WO2024120118A1

WO2024120118A1 - Interrupt request processing method, system, device and computer-readable storage medium

Info

Publication number: WO2024120118A1
Application number: PCT/CN2023/130987
Authority: WO
Inventors: 张景涛; 童贤慧; 余润; 李健; 宦旭
Original assignee: 中兴通讯股份有限公司
Priority date: 2022-12-05
Filing date: 2023-11-10
Publication date: 2024-06-13
Also published as: CN118152063A

Abstract

The present application discloses an interrupt request processing method, a system, a device, and a computer-readable storage medium. The interrupt request processing method comprises: in response to an interrupt request, detecting whether an online virtual central processing unit (vCPU) is included in a preset online list; if a detection result is yes, acquiring scheduling information and load information corresponding to at least one online vCPU; according to the scheduling information and the load information, selecting an online vCPU having the smallest load from the at least one online vCPU as a target vCPU; sending the interrupt request to the target vCPU for interrupt processing.

Description

Interrupt request processing method, system, device and computer readable storage medium

Related Applications

This application claims priority to Chinese patent application No. 202211557937.X filed on December 5, 2022, the entire contents of which are incorporated by reference into this application.

Technical Field

The present application relates to the field of computers, and to an interrupt request processing method, system, device, and computer-readable storage medium.

Background technique

With the development of network technology, input/output virtualization has become a decisive factor in system performance and efficiency in virtualized infrastructure. It integrates multiple virtual machine instances through a hypervisor or virtual machine manager to share the same underlying physical server resources. Virtual I/O (Input/Output) devices simulate real physical I/O devices to process I/O requests and generate corresponding interrupts. The virtual machine manager intercepts and simulates the host system to schedule vCPU (virtual Central Processing Unit) to process interrupts. The virtual machine exit caused by interrupt delivery and the related interrupt processing delay will greatly affect the I/O throughput and response of the virtual machine. Therefore, how to reduce the interrupt request processing time delay of vCPU is an urgent problem to be solved.

Summary of the invention

The main purpose of this application is to provide an interrupt request processing method, system, device and computer-readable storage medium, aiming to solve the technical problem of how to reduce the interrupt request processing time delay of the vCPU.

To achieve the above object, the present application provides an interrupt request processing method, comprising:

In response to the interrupt request, detecting whether a preset online list includes an online virtual central processing unit vCPU;

If the detection result is yes, obtaining scheduling information and load information corresponding to at least one of the online vCPUs;

Selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

The interrupt request is sent to the target vCPU for interrupt processing.

In addition, to achieve the above purpose, the present application also provides an interrupt request processing system, including:

A receiving module, configured to detect, in response to an interrupt request, whether a preset online list includes an online virtual central processing unit vCPU;

An information acquisition module, used for acquiring scheduling information and load information corresponding to at least one of the online vCPUs when the detection result is yes;

A redirection module, configured to select a vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

A processing module is used to send the interrupt request to the target vCPU for interrupt processing.

In addition, to achieve the above-mentioned purpose, the present application also provides an interrupt request processing device, which includes a memory, a processor, and an interrupt request processing program stored in the memory and executable on the processor, and when the interrupt request processing program is executed by the processor, the steps of the interrupt request processing method described above are implemented.

In addition, to achieve the above-mentioned purpose, the present application also provides a computer-readable storage medium, on which an interrupt request processing program is stored, and when the interrupt request processing program is executed by a processor, the steps of the interrupt request processing method as described above are implemented.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a first embodiment of a method for processing an interrupt request of the present invention;

FIG2 is an illustrative schematic diagram of an online list and an offline list of the interrupt request processing method of the present application;

FIG3 is a system distribution framework diagram of the interrupt request processing method of the present application;

FIG4 is another flow chart of the interrupt request processing method of the present invention;

FIG5 is a schematic diagram of an interrupt request processing system of the present application;

FIG6 is a schematic diagram of the terminal\device structure of the hardware operating environment involved in the embodiment of the present application.

The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings.

Detailed ways

It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

When a virtual machine receives an interrupt request, the target vCPU (virtual Central Processing Unit) that handles the interrupt can be set according to the interrupt affinity of the vCPU in the virtual machine, denoted as vCPUA. In the case where the physical CPU core is reused (that is, multiple vCPUs may share a core), when the PI (Posted Interrupt) delivers the interrupt to it, the target vCPU A may be in the scheduling waiting queue. Then the delivered interrupt cannot be processed before vCPU A is rescheduled to the physical CPU. In this case, the scheduling delay of vCPU A will inevitably be introduced into the client's I/O (Input/Output) event processing, causing additional I/O delays.

