processing on each first fault storage unit in the target physical area to obtain fault clustering results. The fault clustering results include multiple second fault storage units, which are at least some of the multiple first fault storage units, and the multiple second fault storage units meet preset conditions.
[0097] In one possible implementation, the clustering direction can be determined by the memory fault type or the target physical region type. For example, when the faulty memory cell is a row fault, the faulty memory cells in that row are clustered according to the column direction; when the faulty memory cell is a column fault, the faulty memory cells in that row are clustered according to the row direction. Similarly, if the target physical region is a target row physical region, the faulty memory cells in that row are clustered according to the column direction; if the target physical region is a target column physical region, the faulty memory cells in that column are clustered according to the row direction.
[0098] For example, if there are N faulty memory cells in a row and the columns are from 1 to N, clustering the faulty memory cells in the row according to the column direction means clustering from column 1 to column N.
[0099] After determining the target physical region and clustering direction, the first faulty storage units within the target physical region are clustered based on clustering parameters, the location information of each first faulty storage unit, and the time information of each first faulty storage unit, resulting in the clustering results for the target physical region. The clustering parameters include the set cluster density, time threshold, and distance threshold. For example, cluster density refers to the minimum number of faulty storage units contained in each cluster after clustering within a certain time period; it can be set to 5 or 10, and this value can be flexibly adjusted. The time threshold refers to a time period, such as 12 hours or 24 hours. The distance threshold refers to the straight-line distance between faulty storage units, and can be set to 3, 4, 5, etc. The clustering parameters can be set according to specific circumstances and are not limited here.
[0100] Taking the first faulty storage unit in the target physical region as an example, the first faulty storage unit is any first faulty storage unit in the target physical region. In the actual clustering process, clustering can start from the first faulty storage unit in the target physical region. For example, there are 8 faulty storage units in the 3rd row, and the corresponding columns are [2,3,8,10,11,12,14,19]. When clustering, the faulty storage unit in the 2nd column is first taken as the first faulty storage unit and then clustered from left to right.
[0101] The clustering process includes: first, determining the time difference between the target first faulty storage unit and other first faulty storage units in the target physical area based on the time information of the target first faulty storage unit; then, determining the distance difference between the target first faulty storage unit and other first faulty storage units in the target physical area based on the location information of the target first faulty storage unit; then, determining multiple sets of faulty storage units in the target physical area based on the time difference, distance difference, and clustering parameters; and finally, determining one or more fault clustering results in the target physical area based on the multiple sets of faulty storage units.
[0102] During clustering, the number of first faulty storage units included in the target physical region determines the number of faulty storage unit sets. However, these sets must meet preset conditions to be considered as fault clustering results. The fault clustering results include multiple second faulty storage units, which are at least a subset of the multiple first faulty storage units. The multiple second faulty storage units included in the clustering results must satisfy the preset conditions.
[0103] The clustering results include multiple second fault storage units that meet preset conditions: 1. The number of the multiple second fault storage units meets the cluster density set in the clustering parameters, that is, the number of second fault storage units included in each clustering result must reach a certain value, such as 10 or 20, etc. This application embodiment does not limit this value; 2. The time difference between any two second fault storage units is less than the time threshold set in the clustering parameters. For example, if the time threshold is set to 12 hours, and the time interval between second fault storage unit 1 and second fault storage unit 2 is 13 hours, then the condition is not met; 3. The distance difference between the cluster center of the clustering result and any one of the multiple second fault storage units is less than the distance threshold set in the clustering parameters. For fault storage units in the same row, the distance difference is the absolute value of the column difference.
[0104] Furthermore, multiple sets of faulty storage units in the target physical region are determined based on time difference, distance difference, and clustering parameters. Specifically, if there exists a first faulty storage unit in the target physical region whose time difference from the target first faulty storage unit is less than the time threshold set in the clustering parameters, and whose distance difference from the first faulty storage unit to the target first faulty storage unit is less than the distance threshold set in the clustering parameters, then the first faulty storage unit is included as an element in the set of faulty storage units corresponding to the target first faulty storage unit. Using the target first faulty storage unit as the cluster center, first faulty storage units that meet the conditions are found according to the above method, and these first faulty storage units are combined into a set of faulty storage units.
