JPH0744405A - Measuring control system for operating time of virtual computer in virtual computer system - Google Patents

Abstract

PURPOSE:To provided an operating time measurement control system in which the performance is improved without use of a real time measurement function for a monitor task managing a physical hardware resource to measure the real time worked for a guest operating system OS. CONSTITUTION:A monitor task 310 acts as controlling a real CPU (PIP), a logic CPU (LIP) task 320 and distribution of a PIP hardware resource to the LIP task 320 as a task and relates to a task control block(TCB) 311. The LIP task 320 acts as controlling control of the LIP itself, start/stop of a guest OS and processing of interception or the like as a task and relates to the TCB 312. The guest OS 330 is an OS working on an LPAR.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus such as a virtual computer system, and more particularly to a virtual computer operating time measurement control system for a virtual computer system.

[0002]

2. Description of the Related Art In recent years, an information processing apparatus called a virtual computer system has been realized, and its usage pattern is becoming general. The virtual computer system is to create a plurality of virtual computers under a single real computer to construct an information processing system. A virtual computer control program (hereinafter referred to as VMCP) is run on the real computer, The operating system (hereinafter referred to as OS) operates under the control of VMCP. As a usage pattern of a computer constituting an information processing system, generally, a method of operating a single OS on a real computer and a virtual computer (hereinafter referred to as VM or VM) operating a plurality of OSs on a single real computer. There is a method called LPAR). The mode in which a single OS operates on a real computer is called the basic mode, and the hardware resources of the real computer by this method are 1
Central processing unit (or CPU or IP)
, One shared main memory (hereinafter referred to as MS), one or more channel paths (hereinafter referred to as CHP)
And)) and. The hardware resources of these real computers are treated as a single resource. A mode in which a plurality of LPARs are constructed on a single real computer and a plurality of OSs are operated is called an LPAR mode.

Generally, in order to realize a plurality of LPARs on a single real computer, a program called VMCP is run on the real computer and a plurality of virtual computers LPAR are generated under the control of this VMCP. Moreover, an independent OS was running on each of these LPARs. Therefore, the VMCP is added with a function of allowing each LPAR to share and use the hardware resources of a single real computer. A single real computer hardware resource for each LP
As a method of sharing the AR, a method of allocating hardware resources under time control under the control of VMCP, or a method of allocating hardware resources
There is a method of logically dividing hardware resources and allocating them to each LPAR exclusively, or a method of allocating them by mixing the above two methods. As a method for allocating them in a mixed manner, for example, the former one of the above-mentioned two methods as a method for sharing real CPUs, that is, a method for allocating real CPUs to each LPAR in a time division manner, Of the above two methods, the latter method, that is, the real CHP and the real M, is used as the sharing method of the MS in the LPAR.
There is a method in which S is logically divided and exclusively allocated to each LPAR.

In such a conventional technique, the above real CPU is used.
As a method of sharing the
The method of allocating to
A conventional method for allocating an actual CPU to an LPAR by time division will be described with reference to FIG. FIG. 1 is a diagram showing a configuration of a method of sharing a real CPU in a conventional technique.

In FIG. 1, the real CPU (hereinafter referred to as PIP) is two PPAs, a PIPA 102 and a PIPB 103.
It has a multiprocessor configuration composed of IP (hereinafter referred to as MP configuration). PIPA102 and PIPB10
3 are each connected to MS 101, and PI
The PA 102 and PIPB 103 are also connected to each other and have an MP configuration. On the PIPA 102 and PIPB 103, a VMCP capable of controlling the MP configuration is operating.
Two LPARs are generated under the control of this VMCP, and LPARA 112 and LPARB 113, respectively.
Is. Each of LPARA 112 and LPARB 113 is composed of two logical CPUs (hereinafter referred to as LIPs). That is, LPARA112 is LIPAA1
21 and LIPAB122 have an MP configuration, and LPARB113 is LIPBA131 and LIPB.
It has an MP configuration composed of B132. Therefore, LP
The OS operating on the ARA 112 controls the LIPAA 121 and the LIPAB 122, and the OS operating on the LPARB 113 controls and operates the LIPBA 131 and the LIPBB 132. Also, LIPAA121 and LIPBA1
31 works only on PIPA 102, LIPAB12
2 and LIPBB 132 operate only on PIPB 103.

Next, the operation of the VMCP and each LIP on each PIP will be described. FIG. 2 shows VMCP, LIPAA121, and LIPB when VMCP and each LIP are dispatched and operate on each PIP.
It is a time chart showing a state of real computer resource allocation to A131, LIPAB122, and LIPBB132. As shown in FIG. 2, as the real computer resources, the sections 201, 203 and 205 are allocated to the VMCP, the sections 202 and 206 are allocated to the LIPAA 121 and LIPAB 122, respectively, and the LIPB is further allocated.
The section 204 is allocated to the A131 and the LIPBB132. This time chart shows a case where LIPAA 121 and LIPBA 131 are fixedly assigned to PIPA 102, and LIPAB 122 and LIPBB 132 are fixedly assigned to PIPB 103.

