JPS63128454A - System for managing extended storage device - Google Patents

Abstract

PURPOSE:To minimize the correction value of an operating system needed for the support of an extended storage device ES by giving an access to the ES via firmware based on a control information table. CONSTITUTION:The firmware is provided for the management of an ES together with a control information table which is used when the firmware gives an access to the ES, a specific command used to call the firmware out of an operating system OS of a data processor, and an exit interface means which calls the OS out of the firmware. In such a constitution, the firmware gives an access to the ES based on the control information table, therefore the correction value of the OS needed for the support of the ES can be minimized.

Description

[Detailed Description of the Invention] [Summary] A data processing device equipped with an expanded storage device (ES) includes firmware for managing the expanded storage device (ES), and a firmware for managing the expanded storage device (ES). A control information table used when accessing, a specific command for calling the firmware from the operating system (OS) of the data processing device, and an exit for calling the operating system (OS) from the firmware. By providing an interface means, the firmware accesses the extended storage device (ES) based on the control information table.

[Industrial application field]

The present invention relates to an expanded storage (ES) management method in a data processing device equipped with an expanded storage (ES).

With recent remarkable advances in semiconductor technology, large-capacity semiconductor storage devices have become economically available. ) is often provided.

The expanded storage device (CES) can transfer data between the main storage device (MS) and the main storage device (MS) by a specific instruction (move instruction) issued by the central processing unit (CPU) of the data processing device. Operating system (O
For example, it appears as a paging device when viewed from S).

However, if the operating system (OS) of the data processing device tries to support the management of such an extended storage device (BS), there is a problem in that the overhead of the operating system (OS) increases. For example, an effective management scheme by firmware is required.

[Prior art and problems to be solved by the invention] FIG. 2 is a diagram illustrating a conventional expanded storage (ES) management system, in which (a) shows an example of the overall configuration of a data processing device;
(b) shows a memory map of the main storage device (MS) 2, and (c) shows an example of the configuration of the control table 21.

In the conventional method, the main memory management routine 22 of the operating system (OS) of the data processing device operates between the extended storage device (1!S) 3 and the main memory bag W (MS) 2 based on the control table 21. Manage data transfer.

Normally, in the data processing device shown in (a), the main memory (MS) 2 uses a virtual memory system, and data is often transferred in units of pages/4KB. Based on page-by-page data access using the method, extended storage (US) 3 and main storage (
A management method for the expanded storage device (BS) when controlling data transfer between the MS and the MS 2 will be explained.

In the data processing device, corresponding to each of the main storage device (MS) 2 and the expanded storage device (ES) 3,
Control table 2 consisting of page queues shown in (b)
1, and data transfer control is performed based on this control table 21.

[1] Page-out operation to the main storage (MS)-0 extended storage (US): ■ First, the MS page management routine 22a of the operating system (hereinafter referred to as O3) checks the MS free page queue 21a. When it is recognized that the value has fallen below a certain value, the page with the lowest φ reference frequency is selected from the MS used page queue 21b by searching the LRU control queue (not shown).

■ From the logical address of the page selected above, select the cassage table entry (PT) of the page.
E) is invalidated.

■ The above MS page management routine 22a of O3 uses control information regarding the MS page (for example, MS real page address, key, owner identifier (virtual memory number), virtual page identifier (virtual page address)) as parameters, Control is passed to the ES page management routine 22b.

■ Processing in the ES page management routine: (i) Number of free page frames on extended storage device (hereinafter referred to as US)>1
Case (1) Look at the queue control block (opening CB) of the Es free page queue 21c and take out the free page at the head of the queue.

(2) Necessary information (key, owner identifier, virtual page identifier) is saved in the ES page queue entry (ESPQE) of the removed BS free page queue 21c.

(3) Based on the control information obtained in step (■) above, the main memory (
Hereinafter referred to as MS) The contents of page frame 2 are 1li
Transfer to the above empty page in s3.

(4) Above, invalidated page table entry (P
TE) stores the page address of the ES page to which the above data was transferred.

(5) ES is stored in the ES usage page queue 21d for each owner.
Queue page queue entries (HSPQE).

(6) Queue the above ES page queue entry (ESr' E) at the end of the LRU control queue (not shown).

(ii) When the number of empty page frames of BS = 1 (11
Queue control block (Q
CB) and take out the empty page.

(2) BS of the retrieved BS free page queue 21c
The page queue entry (ESPQE) contains the necessary information (
key, owner identifier, virtual page identifier).

(3) Based on the control information obtained in (■) above, MS 2
Transfer the contents of the 0 page frame to the above empty page of ES3.

(4) Above, invalidated page table entry (P
TE) stores the page address of the US page to which the above data was transferred.