[0105] It should be noted that a target first faulty storage unit can identify a set of faulty storage units. Each first faulty storage unit in the target physical region has a set of faulty storage units. Therefore, a target physical region can identify multiple sets of faulty storage units. Each set of faulty storage units can be understood as a cluster, and each cluster has a cluster center. For example, if a cluster consists of faulty storage units in the third row with columns [2,3,8,10,11,12,14,19], then the cluster center is the center of columns 10 and 11. As another example, if a cluster consists of faulty storage units in the fifth column with rows [3,5,7,10,11], then the cluster center is the faulty storage unit in the fifth column and seventh row. The cluster center can be virtual or real.
[0106] Furthermore, the fault clustering results are determined based on multiple sets of fault storage units. The number of fault clustering results is less than or equal to the number of sets of fault storage units. Specifically, one or more fault clustering results that determine the target physical region based on multiple sets of fault storage units may include:
[0107] When two adjacent clusters have the same one or more faulty storage units, these two clusters are merged to obtain a single cluster. In other words, the two sets of faulty storage units are merged to obtain a single set of faulty storage units.
[0108] The number of faulty storage units included in the merged set of faulty storage units and the set of faulty storage units that have not been merged is determined. If the number reaches the clustering density set in the clustering parameters, the corresponding set of faulty storage units is determined as the faulty clustering result. For example, two adjacent clusters are composed of faulty storage units with columns [2, 3, 8, 10] in the third row and [8, 10, 12, 14] in the third row, respectively. Since the two clusters have the same storage units: columns 8 and 10 in the third row, the two clusters can be merged to obtain a single cluster composed of faulty storage units with columns [2, 3, 8, 10, 12, 14] in the third row.
[0109] In one possible implementation, the k-means clustering algorithm can be used to cluster the faulty storage units in each target physical region. The k-means clustering algorithm clusters according to a set number of clusters, cluster centers, and radius. The number of clusters represents the number of clusters, and then clustering is performed based on the cluster centers of each cluster.
[0110] S404: Determine the fault isolation region based on the fault clustering results. The fault isolation region is the area formed by the addresses of multiple second fault storage units.
[0111] In one possible implementation, when determining the fault isolation region based on the fault clustering results, it can be determined based on the memory fault type or the type of the target physical region. For example, if the memory fault type of each faulty storage unit in the fault clustering results is a row fault type, or the target physical region is a target row physical region, then in the target physical region, the fault isolation region corresponding to each fault clustering result is determined based on the minimum and maximum values of the column addresses of each faulty storage unit in each fault clustering result. If the memory fault type of each faulty storage unit in the fault clustering results is a column fault type, or the target physical region is a target column physical region, then in the target physical region, the fault isolation region corresponding to each fault clustering result is determined based on the minimum and maximum values of the row addresses of each faulty storage unit in each fault clustering result. Each clustering result includes multiple second faulty storage units, so the corresponding fault isolation region is also a region formed by the addresses of multiple second faulty storage units. Specifically, this region contains not only faulty storage units but also non-faulty storage units, because the fault isolation region is a region, not a single storage unit.
[0112] In other words, when the clustered faulty storage units are of the row fault type, the fault isolation region is determined based on the fault clustering results and the corresponding column addresses; when the clustered faulty storage units are of the column fault type, the fault isolation region is determined based on the fault clustering results and the corresponding row addresses. It's important to note that when the fault type is row fault, in addition to the minimum and maximum values of the column addresses of each faulty storage unit, it also depends on which specific row the fault occurs in. For example, if the fault type is row fault, and the fault is in the fifth row of the target storage array, and the minimum and maximum values of the column addresses of each faulty storage unit in one set of faulty storage units are 8 and 18 respectively, then the corresponding fault isolation region is the area from (5,8) to (5,18), which is the area from column 8 to column 18 of the fifth row.