Section 201 : PI in section 201
VMC running on PA 102 and PIPB 103
P is the LIP held inside the LPARA 112
A time slice value (hereinafter referred to as TIMSLC) that regulates the travel time of the AA 121 and the LIPAB 122 is used to set the operation cutoff time of the LIPAA 121 and the LIPAB 122 in the internal timer of the PIPA 102 and the PIPB 103, and then the LIPAA 121 and the LIPAB 122 are respectively set on the PIPA 102 and the PIPB 103. To start running. At that point, PIPA 102 and PIP
The operation of B103 moves to section 202.

Section 202 : In section 202, L
The IPAA 121 and the LIPAB 122 operate while updating the internal timer set in the section 201 using TIMSLC. LIPAA121 and LIPAB
When the operation abort time of 122 comes, the operations of LIPAA 121 and LIPAB 122 are aborted by a timer interrupt to VMCP and control is transferred to VMCP. When control is transferred to VMCP, PIPA 102 and P
The operation of the IPB 103 moves to section 203.

Section 203 : In section 203,
PIPA 102 and PIPB 103 as in section 201
The VMCP running on the above is the LIPBA 131 and LIPBB 13 held inside the LPARB 113.
PIP using TIMSLC that regulates the running time of 2
LIPBA in the internal timer of A102 and PIPB103
After setting the operation cutoff time of 131 and LIPBB 132, the traveling of LIPBA 131 and LIPBB 132 is started on PIPA 102 and PIPB 103, respectively. At that point, the operations of the PIPA 102 and PIPB 103 move to section 204.

Section 204 : In Section 204, L
The IPBA 131 and the LIPBB 132 operate while updating the internal timer that is set using the TIMSLC in the section 203 as in the section 202. LIPAA1
21 and LIPAB122 operation cutoff time comes,
The operation of LIPBA131 and LIPBB132 is VMC.
Control is terminated by a timer interrupt for P
Moved to CP. When control is transferred to VMCP, P
The operations of the IPA 102 and PIPB 103 move to the section 205.

Section 205 : The operation in the section 205 is the same as that in the section 201, and the VMCP has LIPAA 121 and LIP on the PIPA 102 and PIPB 103, respectively.
Start running IPAB122, PIPA102 and P
The operation of the IPB 103 moves to section 206.

Hereinafter, the LPARA 112 and the LPARB 113 are PIP'ed by repeating the same operation as described above.
It runs on the A102 and PIPB103 in a time-sharing manner. As described above, when a plurality of LIPs are run on one PIP in a time division manner, it is realized by controlling VMCP using TIMSLC.

Next, the relationship between the operations of the VMCP and each LIP will be described with reference to FIG. FIG. 3 is a configuration diagram showing the relationship between a monitor task forming a part of VMCP, a LIP task forming a part of VMCP, and a guest OS operating under the control of the LIP task. Is. In FIG. 3, the monitor task 310 controls the PIP and the LIP task 320, and further controls the LIP task 32.
It manages the allocation of PIP hardware resources to 0, and L through the interface 351 and the interface 353.
It is connected to the IP task 320 and the guest OS 330. The monitor task 310 operates as a task and is associated with a task control block (hereinafter referred to as TCB) 311. The LIP task 320 controls the LIP itself, controls the start / stop of the guest OS, the interception processing, and the like, and is connected to the monitor task 310 and the guest OS 330 via the interfaces 351 and 352. The LIP task 320 operates as a task and is associated with the TCB 312. The guest OS 330 is a guest OS that operates on the LPAR, and the LIP task 320 and the monitor task 3 via the interface 352 and the interface 353.
It is connected with 10. In FIG. 3, guest OS 330
The procedure for activating and operating is as follows. VM
The monitor task 310 is first started with the start-up of the CP. The monitor task 310 dispatches the LIP task 320 via the interface 351 to activate the VM, and passes control to the LIP task 320. The LIP task 320 which has started the operation starts the guest OS 330 via the interface 352 so as to start the guest OS operating on the VM. After that, the user's application runs under the control of the guest OS 330. LI
The SIE instruction is used when the P task 320 activates the guest OS 330 via the interface 352. If any interception condition is detected while the guest OS 330 is running, the SIE instruction being executed is intercepted and control is passed to the LIP task 320 via the interface 352. The interception is a part of the IE function, and its general specification is described in detail in, for example, the aforementioned publications issued by IBM Corporation. The LIP task 320 analyzes the interception factor and executes the corresponding interception process.
In this case, if the interception process is, for example, a wait state interception, the LIP task 320 controls the monitor task 310 via the interface 351.
To, for example, LIP for instruction interception
The task 320 performs a simulation process of the intercepted instruction and restarts the guest OS 330 via the interface 352. If an interrupt condition for some host is detected while the guest OS 330 is running, the SIE instruction being executed is interrupted, and control is passed to the monitor task 310 via the interface 353. The monitor task 310 analyzes an interrupt factor and executes a corresponding interrupt process. In this case, the interrupt processing is, for example, an external interrupt indicating an expiration of the time assigned to the LIP task, the monitor task 310 passes control to another LIP task via the interface 351, and a monitor interrupt is a program interrupt, for example. The task 310 performs a simulation process of a program interrupt and restarts the guest OS 330 via the interface 353.
The user's application runs under the control of the guest OS 330 according to the procedure described above.
In order to add the function of collecting the actual time when S runs as statistical information, it is necessary to realize it so that it is reflected as the operand of the result of the instruction issued by the guest OS itself. In the operand of the result of the instruction issued by the guest OS itself,
To store the real time that this guest OS has run, L
The real-time measurement function must be added to the IP task 320, and in the conventional technique, this measurement function was realized by using the real-time measurement function of the monitor task 310.

