JP2018116389A - Processing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a processing apparatus capable of realizing load distribution among OSs while maintaining isolation between the OSs and portability of the OSs.SOLUTION: In an ECU 70 which includes physical cores 2 to 4 and executes OSs 51 and 52, the OSs 51 and 52 are allocated with virtual cores a to c, and A and B, respectively, and are virtualized. Load ratios of applications 511 and 521 operating in the OSs 51 and 52, respectively, are determined in advance. The OSs include dividing units for dividing a processing time of each of the virtual cores a to c and A and B into a division time, allocation units for evenly allocating the division time to the physical cores 2 to 4 and making ratios between sum totals of the division time load ratios in which the sum totals of the division time for OSs corresponding to the OSs 51 and 52, respectively, are regarded as the sum totals of time of the physical cores 2 to 4 in whole in a predetermined period, and control units 21, 31 and 41 for executing each processing of the corresponding virtual cores a to c and A and B per allocated division time in the physical cores 2 to 4.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a technique for executing a plurality of OSs on a plurality of physical cores.

  The information processing system described in Patent Document 1 is a system that executes a plurality of OSs with a plurality of physical cores, and an OS that uses each physical core is fixed. The information processing system executes a program specific to each OS on each OS and distributes the load among the OSs. Therefore, the processing of the parallelizable program is divided, and the divided processing is divided into the processing load of the physical core. Are executed in parallel on the OS determined according to the above.

JP 2009-163527 A

  However, in the information processing system, since the divided processes are executed in parallel on a plurality of OSs, the isolation between the OSs is impaired as compared with the case where only the programs unique to each OS are executed. Further, in the information processing system, since it is necessary to add a function for executing the divided processing in parallel to each OS, the portability of the OS is also impaired as compared with the case where only the program unique to each OS is executed. On the other hand, in the above information processing system, when parallel processing is not performed and only a program unique to each OS is executed, even if the load between the OSs is uneven, the physical core can handle the operation of the OS with a high load. The OS load distribution cannot be realized by allocating a lot of processing time.

  The present disclosure has been made in view of the above circumstances, and provides a processing apparatus capable of realizing load distribution between OSs while maintaining isolation between OSs and portability of OSs.

  The present disclosure is a processing device (70) that includes a plurality of physical cores (2 to 4) and executes a plurality of OSs (51, 52), and each of the plurality of OSs has a different virtual core (ac). , A, B) are allocated and virtualized, and the application (511, 521) operating on each of the plurality of OSs has a predetermined load ratio between the applications, and the dividing unit, the allocation unit, And control units (21, 31, 41). The division unit is configured to divide each processing time of the virtual core into division times that are preset times. The assigning unit assigns the division time evenly to the plurality of physical cores, and, for all of the plurality of physical cores in a predetermined period, the sum of the division times corresponding to each of the OSs for each OS is defined as the total time. The ratio is configured to be the load ratio. The control unit is configured to execute processing of the corresponding virtual core for each of the division times allocated by the allocation unit in each of the plurality of physical cores.

  According to the present disclosure, the operating time of the OS on the virtual core assigned to each OS is divided into divided times, and the divided times are assigned to a plurality of physical cores, so there is no need to change the processing on the OS. Therefore, isolation between OSs and OS portability are maintained. Further, when the division time is allocated to a plurality of physical cores, the ratio between the total of the division times corresponding to each OS over the plurality of physical cores in a predetermined period is the load of the application operating on each OS. The division times are allocated equally so that the ratio is obtained. Therefore, the operating time of the OS is a time corresponding to the load ratio between the OSs. That is, load balancing between OSs can be realized while maintaining isolation between OSs and portability of OSs.

  In addition, the code | symbol in the parenthesis described in this column and a claim shows the correspondence with the specific means as described in embodiment mentioned later as one aspect, Comprising: The technical scope of this invention is shown. It is not limited.

It is a block diagram which shows schematic structure of an electronic controller. It is explanatory drawing which shows allocation of the processing time of a virtual core to a physical core. It is a time chart which shows the time change of the processing which each physical core performs.

Hereinafter, embodiments for carrying out the invention will be described with reference to the drawings. In the present embodiment, an electronic control device mounted on a vehicle is assumed as the processing device.
<Configuration>
First, with reference to FIG.1 and FIG.2, the structure of the electronic control apparatus (henceforth, ECU) 70 which concerns on this embodiment is demonstrated. The ECU 70 is configured with the arithmetic device 60 as a main body. The arithmetic device 60 includes a memory 50 and four physical cores 1 to 4. The memory 50 and the physical cores 1 to 4 are connected to a single bus (not shown), and the physical cores 1 to 4 share the memory 50. The physical cores 1 to 4 are cores having the same performance.