(5) ES is added to the ES usage base queue 21d for each owner.
Queue page queue entries (ESPQf entries).

(6) Queue the Es page queue entry (ESPQE) at the end of the LRU control queue (not shown).

(7) Look at the ES usage page queue 21d and select the oldest Es page.

(8) From the above ES use page queue 21d, the ES page queue entry (ES
PQE).

(9) Transfer the contents of the selected BS page to the MS 2 page that became vacant in (3). At this time,
Restore key information.

QO) The control information (key, virtual page identifier, owner identifier) regarding the selected IEs page is sent to the O8.
``routine that controls page-out to DASD'' in .

0υ The above empty ES page is transferred to the above BS of ES.
It is returned (queued) to the free page queue 21c.

(2) The ES page management routine 22b passes control to the MS page management routine 22a.

[2] Page fault processing: ■ When a program accesses the MS 2 and the hardware detects a page fault (an event in which the page does not exist in the MS 2), a so-called hardware interrupt causes a specific page fault in the O8 to be executed. Control passes to the page fault control routine.

■ In the page fault control routine, the ES
Prepare an area for receiving data from MS 2 from 3, and perform the following page-in processing.

+1) With the above page out process, from MS 2 to B
When paged out to S3, the invalidated page table entry (PTE) is searched, the page required by the page fault control routine is selected, and the page stored in the page table entry (PTE) is selected. Knowing the actual page number of the ES 3 located in the MS 3, the content of the ES page is transferred to the free page frame of the MS 2. At this time, the key information of the page is also restored.

(2) The IES page transferred to the MS 2 is dequeued from the ES user page queue 21d and LR1l queue, and returned (queued) to the BS free page queue 21c.

(3) Resume the processing of the instruction interrupted by the page fault.

In this way, in the conventional system, from MS 2 to BS 3
When an event that requires page-in or page-out occurs, all processing is performed by O8, which causes a problem in that the overhead at O3 becomes large. In particular, there is a problem in that it takes time to process an interrupt when a page fault occurs.

Additionally, since there are multiple types of O3, when a new function such as adding BS3 is added to a data processing device, the various O3s that manage the resources of the data processing device must support that function. In order to do this, it is necessary to make detailed corrections for each O3, which poses a problem in that a large number of man-hours are required.

In view of the above-mentioned conventional drawbacks, the present invention aims to provide a management method that supports the ES by reducing the overhead in the O3 and minimizing the amount of modification of the O3.

[Means for solving problems]

FIG. 1 is a diagram showing an example of the configuration of an extended storage device (ES) management method according to the present invention.

In the present invention, in a data processing device equipped with an expanded storage device 3, a hardware mechanism 20 for managing the expanded storage device 3 is provided.
1, and the hardware mechanism 201 is the expanded storage device 3.
control information table 210 used when managing the data processing device, and the operating system (OS) of the data processing device.
), a specific instruction for calling the hardware mechanism 201, and an exit interface means for calling the operating system (OS) from the hardware mechanism 201, so that the operating system (OS) , an event that accesses the expanded storage device 3 (page out)
When recognizing this, the hardware mechanism 201 is activated with the specific command, and an event (page fault) in which another hardware mechanism different from the hardware mechanism 201 accesses the expanded storage device 3 is detected. In some cases, the exit ink face means transfers control to an operating system (OS), and the operating system (OS) prepares an area for receiving data in the expanded storage device 3, and executes the data in accordance with the specific command. The hardware mechanism 201 is activated, and the activated hardware mechanism 201 is configured to access the expanded storage device 3 based on the control information table 210.

[Effect]

That is, according to the present invention, in a data processing device equipped with an expanded storage device (BS), firmware for managing the expanded storage device (ES) and the firmware for accessing the expanded storage device (ES) are provided. a control information table to be used at the time, a specific command for calling the firmware from the operating system (OS) of the data processing device, and an exit interface means for calling the operating system (OS) from the firmware. By providing the firmware, the expanded storage device (ES) is configured based on the control information table.
), it is possible to minimize the amount of modification of the operating system (OS) necessary to support the expanded storage device (ES), and reduce operational overhead. It has a mitigating effect.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

The above-mentioned FIG.
1. BS control table 210. O5? ! IJ table 211. and a specific command (page out) issued by the OS to call the ES management firmware 201.

page-in command) and O from the ES management farm 201.
The exit interface means for calling 8 are the means necessary to implement the invention.

Hereinafter, with reference to FIG. 1, an expanded storage device (BS) of the present invention will be explained.
Explain the management method.