Next, the real-time measuring function by the LIP task 320 which is a conventional technique will be described with reference to FIGS. 3, 4 and 5. 4 and 5 are flowcharts showing a processing procedure for realizing the real-time measurement function based on the LIP task 320 which is a conventional technique.

Step 401 : Guest OS on VM
As a pre-process for running a VMP, the VMCP monitor task 310 sends a plurality of LI tasks to be dispatched to the LIP task.
Select from the P task group. A program module that performs this processing is called a scheduler.

Step 402 : Monitor task 310
Activates the LIP task selected by the scheduler from the plurality of LIP task groups, the task-specific information held by the LIP task and hardware information such as the contents of general-purpose registers are transmitted from the TCB of the LIP task. Restore to the hardware register group in PIP.

Step 403 : Monitor task 310
Is an area that holds the cumulative value of the dispatch time of the task placed in the TCB of the LIP task (hereinafter,
In order to update TCCBTIME, a time-of-day clock (hereinafter referred to as TOD) of PIP is used to start counting time. Specifically, the TOD time is the area that holds the dispatch start time of the task placed in the TCB of the LIP task (hereinafter T
CBDSPTM). This processing is a part of the real-time measurement function of the monitor task 310 described above. After that, the program status word of PIP (hereinafter, PSW
"), The old PSW of the LIP saved in the TCB of the LIP task is loaded, and control is passed to the LIP task (hereinafter, the LIP task 320 will be described as an example).

Step 404 : LIP task 320
Initializes a state description (hereinafter referred to as SD), which is an operand of the SIE instruction, in a predetermined manner in preparation for starting the guest OS 330. After that, LIP task 3
Since the guest OS 330 uses the real-time measurement function of the monitor task 310 as a part of the function of the guest OS 330 for measuring the running real-time, it issues a dummy supervisor call instruction (hereinafter referred to as SVC instruction). Issuing this SVC instruction causes an SVC interrupt, and the control is LIP.
It is passed from the task 320 to the monitor task 310.

Step 405 : Monitor task 310
As a part of SVC interrupt processing, TCBTTIME
To maintain. Specifically, the TOD time is an area that holds the dispatch end time of the task placed in the TCB of the LIP task 320 (hereinafter, TCBINTM.
Store). After that, TCBI in the TCB
The contents of TCBDSPTM are subtracted from the contents of NTM and the result is added to TCBTTIME. According to this processing, the dispatched time of the LIP task 320 is accumulated in TCCBTIME in the TCB. After that, when the SVC code that has caused the interrupt is analyzed and it is recognized that the issued SVC instruction is a dummy SVC code, the SVC instruction is regarded as a no operation and the subsequent operation is performed. After that, in order to update TCBTTIME in the TCB, the time counting is started using the PIP TOD. Specifically, the time of TOD is stored in TCBDSPTM of the task placed in the TCB of the LIP task. This process is the same as the above-mentioned step 403. After that, the old PSW of the LIP saved in the TCB of the LIP task is loaded into the PSW of the PIP, and the control is passed to the LIP task 320.

Step 406 : LIP task 320
The TCBTTIME updated in step 405,
Guest O placed in a LIP table (hereinafter referred to as LIPTBL) prepared uniquely to the LIP task
The dispatch start time value of S330 is stored in the area holding it (hereinafter referred to as LIPBIET). Here, LIPBIET holds the dispatch start time value of the guest OS 330.