In addition, the side-core method can be used to assign dedicated cores to be responsible for I/O processing of virtual machines. The side-core method divides the cores into two different sets: one for running the guest system and the other dedicated to virtual I/O processing. However, the number of cores that should be assigned to each group depends on the changing workload. In the worst case, the number of dedicated cores used for virtual I/O processing is much larger (smaller) than the optimal value, and one of the two sets of cores is always overloaded, consuming more CPU cycles, and the performance of other applications (workloads) running on the virtual machine is impaired.

Relevant scheduling algorithms can also be used to reduce the waiting time of each vCPU in the run queue. The scheduling algorithm-related methods can reduce the time slice of the scheduler in the virtual machine manager, timely schedule the vCPU that receives the interrupt, and reduce the waiting time of the vCPU in the scheduling queue. However, frequent scheduling switches between vCPUs in order to quickly respond to interrupts will lead to an astonishing number of context switches, and a large number of CPU cycles will be wasted, causing serious performance degradation in practice. At the same time, the virtual machine manager scheduler treats each vCPU fairly by allocating an equal number of CPU cycles to each vCPU. However, the interrupt load on each vCPU is uneven, and more CPU cycles cannot be allocated to vCPUs with greater interrupt load pressure, which cannot fully utilize system resources.

Therefore, in order to avoid the generation of additional unnecessary context switching overhead when processing interrupt requests, and the inability to achieve interrupt load balancing on the vCPU, resulting in an uneven number of interrupt requests on each vCPU, the inability to dynamically allocate corresponding CPU cycles according to the interrupt load situation on the vCPU, and the inability to fully utilize resources, in this embodiment, when a virtual device or a physical device directly connected to the virtual machine completes the required I/O operation and generates an interrupt request, the virtual machine will respond to the received interrupt request and detect whether there is a vCPU in the preset online list. After determining that there is at least one vCPU in the preset online list, the interrupt request load information and the scheduling information of the vCPU of all current vCPUs in the preset online list are obtained and evaluated, and the interrupt request issued by the device is intelligently redirected to the target vCPU with the smallest load for processing; after detecting that the vCPU is not stored in the preset online list, a vCPU is selected from the preset offline list as the target vCPU to complete the processing of the interrupt request, thereby achieving load balancing of the interrupt request redirection to the target vCPU, reducing the time delay of interrupt request processing, and thereby improving the concurrency and responsiveness of I/O virtualization. It is based on notification interrupts and vCPU scheduling design in Linux KVM, and is applied to the interrupt transmission and delivery processing flow in the virtual I/O event path, reducing the I/O interrupt processing delay caused by the vCPU scheduler.

Furthermore, in an embodiment of the present application, the storage location of the load-related information of the vCPU in the virtual machine can also be sent directly to the virtual machine monitor on the host machine, so that the virtual machine monitor can share and determine the load scheduling strategy based on the vCPU load information, and select the most appropriate target vCPU for the interrupt request.

The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

1 , the present application provides an interrupt request processing method. In a first embodiment of the interrupt request processing method, the interrupt request processing method includes:

Step S10, in response to the interrupt request, detecting whether a preset online list includes an online virtual central processing unit vCPU;

Step S20: if the detection result is yes, obtaining scheduling information and load information corresponding to at least one of the online vCPUs;

Step S30, selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

Step S40: Send the interrupt request to the target vCPU for interrupt processing.

Exemplarily, the interrupt request processing method in the embodiment of the present application is applied to an interrupt request processing system, and a vCPU scheduling information module, a vCPU load information module and a target vCPU redirection module for the interrupt request can be set in the interrupt request processing system. After the virtual device or the direct physical device completes the required I/O operation and generates at least one interrupt request, before being sent to the corresponding specific vCPU through PI (Posted Interrupt, notification interrupt), the interrupt request will be intercepted by the target vCPU redirection module of the interrupt request. In one scenario, two information channels can be created in the interrupt request processing system, and the vCPU obtains the real-time vCPU scheduling status and vCPU load balancing information respectively through the two information channels, and intelligently finds the most suitable vCPU to process the interrupt request according to the vCPU scheduling status and vCPU load balancing information, so as to complete the optimization of the PI mechanism, thereby minimizing the IO event processing delay of the client.