[0113] It should be noted that when there is only one fault clustering result, a corresponding fault isolation region can be determined based on the fault clustering result, and the target physical region can be determined to have one fault isolation region; when there are multiple fault clustering results, the fault isolation region corresponding to each fault clustering result can be determined based on each fault clustering result, and the target physical region can be determined to have multiple fault isolation regions.
[0114] like Figure 5 The image shown is a schematic diagram of a fault clustering result. Figure 5The target physical region shown is a column. After clustering processing using the memory fault handling method provided in this application embodiment, three fault clustering results are obtained (the three boxes in the figure). Based on the row position information of each fault clustering result, three fault isolation regions can be determined. For example, from top to bottom, there are three fault isolation regions: the first fault isolation region is the second row (row minimum) to the fifth row (row maximum) of column 6; the second fault isolation region is the 18th to the 22nd row of column 6; and the third fault isolation region is the 30th to the 34th row of column 6.
[0115] In one possible implementation, the need to expand the isolation region can be further determined based on the number of storage units included in the fault isolation region, thereby improving memory health. Specifically, it can be defined as follows: if the ratio between the number of storage units included in the fault isolation region and the total number of storage units included in the target physical region is greater than a ratio threshold, then the target physical region is designated as the fault isolation region, where the target physical region is an entire row or column in the storage array. This ratio threshold is a percentage and can be set to 0.5 or other values; this value can be customized, and this embodiment does not limit its application.
[0116] For example, if the target physical region is column 13, which contains 100 storage units and 40 faulty storage units, then the ratio of the number of faulty storage units to the total number of storage units in this column is less than 0.5. Therefore, the faulty storage units in this column can be clustered to obtain two fault isolation regions. However, if the ratio of the total number of storage units (including faulty and fault-free storage units) in these two fault isolation regions to the total number of storage units in the column is greater than 0.5, then this column can also be directly used as a fault isolation region.
[0117] In practical applications, when performing isolation, the isolated area is the page corresponding to the fault isolation area. Therefore, after determining the fault isolation area, it is also necessary to determine the page corresponding to the fault isolation area. The process of determining the page corresponding to the fault isolation area can be performed by the processing device, or by the computer device or server.
[0118] In one possible implementation, after determining the fault isolation area, the processing device can send a fault isolation request to the computer device or server to request the isolation of the fault isolation area. When the central processing unit of the computer device or server receives the fault isolation request, it can determine the page corresponding to the fault isolation area and perform the isolation operation, that is, isolate the page corresponding to the fault isolation area.
[0119] In one possible implementation, after the processing device determines the page corresponding to the fault isolation area, the processing device can send a fault isolation request to the computer device or server, requesting the isolation of the page corresponding to the fault isolation area. When the central processing unit of the computer device or server receives the fault isolation request, it can perform an isolation operation, that is, isolate the page corresponding to the fault isolation area.
[0120] In this embodiment, by analyzing memory fault information and clustering faulty storage units within the corresponding target physical area, a fault isolation region is determined. Clustering before determining the fault isolation region allows for a more accurate identification of the faulty area. Based on the fault clustering results, a local isolation region is determined. Compared to overall isolation, local isolation reduces the isolation of available areas, thereby improving memory utilization.
[0121] The memory fault handling method provided in this application can be applied to the baseboard management controller and the operation and maintenance platform, or it can be directly applied to the central processing unit. The following example illustrates its application to the baseboard management controller and the operation and maintenance platform:
[0122] When the memory fault handling method is applied to the baseboard management controller, the flowchart of the corresponding memory fault handling method can be as follows: Figure 6 As shown, the corresponding process at this time is as follows:
[0123] ① The memory reports memory fault information to the Basic Input / Output System (BIOS); When a memory failure occurs during computer operation, the memory can report the fault information to the BIOS. The reported fault information can be real-time, reported at set time intervals, or triggered by a certain number of events, etc.
[0124] ② The Baseboard Management Controller (BMC) obtains memory fault information from the BIOS (also known as the Central Processing Unit). This process can be that the BIOS sends the memory fault information to the Baseboard Management Controller (BMC); the BIOS can actively send it to the BMC; or the BIOS can send the information to the Baseboard Management Controller after receiving the information from the Baseboard Management Controller.