Even if the present invention is implemented, the page out and page fault processing itself performed between the MS 2 and the 1Es3 will be
There is no particular difference from the conventional method, and the ES management firmware 201 controls access to the BS3 during page-out processing and page fault processing using the BS control table 20.
The difference is that this is done with reference to 1.

Therefore, the details of the page-out processing and page fault processing will be omitted, and the description will focus on the interface control between the main memory management routine 220 and the ES management firmware 201, which is the core of the present invention.

[1] Page-out processing: ■ The MS page management routine 220a of the main memory management routine 220 detects that the MS free page queue 211a has become below a certain value, and
When the least frequently referenced page is selected from 1b, control information regarding that MS page (MS real address,
A page-out command is issued to the ES management farm 201 using the keys (key, owner identifier, virtual page identifier) as operands.

■ The ES management firmware 201 started by the page-out command executes the same page-out processing as before while referring to the ES control table 210, and when the series of processing is completed, returns control to the main memory management routine 220. . Specifically, in the program status word (PSW),
The start address of the main memory management routine 220 is set.

[2] Page fault processing: ■ When a program accesses the MS 2 and hardware (not shown) detects a page fault, the MS access command is interrupted and the ES management firmware 201
, and immediately write the program status word (PSW) mentioned above.
) and passes control to the page fault processing routine (exit routine) of the main memory management routine 220. Therefore, in the present invention, since the transition to the exit routine does not involve interrupt processing as in the conventional case, the overhead at O8 can be reduced.

■ In the page fault processing routine, ES2
An area is prepared to receive the data, and control is returned to the ESt firmware 201 by a specific command (page-in command).

■ The ES management farm 201 executes page-in processing while referring to the IES control table 210 in the same manner as before, and when the processing is completed, the address of the suspended instruction is written to the program status word (PS
W) and resumes the suspended instruction.

In this manner, the present invention provides an operating system (O
From S), the processing for managing the ES is moved to the ES management farm, a US control table is provided that is referenced only by the ES management farm, and the area within the ES is allocated to O3/
Release, LRU management of the allocated area (i.e. ES
ES management using a control table) is executed by the ES management farm by referring to the ES control table, and for the O3 to call the ES management farm as an interface between the ES management farm and the O8. of,
It is characterized by the provision of new page-out and page-in commands and an exit interface for the ES management firm to call the exit routine of O3.

〔Effect of the invention〕

As described above in detail, the extended storage management method of the present invention provides firmware for managing the extended storage (ES) and firmware for managing the extended storage (ES) in a data processing device equipped with the extended storage (ES). A control information table used when the firmware accesses the expansion storage bag fi (ES), a specific command for calling the firmware from the operating system (OS) of the data processing device, and an operating system from the firmware. By providing an exit interface means for calling the system (OS), the firmware accesses the extended storage device (ES) based on the control information table. The amount of modification of the operating system (OS) required to support the device (I!S) can be minimized, and the overhead during operation can be reduced.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of the configuration of an extended storage device (ES) management method according to the present invention. FIG. 2 is a diagram illustrating a conventional expanded storage (ES) management method. It is. In the drawing, 1 is the central processing unit (CPU), 2 is the main memory (MS)
. 3 is an expanded storage device (ES). 201 is the ES management farm. 21 is a control table. 210 is an ES control table. 211 is the O8 control table. 21a and 211a are MS free page queues. 21b and 211b are also used page queues. 21c and 210a are ES empty page queues. 21d, 210b are ES usage page groups 322, 22
0 is the main memory management routine. 22a and 220a are MS page management routines. 22b and 201a are ES page management routines. QCB is a queue control block. ESPQE is a BS page queue entry. MSPQE is MS page queue entry. (a) 7μ (i-≧qξhi22-=-),] set (ES, 11 points)
Figure 2 ¥〇1

Claims (1)

[Claims] A data processing device equipped with an expanded storage device (3), comprising: a hardware mechanism (201) for managing the expanded storage device (3); A control information table (210) used when managing the storage device (3) and the operating system (OS) of the data processing device.
), a specific instruction for calling the hardware mechanism (201), and an exit interface means for calling the operating system (OS) from the hardware mechanism (201). When the OS (OS) recognizes the event of accessing the extended storage device (3), the hardware mechanism (20
1), and when an event is detected in which a hardware mechanism different from the hardware mechanism (201) accesses the extended storage device (3), the exit interface means causes the operating system (OS) to Transferring control, the operating system (OS) prepares an area for receiving data in the expanded storage device (3), starts the hardware mechanism (201) with the specific command, and executes the activated hardware mechanism (201). The hardware mechanism (201) that has been installed in the expanded storage device (3) based on the control information table (210)
An extended storage device management method characterized by accessing.