By introducing the vCPU scheduling information module, an information channel is established with the vCPU scheduler of the VMM (virtual machine manager) to collect the current vCPU scheduling information, thereby avoiding the phenomenon of not knowing the scheduling status of the vCPU when allocating the target vCPU for the interrupt request. And by introducing the vCPU load information module, the load balancer in the VMM will collect the interrupt request reception status of the vAPIC Page on the vCPU, maintain the interrupt request status table of the current vCPU, and realize real-time evaluation of the interrupt request load, thereby avoiding the phenomenon of not being able to ensure the load balancing of the vCPU when allocating the target vCPU for the interrupt request. By introducing the target vCPU redirection module for the interrupt request, and after obtaining the vCPU scheduling status and vCPU load balancing status, the interrupt request is redirected to the online idle vCPU, which reduces the virtual machine I/O event processing delay converted by the vCPU scheduling delay, and avoids the impact of the virtual machine manager's scheduling activities on the virtual machine's I/O performance.

In step S10, in response to the interrupt request, detecting whether the preset online list includes an online virtual central processing unit vCPU;

Exemplarily, when a physical device or a virtual device sends an interrupt request, the virtual machine responds to the received interrupt request and detects whether an online vCPU is included in a preset online list. The preset online list is a notifier generated by the vCPU scheduling provided by the virtual machine to maintain an online status table of the vCPU for all virtual machines. If a vCPU thread is currently running on a physical CPU, it will be placed in the preset online list, and the currently online vCPUs will be counted in a first-in-first-out queue to record the order in which the vCPUs are stored.

Wherein, in step S10, after detecting whether the preset online list includes the online virtual central processing unit vCPU, the following steps are included:

Step z: after determining that the preset online list is an empty list, selecting an offline vCPU as a target vCPU from the preset offline list according to a preset storage order, wherein the empty list does not include an online vCPU.

Exemplarily, after detecting that the preset online list is an empty table, it is determined that no online vCPU is stored in the preset online list. At this time, an offline vCPU can be selected from the preset offline list according to the preset storage order as the target vCPU to process the received interrupt request. The preset offline list is a preset online list that uses the notifier generated by the vCPU scheduling provided by the virtual machine to maintain the offline status table of the vCPU for all virtual machines, and counts the currently offline vCPUs in a first-in-first-out (i.e., preset storage order) queue, and records the order in which the vCPUs are stored. The offline vCPU list is sorted to indicate the order in which all vCPUs are canceled. Whenever a vCPU is canceled, it will be removed from the online list and added to the end of the offline list. The offline vCPU may be a vCPU stored in the preset offline list that is in an offline state.

Exemplarily, after the virtual machine receives an interrupt request, if the preset online list is an empty list, it means that there is no online vCPU. At this time, the offline vCPU at the head of the queue can be selected as the target vCPU for processing the interrupt request in a first-in-first-out order. For example, as shown in FIG2, the preset offline list is a first-in-first-out queue structure and stores 4 vCPUs, which are vCPU5, vCPU6, vCPU7, and vCPU8 in the order of storage. When the preset online list is an empty list, the offline vCPU at the head of the queue, that is, vCPU5, is selected as the target vCPU, and the offline vCPU will enter the preset online list after being assigned the task of processing the interrupt request, and then be scheduled to run the processing interrupt request in the preset online list, thereby ensuring that after the interrupt request is assigned to the target vCPU, it will not fall into a long wait, resulting in the phenomenon that the interrupt request cannot be delivered in time.

For step S20, if the detection result is yes, obtaining scheduling information and load information corresponding to at least one of the online vCPUs;

Exemplarily, after determining that the preset online list includes at least one online vCPU, the vCPU scheduling information processing module will establish an information channel with the vCPU scheduler of the operating system to collect the current vCPU scheduling information. The scheduling information records the scheduling status of all vCPUs; the vCPU load information module will establish an information channel with the vCPU scheduler of the operating system to collect the current vCPU load information, and the load information records the number of interrupt loads included in all vCPUs.