[0125] ③ The baseboard management controller analyzes memory fault information to determine the target physical region corresponding to the storage array that needs clustering. Based on Figure 3 In the described embodiments, the baseboard management controller specifically performs analysis of the location and time information of memory fault information, determines the memory fault type of the faulty storage unit based on this information, and determines the target physical region in the target storage array that needs to be clustered based on the memory fault type, etc.
[0126] ④ The baseboard management controller performs clustering processing on the fault storage units included in the target physical area based on the set clustering algorithm and clustering parameters to determine the fault isolation area.
[0127] Then, the baseboard management controller sends the fault isolation region to the central processing unit (CPU). The fault isolation region is used to trigger the CPU to isolate the fault isolation region. The CPU runs processing firmware (such as BIOS) and OS. This process specifically includes the following steps ⑤ and ⑥:
[0128] ⑤ The baseboard management controller reports the fault isolation area to the BIOS;
[0129] ⑥ The BIOS notifies the operating system (OS) to isolate the fault isolation area based on the reported fault isolation area.
[0130] When memory fault handling methods are applied to an operations and maintenance (O&M) platform, which possesses data processing capabilities and can be deployed on a cloud server, the flowchart of the corresponding memory fault handling method can be shown as follows: Figure 7a As shown, the corresponding process at this time is as follows:
[0131] ① The computer equipment or server reports memory fault information to the operation and maintenance platform; similarly, inside the server, the memory first reports the memory fault information to the BIOS, and then the BIOS reports it to the operation and maintenance platform.
[0132] ② The operation and maintenance platform should receive memory fault information and analyze it to determine the target physical region corresponding to the storage array that needs to be clustered.
[0133] ③ The operation and maintenance platform performs clustering processing on the faulty storage units included in the target physical area based on the set clustering algorithm and clustering parameters to determine the fault isolation area.
[0134] ④ The operation and maintenance platform returns the fault isolation area to the computer equipment or server.
[0135] ⑤ The operating system of the computer equipment or server isolates the fault isolation area based on the reported fault isolation area.
[0136] When memory fault handling methods are applied to an operations and maintenance platform that lacks data processing capabilities, the memory fault information needs to be sent to a separate processing device for handling. A flowchart of the corresponding memory fault handling method can be shown below. Figure 7b As shown, the corresponding process at this time is as follows:
[0137] ① Computer equipment or servers report memory fault information to the operation and maintenance platform.
[0138] ② The operation and maintenance platform should receive memory fault information and send it to the processing device.
[0139] ③ The processing device analyzes the memory fault information of the memory to determine the target physical region corresponding to the storage array that needs to be clustered.
[0140] ④ The processing device performs clustering processing on the faulty storage units included in the target physical area based on the set clustering algorithm and clustering parameters to determine the fault isolation area.
[0141] ⑤ The processing equipment returns the fault isolation area to the operation and maintenance platform.
[0142] ⑥ The operation and maintenance platform notifies the central processing unit to isolate the fault isolation area. Specifically, it notifies the operating system of the computer equipment or server to isolate the fault isolation area according to the reported fault isolation area.
[0143] It should be noted that the isolation of the fault isolation area described above can be the isolation of the page corresponding to the fault isolation area, or it can be the isolation of the fault isolation area itself.
[0144] Applying this method to an operations and maintenance platform is beneficial for the operation and maintenance management of clusters and data centers. In summary, the memory fault handling method provided in this application mainly involves local isolation of memory faults, thereby achieving a balance between capacity and memory health, and avoiding the waste of space by isolating entire rows or columns of memory.
[0145] The methods of the embodiments of this application have been described in detail above. In order to facilitate better implementation of the above solutions of the embodiments of this application, the apparatus of the embodiments of this application is provided below.