Step 407 : LIP task 320
Issues an SIE instruction using SD, which is the operand of the SIE instruction initialized in step 404, as an operand, and starts the guest OS 330 on the VM.

Step 408 : Guest OS on VM
330 runs.

Step 409 : When the guest OS 330 running on the VM detects any interception factor, the corresponding interception occurs,
Control is performed from the guest OS 330 on the VM to the LIP task 32.
It is set back to 0.

Step 410 : LIP task 320
Issues a dummy SVC command in order to use the real-time measuring function of the monitor task 310 as a part of the function of the guest OS 330 for measuring the real-time running. Issuing this SVC instruction causes an SVC interrupt, and control is passed from the LIP task 320 to the monitor task 310.

Step 411 : Monitor task 310
Maintains TCBTTIME as a part of SVC interrupt processing. Specifically, as in step 405, TO
The time of D is stored in the area (hereinafter, referred to as TCBINTM) which holds the dispatch end time of the task placed in the TCB of the LIP task 320. Then, from the contents of TCBINTM in the TCB, the TCBDS
Decrement the contents of PTM and add the result to TCBTTIME. By this processing, TCBTTIME in the TCB
In the LIP task 320, the dispatched time is accumulated. After that, the SVC code that caused the interrupt is analyzed, and the issued SVC instruction is a dummy SVC.
When it recognizes that it is a code, it makes the SVC instruction a no-operation, and then the TCBTTI in the TCB.
To update ME, start counting time using PIP TOD. Specifically, the time of TOD is set to the LIP
TCBDSPT of the task placed in the TCB of the task
Store in M. This process is the same as the above-mentioned step 403. After that, the old PSW of the LIP saved in the TCB of the LIP task is loaded into the PSW of the PIP, and the control is passed to the LIP task 320.

Step 412 : LIP task 320
Is the TCBTTIME updated in step 411,
The end value of the dispatch time of the guest OS 330 placed in the LIPTBL prepared for the LIP task is stored in the area (hereinafter, referred to as LIPAIET). Here, LIPAIET is guest O
The time when the dispatch of S330 ends is held.

Step 413 : LIP task 320
Is the LIPTBL prepared for the LIP task.
Area of the guest OS 330 that holds the cumulative value of the dispatch time of the guest OS 330 (hereinafter, LIPETIME).
Say)). According to this processing, the LIPETIME of the LIP table prepared uniquely to the LIP task
In the LIP task 320, the dispatched time is accumulated.

Step 414 : LIP task 320
Analyzes the interception code that caused control to be returned from the guest OS 330 on the VM to the LIP task 320, and prepares to perform interception processing for each code. This process is called an interception first level process.

Step 415 : LIP task 320
Determines whether the intervention of the monitor task 310 is necessary for performing the interception second level process subsequent to the interception first level process. If intervention of the monitor task 310 is needed,
An SVC instruction having a predetermined SVC code is issued and control is passed to the monitor task 310. That is, go to step 417.

Step 416 : LIP task 320
Performs interception second level processing. The specific processing content of this processing is to separately perform the processing corresponding to each interception code. For example, if the interception code indicates instruction interception, the instruction is simulated. After completing this step, the process proceeds to step 404. After that, step 4
The execution of the LIP task 320 is continued from 04.

Step 417 : Monitor task 310
Maintains TCBTTIME as a part of SVC interrupt processing. Specifically, the TOD time is stored in TCBINTM which holds the dispatch end time of the task placed in the TCB of the LIP task 320. After that, from the contents of TCBINTM in the TCB to TCBD
Decrement the contents of SPTM and add the result to TCBTTIME. According to this processing, TCBTTIM in the TCB
In E, the dispatched time of the LIP task 320 is accumulated. After that, the SVC code that has caused the interrupt is analyzed, and if it is recognized that the issued SVC instruction is the SVC code that requires the intervention of the monitor task 310 for the interception processing, the process proceeds to step 418.

Step 418 : Monitor task 310
Performs the interception second level processing subsequent to the interception first level processing by the LIP task 320 in step 414. The specific processing content of this processing is to perform processing corresponding to various interceptions such as input / output request interception and external interrupt interception that cannot be handled by the LIP task 320 alone. If it indicates request interception, monitor task 31
For the purpose of delivering the input / output interrupt information managed by itself to the guest OS 330, 0 simulates the input / output interrupt, and if the interception code indicates the wait state interception, the monitor task 310 Functions to register the LIP task 320 that has caused the wait-state interception in the wait-state task queue list and activate another LIP task that can operate.

Step 419 : Monitor task 310
Determines whether or not it is a case of returning control to the task of the entry source that has made an entry due to interception or interruption. The case of returning control to the entry source task that has made an entry is, for example, the I / O request interception described above, and the case of not returning control to the entry source task that has made an entry is the wait state interface described above. There is a reception. In the case of returning control to the entry source task that has made an entry, go to step 403. Thereafter, the execution of the monitor task 310 is continued from step 403. In the case where control is not returned to the entry source task that has made an entry, go to step 401. After that, the execution of the monitor task 310 is continued from step 401.