  The memory 50 stores two operating systems, OS51 and OS52. In recent years, the number of ECUs in a vehicle has been increasing, but there is a demand for consolidating a plurality of ECUs into one ECU because the vehicle space in which the ECU can be mounted is limited. Therefore, in the present embodiment, the OS 51 and the OS 52 originally mounted on separate ECUs are mounted on one ECU 70.

  Further, the memory 50 stores an application (hereinafter referred to as an application) 511 and an application 512 that operate on the OS 51, and an application 521 that operates on the OS 52. The load ratio between the application 511 and the application 521 is determined in advance. When the load on the application 511 is 3, the load on the application 521 is 2. The application 512 is an application that is more important in terms of vehicle safety than the applications 511 and 521, and is an application related to running such as running, stopping, and turning.

  As shown in FIG. 2, the OS 51 is virtualized with virtual cores x, a, b, and c assigned thereto. The OS 52 is virtualized with virtual cores A and B assigned thereto. The OS 51 processes the application 511 with the virtual cores a, b, and c, and processes the application 512 with the virtual core x. The OS 52 processes the application 521 with the virtual cores A and B. The number of virtual cores that execute the application 511 and the number of virtual cores that execute the application 521 are the same as the settings of the OS originally installed in another ECU.

  The processes of the virtual cores x, a, b, c, A, and B are assigned to the physical cores 1 to 4. At this time, the processing of the virtual core x that executes the application 521 that is important for safety is assigned to the physical core 1 on a one-to-one basis. By assigning only the processing of the virtual core x to the physical core 1, the processing time is surely assigned to the execution of the application 521.

  Here, when the OS 51 and the OS 52 are directly ported to the physical cores 1 to 4, the processes of the virtual cores x, a, b, and c are respectively assigned to the physical cores 1 to 4, and the processes of the virtual cores A and B are They are assigned to physical cores 3 and 4, respectively. Therefore, the time that the processing of the virtual cores b, c, A, and B can occupy the physical cores 3 and 4 is shorter than the time that the processing of the virtual core a can occupy the physical core 2, and the virtual cores b and c , A and B are slower than the processing of the virtual core a.

  On the other hand, as in Patent Document 1, when the OS that uses each physical core is fixed and the application 511 and the application 521 are executed in parallel on the OS 51 and the OS 52, the load on the OS 51 and the OS 52 is distributed. The isolation between OS 51 and OS 52 is impaired. Usually, in the ECU, the importance level is set for the application operating on the OS, and it is required to maintain the importance level for each application. However, if the isolation between the OS 51 and the OS 52 is impaired, the importance of each application cannot be maintained. In addition, the portability of OS 51 and OS 52 is also impaired. Furthermore, overhead is caused by adding functions for executing processes in parallel on the OS 51 and the OS 52 to the OS 51 and the OS 52.

  Therefore, the ECU 70 has functions of a dividing unit and an assigning unit. The functions of the dividing unit and the assigning unit may be provided in any one of the physical cores 2 to 4. The dividing unit divides the operation time during which the OSs 51 and 52 operate on the virtual cores a to c, A, and B into a preset time to generate a divided time. Each division time is associated with a virtual core process executed during the corresponding operation time.

  The assigning unit assigns the generated division time evenly to the physical cores 2 to 4. At that time, the allocation unit is the total time that is the sum of the division times corresponding to the OS 51 and the total sum of the division times that correspond to the OS 52 in the entire physical cores 2 to 4 in the virtualization control cycle T that is a predetermined period. The division time is allocated so that the ratio of the total time becomes the load ratio 3: 2 of the application 511 and the application 521. That is, the ratio of the operating time of the OS 51 and the operating time of the OS 52 on the physical cores 2 to 4 in the virtualization control period T is set to the load ratio of the application 511 and the application 521.

  Specifically, the number of virtual cores that execute the application 511 on the OS 51 is K1, the number of virtual cores that execute the application 521 on the OS 52 is K2, and the load ratio between the application 511 and the application 521 is R1: R2. Further, the number of division times generated from the operation time of each virtual core executing the application 511 is M1, the number of division times generated from the operation time of each virtual core executing the application 521 is M2, and the number of physical cores is L. , N1 and N2 are natural numbers. Then, K1 × M1: K2 × M2 = R1: R2 (1), M1 = (L / K1) × N1 (2), M2 = (L / K2) × N2 (3) The division time numbers M1 and M2 are calculated so as to satisfy the above.