Exemplarily, the virtual machine registers two preemption notifier functions for each vCPU, which are recorded as kvm_sched_in and kvm_sched_out. kvm_sched_in will be called before a vCPU thread is about to be scheduled. When kvm_sched_in is called, the corresponding vCPU is placed in the preset online list, and the calling state of the vCPU is recorded as whether it is a processing interrupt state or an unprocessed interrupt state. kvm_sched_out will be called immediately when a vCPU thread is canceled. When kvm_sched_out is called, the corresponding vCPU is placed in the preset offline list. By using these two notifiers, the vCPU scheduling information processing module can collect and update the vCPU scheduling state of each virtual machine.

Exemplarily, the virtual machine also designs a kvm_create_vcpu system call function. After the user creates a vCPU through the kvm_create_vcpu system call request, the virtual machine creates a corresponding kvm_vcpu structure for each vCPU and performs initialization operations. When the interrupt request is injected into the virtual advanced programmable interrupt controller page, the interrupt request load information recorded in the kvm_vcpu structure is dynamically updated and maintained, and the number of interrupt requests currently directed to the vCPU and the completion status are recorded in real time, and the current vCPU interrupt load pressure situation is evaluated.

Exemplarily, after detecting that there is at least one online vCPU in the preset online list, the scheduling information of all online vCPUs in the preset online list is obtained through the vCPU scheduling information module of the virtual machine monitor, wherein the scheduling information includes the scheduling status corresponding to each online vCPU, and the load information of all online vCPUs in the preset online list is obtained through the vCPU load information module, wherein the load information includes the number of interrupt loads corresponding to each online vCPU.

For step S30, selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

Exemplarily, all online vCPUs in the preset online list can perform interrupt request processing operations, that is, they can immediately receive interrupt requests delivered by the PI without any delay. Therefore, the online vCPU with the smallest load and the lightest load can be selected as the target vCPU from among the online vCPUs based on the obtained scheduling information and load information of the online vCPU. And when determining the target vCPU, the scheduling information of all online vCPUs in the preset online list can be compared, and the load information of all online vCPUs in the preset online list can be compared to determine the online vCPU with the smallest load, and use the online vCPU with the smallest load as the target vCPU.

For step S40, the interrupt request is sent to the target vCPU for interrupt processing.

Exemplarily, after determining the target vCPU, the interrupt request can be directly sent to the target vCPU for interrupt processing, and after the interrupt processing is completed, after receiving a new interrupt request sent by a physical device or a virtual device, the target vCPU can continue to process it.

In addition, to assist in understanding the interrupt request processing flow in this embodiment, an example is given below.

For example, as shown in FIG3 , a virtual machine, hardware, and a virtual machine monitor are included, and a vCPU scheduling information module, a vCPU load information module, and a target vCPU redirection module for interrupt requests are set in the virtual machine monitor. After the virtual machine monitor detects that a virtual device or a physical device generates an interrupt request, the target vCPU redirection module for the interrupt request will intercept the interrupt request. Two information channels will be created, and the vCPUs will be sent to the virtual machine monitor through the two information channels. Get real-time vCPU scheduling status and vCPU load balancing information, that is, the vCPU scheduling information module monitors the vCPU scheduler through an information channel to obtain vCPU scheduling information. The vCPU load information module monitors the notification interrupts of each vCPU through another information channel to obtain vCPU interrupt load information. The virtual machine monitor intelligently finds the most suitable vCPU to handle the interrupt request based on the vCPU scheduling status and vCPU load balancing information to complete the optimization PI mechanism, thereby minimizing the client's IO event processing delay.

For another example, as shown in Figure 4, after receiving an interrupt request, the scheduling information and interrupt load information (i.e., load information) of the vCPU are obtained, and a vCPU with the smallest load is selected as the target vCPU for processing the interrupt request based on the scheduling information and interrupt load information of the vCPU. Then, it is determined whether there are online vCPUs stored in the preset online list and whether the average interrupt load number of all online vCPUs exceeds the threshold. If the conditions are met, an interrupt is passed to the current target vCPU for interrupt processing. If the online vCPU cannot be found or the average interrupt load number of the online vCPU exceeds the threshold, the head vCPU of the preset offline queue is used as the target vCPU, and the interrupt is notified to the target vCPU. After the target vCPU for processing the interrupt request is determined, the interrupt load information of the vCPU is updated according to the interrupt request.