Step 421 : When the guest OS 330 running on the VM detects any interrupt factor, a corresponding interrupt is generated, and control is performed by the guest OS 3 on the VM.
It is returned from 30 to the monitor task 310.

Step 422 : Monitor task 310
Performs maintenance of TCBTTIME as a part of the interrupt processing generated in step 421. Specifically, as in step 405, the time of TOD is stored in TCBINTM placed in the TCB of the LIP task 320. After that, from the contents of TCBINTM in the TCB to TCBD
Decrement the contents of SPTM and add the result to TCBTTIME. According to this processing, TCBTTIM in the TCB
The dispatched time of the LIP task 320 is accumulated in E.

Step 423 : Monitor task 310
Analyzes the type of interrupt that caused the interrupt and the interrupt code, and after performing the specified interrupt processing,
Go to step 419. After that, the execution of the monitor task 310 is continued from step 419.

As described above, the guest O according to the conventional technique is
In the method of measuring the actual traveling time of S, it is necessary to pass control to the monitor task by passing through an interrupt before starting the guest OS and at the end of traveling of the guest OS. It was realized according to the designation. With this method, every time the SIE instruction is issued once, 2
Intervention of SVC interrupts due to the issuance of SVC instructions occurs, overhead of SVC interrupt generation processing by hardware and overhead of SVC first level interrupt processing and SVC second level interrupt processing by monitor task occur. Overhead was a major factor in the performance degradation of the virtual computer system, and was a major problem that could not be ignored.

[0038]

In the above-mentioned prior art, when measuring the running real time of the guest OS, it was necessary to intervene in the monitor task to acquire the data for measuring the running real time. For the purpose of switching the control to the above, before the guest OS is started and at the time when the running of the guest OS is finished, even if the interrupt is not originally required, the interrupt is passed, so that the SVC interrupt is generated due to the issuance of the SVC instruction. Therefore, control was passed to the monitor task. In this case, every time the SIE instruction is issued once, the intervention of the SVC interrupt due to the issuance of the SVC instruction occurs twice, and the overhead of the SVC interrupt generation processing by the hardware and the SVC first level interrupt processing by the monitor task occur. In addition, the overhead of the SVC second level interrupt processing occurs, and these overheads are a major cause of the performance degradation of the virtual computer system, which is a major problem that cannot be ignored in the performance of the virtual computer system.

An object of the present invention is to solve the above-mentioned problems of the prior art. When measuring the running real time of the guest OS, originally, the acquisition of the data for measuring the running real time when the interrupt is not necessary In this case, the intervention of the monitor task is unnecessary, and before the guest OS is started and when the guest OS has finished running, if the interrupt is originally unnecessary, the SVC instruction is not issued and, as a result, the SVC interrupt is generated. Depending on the removal, S is issued twice for each SIE command issued.
The SVC interrupt due to the issue of the VC instruction is removed, and the overhead of the SVC interrupt generation processing by hardware and the SVC first level interrupt processing and SV by the monitor task are eliminated.
The purpose of the present invention is to provide a virtual computer operating time measurement control method for a virtual computer system, in which the performance is remarkably improved by removing the overhead of the C second level interrupt processing.

[0040]

According to the present invention, the above-mentioned object is to measure the running real time of the guest OS.
This is achieved by changing the method of acquiring the data for measuring the traveling real time when the interrupt is not necessary as follows.

That is, before the guest OS is started and the guest O
At the end of the running of S, if the interrupt is not originally required, the cumulative value of the task running time measured by the monitor task controlled by the generation of the SVC interrupt by issuing the SVC command is calculated as the guest OS. In order to measure the running real time of the guest OS, the value of TOD of the actual computer obtained by the Sore Clock command directly issued by the LIP task is changed by changing the method of using the measurement data for measuring the running real time of This can be achieved by changing the method used as the measurement data used for.

[0042]

According to the virtual machine operation time measurement control method of the virtual machine system according to the present invention, in FIG. 3, the LIP task 320 executes the LIPTBL prepared for the LIP task before issuing the SIE instruction. The TOD value of the PIP is stored in the LIPBIET holding the dispatch start time of the guest OS 330 placed therein by using the Store Clock instruction. After that, the LIP task 320 issues the SIE instruction using SD, which is the operand of the SIE instruction, as an operand, and the guest O on the VM
Start S330. When some kind of interception occurs while the booted guest OS 330 is running, control is returned from the guest OS 330 on the VM to the LIP task 320. The LIP task 320 stores the TOD value of the PIP in the LIPAIET holding the end time of the dispatch time of the guest OS 330 placed in the LIPTBL prepared for the LIP task, using the Store Clock instruction. Then L
The IP task 320 uses the contents of LIPAIET of LIPTBL prepared for the LIP task as the LIPB.
LIPETIM that holds the cumulative value of the dispatch time of the guest OS 330 by subtracting the content of IET
Add to E. According to this processing, the LI is set in the LIPETIME of the LIPTBL prepared for the LIP task.
The dispatched time of the P task 320 is accumulated.