  In this embodiment, since K1 = 3, K2 = 2, and L = 3, M1 = M2 = 3 × N, where N is a natural number. In this embodiment, N = 1 and M1 = M2 = 3. In other words, the virtualization control period T is a period during which the processes corresponding to the minimum number of division times satisfying the expressions (1) to (3) are operated through the virtual cores a to c, A, and B.

  As a result, as shown in FIG. 2, three divided times are generated from the operating times of the virtual cores a to c, respectively, and three divided times are generated from the operating times of the virtual cores A and B, respectively. . Then, a total of 15 division times are equally allocated to the three physical cores 2 to 4, and a period corresponding to the five division times becomes the virtualization control period T. Therefore, in the entire physical cores 2 to 4 in the virtualization control period T, the division time corresponding to the OS 51 is nine and the division time corresponding to the OS 52 is six, and the ratio between the operations of the OS 51 and the OS 52 is the load ratio. Like the above, it becomes 3: 2.

  In FIG. 2, the physical core 2 is assigned division times a1, a2 and a3 for processing the virtual core a and division times b1 and b2 for processing the virtual core b. Also, the physical core 3 is allocated with divided times c1, c2, and c3 for processing the virtual core c, a divided time A3 for processing the virtual core A, and a divided time b3 for processing the virtual core b. It has been. Also, the physical core 4 is allocated with divided times A1 and A2 for processing the virtual core A and divided times B1, B2 and B3 for processing the virtual core B. An OS that uses a specific physical core is not fixed.

  Further, the allocation unit generates the processing order in each of the virtual cores a to c, A, and B so that the processing order of the virtual cores a to c, A, and B by the physical cores 2 to 4 is maintained. Assign a split time. For example, for the divided times b1, b2, and b3 generated from the operating time of the virtual core b, the processing of the corresponding virtual core b needs to be executed in the order of the divided times b1 → b2 → b3. Therefore, when assigning the divided times b1, b2, and b3 to any of the physical cores 2 to 4, after the process corresponding to the divided time b1, the process corresponding to the divided time b2, and then the divided time b3. Make the flow of the corresponding processing.

  In FIG. 2, the process corresponding to the division time b1 is assigned to the third process of the physical core 2, and the process corresponding to the division time b2 is assigned to the fourth process of the physical core 2. A process corresponding to the division time b3 is assigned to the fifth process of the physical core 3. Therefore, when the processing of the virtual core b is executed by the physical cores 2 and 3, the processing order by the virtual core b is maintained. If the processing order of each of the virtual cores a to c, A, and B is maintained when the processing of the virtual cores a to c, A, and B by the physical cores 2 to 4 is performed, It may be different from FIG.

  The physical cores 2 to 4 have the functions of the virtualization control units 21, 31, and 41. In each of the physical cores 2 to 4, the virtualization control units 21, 31, and 41 execute a virtual core process corresponding to the allocated time for each allocated division time. Specifically, the physical cores 2 to 4 read the virtual core process corresponding to the allocation time from the memory 50 and execute it for each allocation time. In the present embodiment, the virtualization controllers 21, 31, 41 correspond to the controller.

<Operation>
Next, the operation of the ECU 70 will be described with reference to the time chart of FIG. The virtualization control cycle T is composed of five periods t1 to t5, and each of the periods t1 to t5 corresponds to a divided time. In the period t1, the physical core 2 executes processing of the virtual core a corresponding to the division time a1, the physical core 3 executes processing of the virtual core c corresponding to the division time c1, and the physical core 4 is divided. The processing of the virtual core A corresponding to the time A1 is executed. Subsequently, in the period t2, the physical core 2 executes the processing of the virtual core a corresponding to the division time a2, the physical core 3 executes the processing of the virtual core c corresponding to the division time c2, and the physical core 4 Executes the processing of the virtual core A corresponding to the division time A2.

  Subsequently, in the period t3, the physical core 2 executes the processing of the virtual core b corresponding to the division time b1, the physical core 3 executes the processing of the virtual core c corresponding to the division time c3, and the physical core 4 Executes the processing of the virtual core B corresponding to the division time B1. Subsequently, in the period t4, the physical core 2 executes the process of the virtual core b corresponding to the division time b2, the physical core 3 executes the process of the virtual core A corresponding to the division time A3, and the physical core 4 Executes the processing of the virtual core B corresponding to the division time B2. Subsequently, in the period t5, the physical core 2 executes the processing of the virtual core a corresponding to the division time a3, the physical core 3 executes the processing of the virtual core b corresponding to the division time b3, and the physical core 4 Executes the processing of the virtual core B corresponding to the division time B3. When the period t5 ends, the process returns to the period t1, and the virtualization control cycle T including the periods t1 to t5 is repeated.