This embodiment detects whether an online virtual central processing unit vCPU is included in a preset online list in response to an interrupt request; if the detection result is yes, obtains scheduling information and load information corresponding to at least one of the online vCPUs; selects an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information; and sends the interrupt request to the target vCPU for interrupt processing, thereby redirecting the interrupt request to the target vCPU with the smallest load for processing according to the scheduling information and load information of all vCPUs in the online list, overcoming the technical defects of the prior art in which expensive context switching between the virtual machine manager and the virtual machine when processing interrupt requests, or the inability to dynamically select the vCPU with the smallest load to process the received interrupt request according to the scheduling and load conditions of the vCPU, thereby reducing the interrupt request processing time delay of the vCPU.

Based on the first embodiment of the present application, a second embodiment of the interrupt request processing method of the present application is proposed. In this embodiment, in step S30 of the above embodiment, selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information includes:

Step a, obtaining the interruption load quantity corresponding to the online vCPU based on the load information, and obtaining the scheduling state corresponding to the online vCPU based on the scheduling information;

Step b: after all the scheduling states are in the unprocessed interrupt state, select an online vCPU with the smallest load from at least one of the online vCPUs as a target vCPU according to the interrupt load quantity.

Exemplarily, since the scheduling information of the vCPU records the scheduling status of each vCPU, the scheduling status of the online vCPU includes the processing interrupt status and the unprocessed interrupt status. The processing interrupt status refers to the state in which a certain online vCPU is selected to process an interrupt, and all states except the processing interrupt status are not in the processing interrupt status. If it is determined that the scheduling status of each online vCPU is the unprocessed interrupt status according to the scheduling information of all online vCPUs in the preset online list, and then the number of interrupt loads on each online vCPU is determined according to the load information of all online vCPUs in the preset online list, an online vCPU with the least number of interrupt loads is screened out as the target vCPU for receiving the processed interrupt request. For example, referring to Figure 4, if there are 4 online vCPUs in the preset online list of the first-in-first-out queue structure, namely vCPU1, vCPU2, vCPU3 and vCPU4, and the scheduling status of the 4 online vCPUs is the unprocessed interrupt status, the number of interrupt loads corresponding to vCPU1, vCPU2, vCPU3 and vCPU4 is 10, 4, 15, 8, then the online vCPU (i.e., vCPU2) corresponding to the smallest number of interrupt loads (i.e., 4) is selected as the target vCPU.

In this embodiment, after receiving an interrupt request, the interrupt request is redirected to the vCPU with the smallest online load for processing by obtaining the online vCPU scheduling status and the number of vCPU interrupt loads, thereby reducing the virtual machine I/O event processing delay converted by the vCPU scheduling delay, reducing the processing time delay of the vCPU interrupt request, and thereby avoiding the impact of the long interrupt request processing time delay on the I/O performance of the virtual machine.

In one embodiment, selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the interrupt load quantity includes:

Step c, determining an average interruption load quantity corresponding to all the interruption load quantities;

Step d, after the average interruption load quantity is less than a first preset threshold, determining a minimum interruption load quantity among all the interruption load quantities;

Step e: taking the online vCPU corresponding to the minimum interrupt load number as the target vCPU.

Exemplarily, after determining that all online vCPUs are in an unprocessed interrupt state, the number of interrupt loads corresponding to all online vCPUs is obtained, and the average of all interrupt load numbers is taken to obtain the average interrupt load number. The average interrupt load number is then compared with the first preset threshold. After the average interrupt load number is less than the first preset threshold, the online vCPU with the smallest load is selected from all online vCPUs as the target vCPU, that is, the online vCPU corresponding to the smallest interrupt load number is selected as the target vCPU. After the average interrupt load number is greater than or equal to the first preset threshold, an offline vCPU that meets the conditions is selected in the preset offline list as the target vCPU. Among them, the first preset threshold can be determined based on the ratio of the vCPU scheduling delay and the average interrupt request processing time. For example, if there are four online vCPUs (vCPU1, vCPU2, vCPU3, vCPU4) that are not in the interrupt processing state, the corresponding interrupt load numbers are (10, 4, 15, 8), and the average interrupt load number of the four online vCPUs is calculated to be 9.25. At this time, if the first preset threshold is 8, the average interrupt load number is less than the first preset threshold, and the online vCPU corresponding to the minimum interrupt load number, that is, vCPU2, is selected as the target vCPU.