As described above, as a result of applying the virtual machine operation time measurement control method of the virtual machine system of the present invention, the LIP task 320 has the real time of the monitor task 310 as a part of the function of measuring the real time when the guest OS 330 runs. Do not use the measurement function of. Therefore, it is not necessary to issue a dummy SVC instruction, and the overhead of the hardware SVC interrupt generation processing that occurs with the issuance of the SVC instruction and the overhead of the SVC first level interrupt processing and the SVC second level interrupt processing due to the monitor task Can be eliminated, and a virtual computer operating time measurement control method of a virtual computer system with significantly improved performance can be realized.

[0044]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a virtual computer operating time measurement control system of a virtual computer system according to the present invention will be described in detail below with reference to the drawings.

FIGS. 5 and 6 are processing flowcharts showing the processing procedure of the virtual computer operation time measurement control method of the virtual computer system according to the present invention. The processing procedure will be described in detail below with reference to FIGS. 3, 5, and 6.

In FIGS. 5 and 6, reference numerals 501 to 526 indicate processing steps.

Step 501 : Guest OS on VM
As a pre-process for running a VMP, the VMCP monitor task 310 sends a plurality of LI tasks to be dispatched to LIP tasks.
Select from the P task group. A program module that performs this processing is called a scheduler.

Step 502 : Monitor task 310
Activates the LIP task selected by the scheduler from the plurality of LIP task groups, the task-specific information held by the LIP task and hardware information such as the contents of general-purpose registers are transmitted from the TCB of the LIP task. Restore to the hardware register group in PIP.

Step 503 : Monitor task 310
Is an area that holds the cumulative value of the dispatch time of the task placed in the TCB of the LIP task (hereinafter,
In order to update TCCBTIME, a time-of-day clock (hereinafter referred to as TOD) of PIP is used to start counting time. Specifically, the TOD time is TCBDSPTM that holds the dispatch start time of the task placed in the TCB of the LIP task.
Store at. This process forms part of the existing measurement function. After that, the old PSW of the LIP saved in the TCB of the LIP task is loaded into the PSW of the PIP, and control is performed on the LIP task (hereinafter, LIP task 320).
Will be explained as an example).

Step 504 : LIP task 320
Initializes a state description (hereinafter referred to as SD), which is an operand of the SIE instruction, in a predetermined manner in preparation for starting the guest OS 330.

Step 505 : LIP task 320
Sets the execution start time of the guest OS 330 in LIPBIET as part of the execution time measurement process of the guest OS 330. Specifically, the Store Clock instruction is issued and the current time of TOD is stored in the LIPBIET holding the dispatch start time value of the guest OS 330 placed in the LIPTBL prepared uniquely to the LIP task 320. To do. LIPBIET here
Holds the dispatch start time value of the guest OS 330.

Step 506 : LIP task 320
Issues an SIE instruction using SD of the SIE instruction initialized in step 504 as an operand, and activates the guest OS 330 on the VM.

Step 507 : Guest OS on VM
330 runs.

Step 508 : When the guest OS 330 running on the VM detects any interception factor, corresponding interception occurs,
Control is performed from the guest OS 330 on the VM to the LIP task 32.
It is set back to 0.

Step 509 : LIP task 320
Sets the execution end time of the guest OS 330 in LIPAIET as part of the execution time measurement process of the guest OS 330. Specifically, the Store Clock instruction is issued and the current time of TOD is stored in the LIPAIET that holds the end time value of the dispatch of the guest OS 330 placed in the LIPTBL prepared uniquely for the LIP task 320. To do. LIPAIET here
Holds the dispatch end time value of the guest OS 330.

Step 510 : LIP task 320
Is the LIPTBL prepared for the LIP task.
The contents of LIPBIET are subtracted from the contents of LIPAIET, and the result is added to LIPETIM which holds the cumulative value of the dispatch time of the guest OS 330. According to this processing, the LI prepared uniquely to the LIP task
LIP task 320 for LIPETIME of P table
The dispatched time of is accumulated.

Step 511 : LIP task 320
Analyzes the interception code that caused control to be returned from the guest OS 330 on the VM to the LIP task 320, and prepares to perform interception processing for each code. This process is called an interception first level process.