[0146] Please see Figure 8 , Figure 8 This is a schematic diagram of a memory fault handling device provided in an embodiment of this application. The memory fault handling device can be installed in the processing equipment or operation and maintenance platform provided in the embodiment of this application. Figure 8 The memory fault handling device shown can be a computer program (including program code) running on a processing device or operation and maintenance platform. This memory fault handling device can be used to execute... Figure 4 , Figure 6 , Figure 7a as well as Figure 7b Some or all of the steps in the method embodiments shown. Please refer to [link / reference]. Figure 8 The memory fault handling device may include the following units:
[0147] The determining unit 801 is used to determine the fault information of each faulty storage unit among multiple faulty storage units based on the memory fault information of the memory; if the number of faulty storage units included in the target storage array reaches a preset threshold, then the target physical region of the target storage array is determined, wherein the target storage array is any one of one or more storage arrays determined based on the location information of the multiple faulty storage units, and the target physical region includes multiple first faulty storage units, wherein the multiple first faulty storage units are at least some of the multiple faulty storage units, and the multiple first faulty storage units are located in the same row or column;
[0148] The processing unit 802 is configured to perform clustering processing on each of the first fault storage units in the target physical area based on the fault information of each first fault storage unit, and obtain a fault clustering result. The fault clustering result includes a plurality of second fault storage units, which are at least some of the first fault storage units among the plurality of first fault storage units, and the plurality of second fault storage units satisfy preset conditions.
[0149] The determining unit 801 is further configured to determine a fault isolation region based on the fault clustering result, wherein the fault isolation region is the region formed by the addresses of the plurality of second fault storage units.
[0150] In one possible implementation, the fault information of each faulty storage unit includes location information. When determining the target physical region of the target storage array, the determining unit 801 specifically performs the following steps:
[0151] The memory fault type of each first faulty storage unit is determined based on the location information of the first faulty storage unit included in the target storage array. The memory fault type includes row fault type and column fault type.
[0152] The target physical region is determined based on the memory fault type of each first fault storage unit. The target physical region includes: the target row physical region corresponding to the row fault type, or the target column physical region corresponding to the column fault type. The multiple first fault storage units included in the target row physical region are located in the same row, and the multiple first fault storage units included in the target column physical region are located in the same column.
[0153] In one possible implementation, the fault information of each faulty storage unit further includes time information. When the processing unit 802 performs clustering processing on each of the first faulty storage units in the target physical area based on the fault information of each first faulty storage unit to obtain the fault clustering result, it specifically performs the following steps:
[0154] Based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit, the first fault storage units in the target physical area are clustered to obtain the fault clustering result of the target physical area.
[0155] In one possible implementation, when processing unit 802 performs clustering processing on the first fault storage units within the target physical region based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit, and obtains the fault clustering result of the target physical region, it specifically performs the following steps:
[0156] For any target first fault storage unit in the target physical region, the time difference between the target first fault storage unit and other first fault storage units in the target physical region is determined based on the time information of the target first fault storage unit; the target first fault storage unit is any one of the first fault storage units in the target physical region.
[0157] The distance difference between the target first faulty storage unit and other first faulty storage units in the target area physical region is determined based on the location information of the target first faulty storage unit;
[0158] Based on the clustering parameters, the time difference, and the distance difference, determine the set of fault storage units corresponding to each first fault storage unit included in the target physical region;
[0159] The fault clustering result of the target physical region is determined based on the set of fault storage units corresponding to each first fault storage unit. The fault clustering result includes multiple second fault storage units, and the multiple second fault storage units satisfy preset conditions.
[0160] In one possible implementation, the plurality of second fault storage units satisfying preset conditions include: the number of the plurality of second fault storage units satisfies the clustering density set in the clustering parameters; the time difference between any two second fault storage units is less than the time threshold set in the clustering parameters; and the distance difference between the cluster center of the fault clustering result and any one of the second fault storage units is less than the distance threshold set in the clustering parameters.
[0161] In one possible implementation, when determining the fault isolation zone based on the fault clustering results, the determining unit 801 specifically performs the following steps:
[0162] If the fault type of the second fault storage unit included in the fault clustering result is the row fault type, then the fault isolation region is determined according to the minimum and maximum values in the column addresses of each second fault storage unit.
[0163] If the fault type of the second fault storage unit included in the fault clustering result is the column fault type, then the fault isolation region is determined according to the minimum and maximum values in the row address of each second fault storage unit.