In this embodiment, after determining that all online vCPUs in the preset online list are in an unprocessed interrupt state, the online vCPU corresponding to the minimum interrupt load number can be selected as the target vCPU for processing the interrupt request based on the load information of the online vCPU, thereby achieving load balancing of the interrupt request distribution to the target vCPU, setting the average threshold of the interrupt load on the vCPU, and ensuring that the processing delay of the interrupt request is always the minimum value of the total processing time of the existing interrupt request queue on the vCPU and the vCPU scheduling delay overhead, thereby reducing the waiting time of the interrupt request.

In one embodiment, after determining the average interruption load number corresponding to all the interruption load numbers, the method includes:

Step f: after the average interruption load number is greater than or equal to the first preset threshold, select an offline vCPU from the preset offline list as the target vCPU according to the preset storage order.

Exemplarily, since both the preset online list and the preset offline list are first-in-first-out queue structures, if the average interrupt load of all online vCPUs is greater than or equal to the first preset threshold, the offline vCPU at the head of the preset offline list can be selected as the target vCPU according to the first-in-first-out storage order (i.e., the preset storage order). The head vCPU in the preset offline queue is the vCPU with the longest offline time, and will be the first vCPU to resume online status and the first vCPU to enter the running status, thereby avoiding the occurrence of abnormal situations such as CPU operation abnormality or abnormal exit of interrupt request processing due to excessive interrupt load on the online vCPU, and ensuring that all interrupt requests on all online CPUs can be effectively and accurately processed.

For example, if all four vCPUs are online and not in the interrupt processing state (vCPU1, vCPU2, vCPU3, vCPU4), the corresponding interrupt load quantity is (10, 4, 15, 8), and the average interrupt load quantity of the 4 online vCPUs is calculated to be 9.25. If the first preset threshold is 10, the average interrupt load quantity is greater than the first preset threshold. At this time, the offline vCPU that meets the conditions in the preset offline list is selected as the target vCPU.

In one embodiment, after obtaining the scheduling state corresponding to the online vCPU based on the scheduling information, the method includes:

Step g, after there is a processing interruption state in all the scheduling states, determining the maximum interruption load number and the minimum interruption load number among all the interruption load numbers;

Step h, determining the ratio between the maximum interruption load number and the minimum interruption load number;

Step i, after the ratio is greater than a second preset threshold, continue to execute the step of selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the number of interrupt loads.

Exemplarily, if one of the online vCPUs is selected to handle an interrupt, that is, it is in the interrupt processing state, and the ratio between the maximum number of interrupt loads and the minimum number of interrupt loads on all current online vCPUs is greater than a preset experience value (i.e., the second preset threshold), the online vCPU corresponding to the minimum number of interrupt loads will still be selected as the target vCPU. If the ratio is less than or equal to the preset experience value, it is considered that the online vCPU that is processing the interrupt is selected as the target vCPU at this time, where the preset experience value can be any value set in advance by the user. For example, if there are 4 online vCPUs (vCPU1, vCPU2, vCPU3, vCPU4), the corresponding interrupt load numbers are (10, 4, 15, 8), and vCPU4 is in the interrupt processing state, vCPU1, vCPU2 and vCPU3 are all in the unprocessed interrupt state, and the ratio of the maximum interrupt load number to the minimum interrupt load number is calculated to be 3.75 and the average interrupt load number is 9.25. At this time, if the preset test value set in advance is 4 and the first preset threshold is 8, the ratio is greater than the preset experience value and the average interrupt load number is less than the first preset threshold. At this time, the online vCPU corresponding to the minimum interrupt load number, that is, vCPU2, is selected as the target vCPU. If the preset test value set in advance is 4 and the first preset threshold is 10, the ratio is greater than the preset experience value and the average interrupt load is greater than the first preset threshold. At this time, the connected vCPU that meets the conditions in the preset connection list is selected as the target vCPU; if the preset test value set in advance is 3, the ratio is less than the preset experience value. At this time, the online vCPU in the interrupt processing state, that is, vCPU4, is selected as the target vCPU.

In this embodiment, after there is an interrupt processing state in all scheduling states, the ratio between the maximum interrupt load number and the minimum interrupt load number among all interrupt load numbers is determined, and after the ratio is greater than a second preset threshold, the online vCPU with the smallest load is selected as the target vCPU among each online vCPU, thereby ensuring the validity of the determined target vCPU.