Step 512 : LIP task 320
Determines whether the intervention of the monitor task 310 is necessary for performing the interception second level process subsequent to the interception first level process. If intervention of the monitor task 310 is needed,
An SVC instruction having a predetermined SVC code is issued and control is passed to the monitor task 310. That is, go to step 514.

Step 513 : LIP task 320
Performs interception second level processing. The specific processing content of this processing is to separately perform the processing corresponding to each interception code. For example, if the interception code indicates instruction interception, the instruction is simulated. After completing this step, the process proceeds to step 504. After that, step 5
The execution of the LIP task 320 is continued from 04.

Step 514 : Monitor task 310
Maintains TCBTTIME as a part of SVC interrupt processing. Specifically, the TOD time is stored in TCBINTM which holds the dispatch end time of the task placed in the TCB of the LIP task 320. After that, from the contents of TCBINTM in the TCB to TCBD
Decrement the contents of SPTM and add the result to TCBTTIME. According to this processing, TCBTTIM in the TCB
In E, the dispatched time of the LIP task 320 is accumulated. After that, the SVC code that has caused the interrupt is analyzed, and if it is recognized that the issued SVC instruction is the SVC code that requires the intervention of the monitor task 310 for the interception processing, the process proceeds to step 515.

Step 515 : Monitor task 310
Performs the interception second level process subsequent to the interception first level process by the LIP task 320 in step 512. The specific processing content of this processing is to perform processing corresponding to various interceptions such as input / output request interception and external interrupt interception that cannot be handled by the LIP task 320 alone. If it indicates request interception, monitor task 31
For the purpose of delivering the input / output interrupt information managed by itself to the guest OS 330, 0 simulates the input / output interrupt, and if the interception code indicates the wait state interception, the monitor task 310 Functions to register the LIP task 320 that has caused the wait-state interception in the wait-state task queue list and activate another LIP task that can operate.

Step 516 : Monitor task 310
Determines whether or not it is a case of returning control to the task of the entry source that has made an entry due to interception or interruption. The case of returning control to the entry source task that has made an entry is, for example, the I / O request interception described above, and the case of not returning control to the entry source task that has made an entry is the wait state interface described above. There is a reception. In the case of returning control to the entry source task that has made an entry, go to step 517. In the case where control is not returned to the entry source task that has made an entry, go to step 501. After that, the execution of the monitor task 310 is continued from step 501.

Step 517 : Monitor task 310
Is the LIP saved in the TCB of the LIP task.
The instruction pointed to by the old PSW instruction address part is SIE
It is determined whether it is an instruction. If the instruction pointed to by the instruction address part of the old PSW of the LIP is not the SIE instruction, the process proceeds to step 503. After that, the execution of the monitor task 310 is continued from step 503. Old P of the LIP
If the instruction pointed to by the instruction address part of SW is the SIE instruction, the process proceeds to step 518.

Step 518 : Monitor task 310
Is the TCBT as part of the LIP task 320 activation process.
Perform maintenance of TIME. Specifically, as in step 503, the TOD time is stored in the TCBDSPTM that holds the dispatch start time of the task placed in the TCB of the LIP task 320.

Step 519 : Monitor task 310
Sets the execution start time of the guest OS 330 in LIPBIET as part of the execution time measurement process of the guest OS 330. Specifically, the Store Clock command is issued and the current time of TOD is stored in the LIPBIET that holds the dispatch start time value of the guest OS 330 placed in the LIPTBL prepared uniquely for the LIP task 320. To do. LIPBIET here
Holds the dispatch start time value of the guest OS 330. After that, TC of the LIP task is added to the PSW of the PIP.
The old PSW of the LIP saved in B is loaded, and control is passed to the LIP task 320. In this case, since the first instruction to be executed is the SIE instruction, execution of the guest OS in step 507 is directly activated.

Step 521 : When the guest OS 330 running on the VM detects any interrupt factor, a corresponding interrupt is generated, and control is performed by the guest OS 3 on the VM.
It is returned from 30 to the monitor task 310.

Step 522 : Monitor task 310
Indicates that the instruction pointed to by the instruction address section of the interrupt old PSW in the prefix save area (hereinafter referred to as PSA) is SI.
It is determined whether it is an E instruction. If the instruction pointed to by the instruction address part of the old PSW is not the SIE instruction, the process proceeds to step 525. If the instruction pointed to by the instruction address part of the old PSW is the SIE instruction, the process proceeds to step 523.

Step 523 : Monitor task 310
Sets the execution end time of the guest OS 330 in LIPAIET as part of the execution time measurement process of the guest OS 330. Specifically, the Store Clock instruction is issued and the current time of TOD is stored in the LIPAIET that holds the end time value of the dispatch of the guest OS 330 placed in the LIPTBL prepared uniquely for the LIP task 320. To do. LIPAIET here
Holds the dispatch end time value of the guest OS 330.