[0164] In one possible implementation, the processing unit 802 is further configured to perform the following steps:
[0165] If the ratio between the number of storage units included in the fault isolation region and the total number of first storage units included in the target physical region is greater than a ratio threshold, then the target physical region is designated as the fault isolation region, where the target physical region is an entire row or column of the target storage array.
[0166] In one possible implementation, the memory fault information is obtained from the central processing unit; the processing unit 802 is further configured to perform the following steps:
[0167] The fault isolation region is sent to the central processing unit (CPU), and the fault isolation region is used to trigger the CPU to isolate the fault isolation region.
[0168] In one possible implementation, the memory fault information is obtained from the operation and maintenance platform; the processing unit 802 is further configured to perform the following steps:
[0169] The fault isolation area is sent to the operation and maintenance platform so that the operation and maintenance platform notifies the central processing unit to isolate the fault isolation area.
[0170] According to another embodiment of this application, Figure 8 The memory fault handling device shown can be constructed by combining the various units into one or more other units, or one or more of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. The above units are based on logical function division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the memory fault handling device may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.
[0171] In this embodiment, the location information of the faulty storage unit is first determined, and then the faulty storage units in the corresponding target physical area are clustered according to the location information. The fault isolation area is determined according to the obtained fault clustering results. This method can achieve local isolation of faulty storage units, reduce the isolation of some available storage units, avoid the waste of storage units, and thus improve memory utilization.
[0172] Based on the above methods and apparatus embodiments, this application provides a processing device. Please refer to... Figure 9 , Figure 9 This is a schematic diagram of the structure of a processing device provided in an embodiment of this application. The processing device can be... Figure 1 The baseboard management controller 102 shown can also be an operation and maintenance platform. Figure 9 The processing device shown includes at least a processor 901, an input interface 902, an output interface 903, and a computer-readable storage medium 904. The processor 901, input interface 902, output interface 903, and computer-readable storage medium 904 can be connected via a bus or other means.
[0173] Computer-readable storage medium 904 can be stored in the memory of a computer device. Computer-readable storage medium 904 is used to store computer programs, including computer instructions. Processor 901 is used to execute the program instructions stored in computer-readable storage medium 904. Processor 901 (or CPU (Central Processing Unit)) is the computing and control core of the processing device, suitable for implementing one or more computer instructions, specifically suitable for loading and executing one or more computer instructions to achieve corresponding method flows or corresponding functions.
[0174] This application also provides a computer-readable storage medium (Memory), which is a memory device in a processing device used to store programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the processing device and extended storage media supported by the processing device. The computer-readable storage medium provides storage space that stores the operating system of the processing device. Furthermore, the storage space also stores one or more computer instructions suitable for loading and execution by a processor. These computer instructions may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here may be high-speed RAM or non-volatile memory, such as at least one disk storage device; optionally, it may also be at least one computer-readable storage medium located remotely from the aforementioned processor.
[0175] In some embodiments, the processor 901 may load and execute one or more computer instructions stored in the computer-readable storage medium 904 to implement the corresponding steps of the memory fault handling method described above. Specifically, the computer instructions in the computer-readable storage medium 904 are loaded and executed by the processor 901 in the following steps:
[0176] The fault information of each faulty storage unit in multiple faulty storage units is determined based on the memory fault information of the memory.
[0177] If the number of faulty storage units included in the target storage array reaches a preset threshold, then the target physical region of the target storage array is determined. The target storage array is any one of one or more storage arrays determined based on the location information of the multiple faulty storage units. The target physical region includes multiple first faulty storage units. The multiple first faulty storage units are at least some of the multiple faulty storage units. The multiple first faulty storage units are located in the same row or column.
[0178] Based on the fault information of each first fault storage unit, clustering processing is performed on each first fault storage unit in the target physical area to obtain a fault clustering result. The fault clustering result includes multiple second fault storage units, which are at least some of the multiple first fault storage units, and the multiple second fault storage units satisfy preset conditions.