In one embodiment, after determining the ratio of the maximum interruption load number to the minimum interruption load number, the method further comprises:

Step j: after the ratio is less than or equal to a second preset threshold, select an online vCPU in a processing interrupt state among all the online vCPUs as a target vCPU.

Exemplarily, if the ratio of the maximum number of interrupt loads to the minimum number of interrupt loads of all online vCPUs is less than or equal to a second preset threshold value, the online vCPU that is processing the interrupt status is selected as the target vCPU, and the vCPU load information is updated in real time, and subsequent interrupt requests are redirected to this online vCPU until it is canceled by the vCPU scheduler or the ratio is greater than a preset empirical value, triggering redirection to select a new online vCPU with the smallest interrupt load as the redirection target vCPU, thereby ensuring that when an interrupt request arrives, if there is an online vCPU that is processing an interrupt, the interrupt request will be sent to this online vCPU for processing. Since this online vCPU is processing an interrupt, delivering the interrupt request to this online vCPU at this time can ensure that the interrupt request can be processed immediately and reduce The processing time delay is reduced, and the interrupt request is redirected to the online vCPU that is processing the interrupt only when the above ratio is less than or equal to the preset experience value, thereby ensuring the load balancing of each online vCPU.

In addition, referring to FIG. 5 , an embodiment of the present application provides an interrupt request processing system, including:

A detection module A10, configured to detect, in response to an interrupt request, whether a preset online list includes an online virtual central processing unit vCPU;

The information acquisition module A20 is used to obtain the scheduling information and load information corresponding to at least one of the online vCPUs when the detection result is yes;

A redirection module A30, configured to select a vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

The processing module A40 is used to send the interrupt request to the target vCPU for interrupt processing.

Wherein, the redirection module A30 is further used for:

Acquire the interruption load quantity corresponding to the online vCPU based on the load information, and acquire the scheduling state corresponding to the online vCPU based on the scheduling information;

After all the scheduling states are in an unprocessed interrupt state, an online vCPU with the smallest load is selected as a target vCPU from at least one of the online vCPUs according to the interrupt load quantity.

Wherein, the redirection module A30 is further used for:

Determine an average interruption load quantity corresponding to all of the interruption load quantities;

After the average interruption load quantity is less than a first preset threshold, determining a minimum interruption load quantity among all the interruption load quantities;

The online vCPU corresponding to the minimum interrupt load number is used as the target vCPU.

Wherein, the redirection module A30 is further used for:

After the average interrupt load number is greater than or equal to the first preset threshold, an offline vCPU is selected from the preset offline list as the target vCPU according to the preset storage order.

Wherein, the redirection module A30 is further used for:

After a processing interruption state exists in all the scheduling states, determining a maximum interruption load number and a minimum interruption load number among all the interruption load numbers;

determining a ratio between the maximum interruption load number and the minimum interruption load number;

After the ratio is greater than a second preset threshold, the step of selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the interrupt load quantity is continued.

Wherein, the redirection module A30 is further used for:

After the ratio is less than or equal to the second preset threshold, an online vCPU in a processing interrupt state among all the online vCPUs is selected as a target vCPU.

Wherein, the information acquisition module A20 is used to:

After determining that the preset online list is an empty list, an offline vCPU is selected as a target vCPU from the preset offline list according to a preset storage order, wherein the empty list does not include an online vCPU.

The specific implementation of the interrupt request processing system of the present application is basically the same as the above-mentioned embodiments of the interrupt request processing method, and will not be repeated here.

In addition, the present application also provides an interrupt request processing device, the interrupt request processing device includes a memory, a processor and an interrupt request processing program stored in the memory and executable on the processor, wherein the interrupt request processing program implements the steps of the interrupt request processing method as described above when executed by the processor.

In addition, in one embodiment, FIG6 is a schematic diagram of the structure of an interrupt request processing device in one embodiment of the present application. As shown in FIG6, at the hardware level, the interrupt request processing device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. Among them, the memory may include a memory, such as a high-speed random access memory (Random-Access Memory, RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage, etc. Of course, the interrupt request processing device may also include hardware required for other services. The processor, the network interface, and the memory may be interconnected through an internal bus, and the internal bus may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one bidirectional arrow is used in FIG6, but it does not mean that there is only one bus or one type of bus. The memory is used to store programs. In one embodiment, the program may include a program code, and the program code includes a computer operation instruction. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a shared resource access control device at the logical level. The processor executes the program stored in the memory and is specifically used to execute the steps of the above-mentioned anti-channelling control method.