Step 524 : Monitor task 310
Is the LIPTBL prepared for the LIP task.
The contents of LIPBIET are subtracted from the contents of LIPAIET, and the result is added to LIPETIM which holds the cumulative value of the dispatch time of the guest OS 330. According to this processing, the LI prepared uniquely to the LIP task
LIP task 320 for LIPETIME of P table
The dispatched time of is accumulated.

Step 525 : Monitor task 310
Performs maintenance of TCBTTIME as a part of the interrupt processing generated in step 521. Specifically, as in step 514, the TOD time is stored in TCBINTM placed in the TCB of the LIP task 320. After that, from the contents of TCBINTM in the TCB to TCBD
Decrement the contents of SPTM and add the result to TCBTTIME. According to this processing, TCBTTIM in the TCB
The dispatched time of the LIP task 320 is accumulated in E.

Step 526 : Monitor task 310
Analyzes the type of interrupt that caused the interrupt and the interrupt code, and after performing the specified interrupt processing,
Go to step 516. Thereafter, execution of the monitor task 310 is continued from step 516.

As described above, according to the present invention, it is not necessary to use the real-time measuring function of the monitor task that has the role of managing the physical hardware resources in realizing the function of measuring the real time when the guest OS runs. Therefore, it is not necessary to switch control between tasks via dummy interrupts. As a result, the overhead of the interrupt generation processing of the hardware and the overhead of the interrupt processing due to the monitor task can be removed, and an efficient guest OS running real-time measurement function can be realized.

In this example, in order to realize the function of measuring the real time when the guest OS runs, the monitor task 31
0 or the software called LIP task 320 shows an example of programmatically accumulating the dispatch time of the LIP task in LIPETIME of the LIP table prepared for the LIP task. However, the internal processing of the SIE instruction depends on the microcode. The LIPAIET and LIPBI described in this embodiment are used for certain entrance processing and exit processing.
The table entry of ET is accessed, the entry data is created, and the calculation of LIPETIME is also performed.
A method may be adopted in which the internal microcode of the SIE instruction is used and the cumulative value of the dispatch time, which is the calculation result, is stored in the entry, and the LIPAIE described in this embodiment is used inside the SIE instruction by the microcode.
Create data corresponding to T and LIPBIET,
Further, the calculation of LIPETIME may be realized by the internal microcode of the SIE instruction, and the cumulative value of the dispatch time as the calculation result may be reflected in the result operand of the SIE instruction.

[0074]

As described above, according to the present invention, the real-time measuring function of the monitor task, which plays a role of managing the physical hardware resources in realizing the function of measuring the real time when the guest OS is running. It is possible to provide a virtual computer operating time measurement control method for a virtual computer system that is highly efficient and has significantly improved performance by not using the.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration concept of a virtual computer system when a virtual computer operation time measurement control method of a virtual computer system according to the present invention is applied.

FIG. 2 is a time chart showing the concept of the transition of the operating state of the logical processor when the virtual computer operation time measurement control method of the virtual computer system according to the present invention is applied.

FIG. 3 is a diagram showing a configuration concept of a virtual computer control program used for realizing the virtual computer system when the virtual computer operation time measurement control method of the virtual computer system according to the present invention is applied.

FIG. 4 is a flowchart showing a processing procedure of a control method for measuring an operation time of a conventional virtual machine.

FIG. 5 is a flowchart showing a processing procedure of a control method for measuring an operation time of a conventional virtual machine. (Continued from Fig. 4)

FIG. 6 is a flowchart showing a processing procedure of a control method for measuring an operation time of a virtual computer when the virtual computer operation time measurement control method of the virtual computer system according to the present invention is applied.

FIG. 7 is a flowchart showing a processing procedure of a control method for measuring an operation time of a virtual computer when the virtual computer operation time measurement control method of the virtual computer system according to the present invention is applied. (Continued from Fig. 6)

[Explanation of symbols]

 101 Main Storage Device 102, 103 Real Processing Device 111 Virtual Machine Control Program 112, 113 Virtual Machine 121, 122, 131, 132 Logical Processing Device 310 Monitor Task 320 Logical Processor Task 330 Guest Operating System

Claims (1)

[Claims] 1. A virtual computer system constructed on an information processing system comprising one or more real central processing units and a main storage device shared by each of the real central processing units. As a means for operating a plurality of logical central processing units built on the central processing unit, a monitor unit that directly controls the real central processing unit and a logical central processing unit control unit that controls the logical central processing unit are configured. When the virtual computer control program measures the operating time of one or more virtual computers, the logical central processing unit control unit has a means for measuring the operating time of the virtual computer. Computer operating time measurement control method.