[0179] Based on the fault clustering results, a fault isolation region is determined, which is the region formed by the addresses of the plurality of second fault storage units. In one possible implementation, the fault information of each fault storage unit further includes time information. When the computer instructions in the computer-readable storage medium 904 are loaded and executed by the processor 901 to perform clustering processing on each of the first fault storage units in the target physical region based on the fault information of each first fault storage unit, and to obtain the fault clustering results, the following steps are specifically performed:
[0180] Based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit, the first fault storage units in the target physical area are clustered to obtain the fault clustering result of the target physical area.
[0181] In one possible implementation, when the computer instructions in the computer-readable storage medium 904 are loaded and executed by the processor 901 to perform clustering processing on the first faulty storage units within the target physical region based on clustering parameters, location information of each first faulty storage unit, and time information of each first faulty storage unit, and to obtain the fault clustering result of the target physical region, the instructions specifically perform the following steps:
[0182] For any target first fault storage unit in the target physical region, the time difference between the target first fault storage unit and other first fault storage units in the target physical region is determined based on the time information of the target first fault storage unit; the target first fault storage unit is any one of the first fault storage units in the target physical region.
[0183] The distance difference between the target first faulty storage unit and other first faulty storage units in the target area physical region is determined based on the location information of the target first faulty storage unit;
[0184] Based on the clustering parameters, the time difference, and the distance difference, determine the set of fault storage units corresponding to each first fault storage unit included in the target physical region;
[0185] The fault clustering result of the target physical region is determined based on the set of fault storage units corresponding to each first fault storage unit. The fault clustering result includes multiple second fault storage units, and the multiple second fault storage units satisfy preset conditions.
[0186] In one possible implementation, the plurality of second fault storage units satisfying preset conditions include: the number of the plurality of second fault storage units satisfies the clustering density set in the clustering parameters; the time difference between any two second fault storage units is less than the time threshold set in the clustering parameters; and the distance difference between the cluster center of the fault clustering result and any one of the second fault storage units is less than the distance threshold set in the clustering parameters.
[0187] In one possible implementation, the computer instructions in the computer-readable storage medium 904, when loaded and executed by the processor 901 to determine the fault isolation area based on the fault clustering results, are specifically used to perform the following steps:
[0188] If the fault type of the second fault storage unit included in the fault clustering result is the row fault type, then the fault isolation region is determined according to the minimum and maximum values in the column addresses of each second fault storage unit.
[0189] If the fault type of the second fault storage unit included in the fault clustering result is the column fault type, then the fault isolation region is determined according to the minimum and maximum values in the row address of each second fault storage unit.
[0190] In one possible implementation, the computer instructions in the computer-readable storage medium 904 are further loaded by the processor 901 and executed as follows:
[0191] If the ratio between the number of storage units included in the fault isolation region and the total number of first storage units included in the target physical region is greater than a ratio threshold, then the target physical region is designated as the fault isolation region, where the target physical region is an entire row or column of the target storage array.
[0192] In one possible implementation, the memory fault information is obtained from the central processing unit; the computer instructions in the computer-readable storage medium 904 are further loaded by the processor 901 and executed as follows:
[0193] The fault isolation region is sent to the central processing unit (CPU), and the fault isolation region is used to trigger the CPU to isolate the fault isolation region.
[0194] In one possible implementation, the memory fault information is obtained from the operation and maintenance platform; the computer instructions in the computer-readable storage medium 904 are further loaded by the processor 901 and executed as follows:
[0195] The fault isolation area is sent to the operation and maintenance platform so that the operation and maintenance platform notifies the central processing unit to isolate the fault isolation area.
[0196] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.
[0197] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, and the division of modules described above is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed.