The specific implementation of the interrupt request processing device of the present application is basically the same as the embodiments of the above-mentioned network optimization method, and will not be repeated here.

In addition, the present application also provides a storage medium, which can be a computer-readable storage medium, and the computer-readable storage medium stores one or more programs. The one or more programs can also be executed by one or more processors to implement the steps of each embodiment of the above-mentioned memory access method.

The specific implementation of the computer-readable storage medium of the present application is basically the same as the above-mentioned embodiments of the memory access method, and will not be repeated here.

It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

The serial numbers of the embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) as described above, including a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server or network device, etc.) to execute the methods described in each embodiment of the present application.

The above are only optional embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

A method for processing an interrupt request, wherein the method comprises:

In response to the interrupt request, detecting whether a preset online list includes an online virtual central processing unit vCPU;

If the detection result is yes, obtaining scheduling information and load information corresponding to at least one of the online vCPUs;

Selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

The interrupt request is sent to the target vCPU for interrupt processing.
The interrupt request processing method according to claim 1, wherein the selecting, according to the scheduling information and the load information, an online vCPU with the smallest load as the target vCPU from at least one of the online vCPUs comprises:

Acquire the interruption load quantity corresponding to the online vCPU based on the load information, and acquire the scheduling state corresponding to the online vCPU based on the scheduling information;

After all the scheduling states are in an unprocessed interrupt state, an online vCPU with the smallest load is selected as a target vCPU from at least one of the online vCPUs according to the interrupt load quantity.
The interrupt request processing method according to claim 2, wherein the selecting, according to the interrupt load quantity, an online vCPU with the smallest load from at least one of the online vCPUs as the target vCPU comprises:

Determine an average interruption load quantity corresponding to all of the interruption load quantities;

After the average interruption load quantity is less than a first preset threshold, determining a minimum interruption load quantity among all the interruption load quantities;

The online vCPU corresponding to the minimum interrupt load number is used as the target vCPU.
The interrupt request processing method according to claim 3, wherein after determining the average interrupt load number corresponding to all the interrupt load numbers, the method further comprises:

After the average interrupt load number is greater than or equal to the first preset threshold, an offline vCPU is selected from the preset offline list as the target vCPU according to the preset storage order.
The interrupt request processing method according to claim 2, wherein after acquiring the scheduling state corresponding to the online vCPU based on the scheduling information, the method further comprises:

After a processing interruption state exists in all the scheduling states, determining a maximum interruption load number and a minimum interruption load number among all the interruption load numbers;

determining a ratio between the maximum interruption load number and the minimum interruption load number;

After the ratio is greater than a second preset threshold, the step of selecting an online vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the interrupt load quantity is continued.
The interrupt request processing method according to claim 5, wherein the determining of the maximum interrupt load number The ratio of the minimum interruption load number includes:

After the ratio is less than or equal to the second preset threshold, an online vCPU in a processing interrupt state among all the online vCPUs is selected as a target vCPU.
The interrupt request processing method according to claim 1, wherein after detecting whether the preset online list includes an online virtual central processing unit vCPU, the method further comprises:

After determining that the preset online list is an empty list, an offline vCPU is selected as a target vCPU from the preset offline list according to a preset storage order, wherein the empty list does not include an online vCPU.
An interrupt request processing system, wherein the system comprises:

A detection module, configured to detect whether a preset online list includes an online virtual central processing unit vCPU in response to an interrupt request;

An information acquisition module, configured to acquire scheduling information and load information corresponding to at least one of the online vCPUs when the detection result is yes;

A redirection module, configured to select a vCPU with the smallest load as a target vCPU from at least one of the online vCPUs according to the scheduling information and the load information;

The processing module is configured to send the interrupt request to the target vCPU for interrupt processing.
An interrupt request processing device, wherein the interrupt request processing device comprises a memory, a processor, and an interrupt request processing program stored in the memory and executable on the processor, and when the interrupt request processing program is executed by the processor, the steps of the interrupt request processing method as described in any one of claims 1 to 7 are implemented.
A computer-readable storage medium, wherein an interrupt request processing program is stored on the computer-readable storage medium, and when the interrupt request processing program is executed by a processor, the steps of the interrupt request processing method according to any one of claims 1 to 7 are implemented.