  In each virtualization control cycle T, the loads of the application 511 and the application 521 are evenly distributed to the physical cores 2 to 4. In each virtualization control period T, the processing order in each of the virtual cores a to c, A, and B is maintained. Further, in each virtualization control period T, the virtual cores of the same OS operate uniformly, and the ratio of the operating time of the OS 51 and the operating time of the OS 52 is the ratio of the load of the application 511 and the application 521.

<Effect>
According to the first embodiment described above, the following effects can be obtained.
(1) A divided time is generated from the operating time of the OS 51 on the virtual cores a to c and the operating time of the OS 52 on the virtual cores A and B, and the generated divided time is equally allocated to the physical cores 2 to 4 . At that time, the division time is allocated so that the ratio of the operating time of the OS 51 and the OS 52 becomes the load ratio of the application 511 and the application 521 in the entire physical cores 2 to 4 in the virtualization control cycle T. Therefore, it is possible to realize load distribution between the OSs while maintaining the isolation between the OSs and the portability of the OSs by setting the operation time of the OSs 51 and 52 according to the load ratio of the applications 511 and 521. In addition, since the operating time of the OSs 51 and 52 is defined in advance according to the load ratio of the OSs 51 and 52, it is not necessary to dynamically detect the load of the OSs 51 and 52, thereby suppressing the occurrence of overhead. it can.

  (2) The division times a1 to a3, b1 to b3, c1 to c3, A1 to A3, and B1 to B3 are the physical cores 2 to 4 without changing the processing order in the virtual cores a to c, A, and B. Assigned to. Therefore, the real-time property of processing by the virtual cores a to c, A, and B can be ensured.

(Other embodiments)
As mentioned above, although the form for implementing this invention was demonstrated, this invention is not limited to the above-mentioned embodiment, It can implement in various deformation | transformation.

(A) In the above embodiment, the two OSs 51 and 52 are mounted on the ECU 70, but three or more OSs may be mounted.
(B) In the said embodiment, ECU70 which is a vehicle-mounted apparatus is assumed as a processing apparatus, However, A processing apparatus is not restricted to a vehicle-mounted apparatus, What kind of processing apparatus may be sufficient.

  (C) A plurality of functions of one constituent element in the above embodiment may be realized by a plurality of constituent elements, or a single function of one constituent element may be realized by a plurality of constituent elements. . Further, a plurality of functions possessed by a plurality of constituent elements may be realized by one constituent element, or one function realized by a plurality of constituent elements may be realized by one constituent element. Moreover, you may abbreviate | omit a part of structure of the said embodiment. In addition, at least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other embodiment. In addition, all the aspects included in the technical idea specified only by the wording described in the claim are embodiment of this invention.

  (D) In addition to the processing apparatus described above, the present invention can be implemented in various forms such as a system including the processing apparatus as a component, a vehicle control apparatus, and a load distribution method.

  1, 2, 3, 4 ... physical core, 21, 31, 41 ... virtualization controller, 70 ... ECU, 51, 52 ... OS, 511, 521 ... application, x, a, b, c, A, B ... Virtual core.

Claims (2)

A processing device (70) having a plurality of physical cores (2-4) and executing a plurality of OSs (51, 52),
The plurality of OSs are virtualized by allocating different virtual cores (ac, A, B),
The application (511, 521) operating on each of the plurality of OSs has a predetermined load ratio between the applications,
A division unit configured to divide each processing time of the virtual core into division times that are preset times;
The division time is equally allocated to the plurality of physical cores, and the total of the division times corresponding to each of the OSs for each of the OSs in the entire plurality of physical cores in a predetermined period is defined as a total time. An allocator configured to use a ratio between hours as the load ratio;
Each of the plurality of physical cores includes a control unit (21, 31, 41) configured to execute processing of the corresponding virtual core for each of the division times allocated by the allocation unit. , Processing equipment.   The allocating unit is configured to allocate the division time to the plurality of physical cores so that an order of processing in each of the virtual cores is maintained when the processing of the virtual cores by the physical core is performed. The processing apparatus according to claim 1.

Images