[0198] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A memory failure handling method, characterized by, The method includes: The fault information of each faulty storage unit in multiple faulty storage units is determined based on the memory fault information of the memory. If the number of faulty storage units included in the target storage array reaches a preset threshold, then the target physical region of the target storage array is determined. The target storage array is any one of one or more storage arrays determined based on the location information of the multiple faulty storage units. The target physical region includes multiple first faulty storage units. The multiple first faulty storage units are at least some of the multiple faulty storage units. The multiple first faulty storage units are located in the same row or column. Based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit, clustering processing is performed on the first fault storage units within the target physical area to obtain a fault clustering result for the target physical area. The fault clustering result includes multiple second fault storage units, which are at least some of the multiple first fault storage units. The multiple second fault storage units satisfy preset conditions. The multiple second fault storage units satisfying preset conditions include: the number of the multiple second fault storage units meets the clustering density set in the clustering parameters; the time difference between any two second fault storage units is less than the time threshold set in the clustering parameters; and the distance difference between the cluster center of the fault clustering result and any one of the multiple second fault storage units is less than the distance threshold set in the clustering parameters. The fault isolation region is determined based on the fault clustering results. The fault isolation region is the region formed by the addresses of the plurality of second fault storage units.
2. The method of claim 1, wherein, The fault information of each faulty storage unit includes location information, and determining the target physical region of the target storage array includes: The memory fault type of each first faulty storage unit is determined based on the location information of the first faulty storage unit included in the target storage array. The memory fault type includes row fault type and column fault type. The target physical region is determined based on the memory fault type of each first fault storage unit. The target physical region includes: the target row physical region corresponding to the row fault type, or the target column physical region corresponding to the column fault type. The multiple first fault storage units included in the target row physical region are located in the same row, and the multiple first fault storage units included in the target column physical region are located in the same column.
3. The method of claim 1, wherein, The first fault storage units within the target physical region are clustered based on clustering parameters, the location information of each first fault storage unit, and the time information of each first fault storage unit to obtain the fault clustering result of the target physical region, including: For any target first fault storage unit in the target physical region, the time difference between the target first fault storage unit and other first fault storage units in the target physical region is determined based on the time information of the target first fault storage unit; the target first fault storage unit is any one of the first fault storage units in the target physical region. The distance difference between the target first faulty storage unit and other first faulty storage units in the target physical area is determined based on the location information of the target first faulty storage unit; Based on the clustering parameters, the time difference, and the distance difference, determine the set of fault storage units corresponding to each first fault storage unit included in the target physical region; The fault clustering result of the target physical region is determined based on the set of fault storage units corresponding to each first fault storage unit. The fault clustering result includes multiple second fault storage units, and the multiple second fault storage units satisfy preset conditions.
4. The method of claim 2, wherein, The step of determining the fault isolation zone based on the fault clustering results includes: If the fault type of the second fault storage unit included in the fault clustering result is the row fault type, then the fault isolation region is determined according to the minimum and maximum values in the column addresses of each second fault storage unit. If the fault type of the second fault storage unit included in the fault clustering result is the column fault type, then the fault isolation region is determined according to the minimum and maximum values in the row address of each second fault storage unit.
5. The method of claim 4, wherein, The method further includes: If the ratio between the number of storage units included in the fault isolation region and the total number of first storage units included in the target physical region is greater than a ratio threshold, then the target physical region is designated as the fault isolation region, where the target physical region is an entire row or column of the target storage array.
6. The method of claim 1, wherein, The memory fault information is obtained from the central processing unit; the method further includes: The fault isolation region is sent to the central processing unit (CPU), and the fault isolation region is used to trigger the CPU to isolate the fault isolation region.
7. The method of any one of claims 1, wherein, The memory fault information is obtained from the operation and maintenance platform; the method further includes: The fault isolation area is sent to the operation and maintenance platform so that the operation and maintenance platform notifies the central processing unit to isolate the fault isolation area.
8. A processing device, characterized by The processing equipment includes: A processor, suitable for implementing one or more computer programs; and, A computer storage medium storing one or more computer programs, said one or more computer programs being adapted to be loaded by the processor and executed as described in any one of claims 1-7.
9. An operation and maintenance platform, characterized in that, The operation and maintenance platform includes: A processor, suitable for implementing one or more computer programs; and, A computer storage medium storing one or more computer programs, said one or more computer programs being adapted to be loaded by the processor and executed as described in any one of claims 1-7.
Citation Information
Patent Citations
Memory fault analysis method and computer equipment
CN115686901A
Memory fault processing method and computer equipment
CN116483600A
