JPS63180150A - Processing system for input/output buffer area acquisition - Google Patents

Abstract

PURPOSE:To decrease the number of dynamic steps needed to fix pages of an input/output buffers by using a means which allocates the real pages instantaneously when a request is given for acquisition of an input/output buffer area of an extended region, etc. CONSTITUTION:When an input/output buffer area request part 10 request an extended region by an area acquisition macroinstruction 13, an area acquisition processing part 11 secures an idle area. Then it is decided whether said secured area is equal to a new page or the extended area. If a new page of the extended area is confirmed, an extended region real page allocation processing part 12 is called out and an idle real page is allocated to a logical page of the secured area. Then the allocated real page number is set at the corresponding page table entry and the CPU-DAT invalid display is erased. When the part 10 is reset, a real page is allocated to an acquired page and a page fixing error is not reset with the following process of an input/output request macroinstruction 14. Thus the useless dynamic steps are eliminated.

Description

Detailed Description of the Invention [Summary] By providing means for immediately allocating a real page when a request is made to acquire an area for an input/output buffer such as an extended region used by program management.

Reduce the number of dynamic steps required to fix pages in the I/O buffer.

[Industrial application field]

The present invention is an input/output buffer area acquisition processing method.

Especially the area l! This relates to an input/output buffer area acquisition processing method that improves processing performance during input/output requests such as program fetch processing in a data processing system that has a processing function that acquires free space in logical space using Fuma macro instructions. be.

[Conventional technology]

FIG. 3 is an explanatory diagram of program fetch processing, and FIG. 4 shows an example of a conventional method.

In FIG. 3, 30 represents a memory, 31 an extended region, 32 a region, 33 an input/output bus sofa, and 34 a program library.

The program fetch process loads the program into the region 32 which is the user area.

It is done as follows.

+11 The human output buffer 33 is secured from the extended region 31 which is the system work area. The size is.

For example, it is 16KB or 32KB.

(2) Secure a program area in region 32.

(3) Create a channel program for reading program 8 in the program library 34 into the input/output buffer 33.

(4) A so-called EXCP macro instruction is used to control input/output to the input/output control unit.

(5) The text data read into the input/output buffer 33 is transferred to the program area and rearranged.

(6) Return the input/output buffer 33.

In the above processing, a program as shown in FIG. 4 is executed.

The program fetch processing unit 20 uses an area acquisition macro instruction (
GE, TMA IN) L 3 is issued.

This causes the GETMA IN processing section 40 to operate.

Free space in the logical space is secured.

To load the program, input/output request macro instruction (E
XCP) 14 is issued. Due to this.

The input/output request processing unit 22 operates. Input/output request processing unit 2
2 calls the page fixation processing routine.

This page fixing process, which fixes the page (channel fixation) of a single output buffer to enable channel DAT, does not hold real pages.

An error will be returned. At this time, the page fixation from the beginning to the middle of the buffer is also canceled. The input/output request processing unit 22 determines the return code of the page fixing process, issues a page load request to the single output buffer in the event of an error, allocates a real page, and calls the page fixing process routine again.

If the page fixing is completed successfully, start the three-person output,
Performs text data reading processing.

Thereafter, the page fixing is canceled by page free processing and control is returned to the issuer of the input/output request macro instruction 14.

The program fetch processing unit 20 transfers the program to the program area, and removes unnecessary crowd covers.
fiI area release macro instruction (FREEMAIN
) 15 calls the FREEMAIN processing unit 23 and releases it.

[Problem that the invention seeks to solve]

I/O buffer for loading programs.

For example, 16 KB to 32 KB is required, and in most cases, real pages are not allocated to all of the buffer area at the time of an input/output request. Therefore, when the input/output request processing unit 22 issues a channel fixing request to the page fixing processing routine, the processing cannot be completed and an error returns. Therefore, the input/output request processing unit 22 performs channel fixing again after performing page load processing, but since channel fixing processing that causes an error is completely wasteful processing, the number of dynamic steps for input/output processing is increased. There is a problem with letting it happen.

To solve this problem, for example, if an input/output buffer is secured in a page fixed area, the real page will be occupied until the ELL area is released, which will impede the effective use of memory. It is also possible that the two-person output request processing unit 22 loads the page after line 9 and then fixes the channel, but the page load process requires a large number of processing steps, and if the real page is allocated to the buffer area. has the problem that page load processing is wasted.

The present invention aims to solve the above problems and to reduce the number of dynamic steps due to memory fixing processing in input/output requests issued by program fetch processing and the like.

[Means for solving problems]

FIG. 1 shows a block diagram of the principle of the present invention.

In FIG. 1, 10 is an input/output buffer area requesting unit that requests an area in an extended region, 11 is an area acquisition processing unit that manages free space and allocates the area to a request source, and 12 is an area requesting unit that requests an area in an extended region. The extended region real page allocation processing unit performs the process of immediately allocating a real page in a certain case, 13 represents an area acquisition macro instruction, and 14 represents an input/output request macro instruction.

When the input/output buffer area requesting unit 10 requests an input/output buffer area (extension region) using the area acquisition macro instruction 13, the +81 area acquisition processing unit 11 secures an empty area in the requested logical space. In the case of the present invention, the area acquisition processing unit 11 further determines whether the secured area is a new page, and also determines whether or not it is a request for an extended region.

If the secured area is a new page in the extended region, the processing unit 12 is called for real page allocation for the extended region, and the processing unit 12 calls the processing unit 12 for real page allocation for the extended region. Allocate free real pages. and.

8. Page table entry (PGTE)
Set the real page number assigned to CPU-DAT.
Perform processing such as erasing the invalid display. That is, the processing that was conventionally performed in the page load processing in the input/output request processing section is performed as an extension of the processing of the +11 area acquisition processing section 11.

After that, when returning to the input/output buffer area requesting unit 10, a real page has been allocated to the acquired page, so in the processing of the subsequent input/output request macro instruction 14, error recovery due to page fixation is eliminated, resulting in unnecessary error recovery. The number of dynamic steps is reduced.

[Effect]

As for the function of allocating a real page to a logical page, conventionally, when a zero-person output buffer with a page fault and a page load does not hold a real page, allocation is generally performed by a page load.

The present invention acquires an input/output buffer area from an extended region during program fetch processing and the like.

This focuses on the fact that an input/output request is made immediately after acquiring the human output buffer, and a real page is allocated when acquiring the area of the extended region.

The number of dynamic steps can be reduced without making any changes to the input/output request processing section.

Also, in the process of allocating real pages, it is possible to omit checking the necessity of so-called page reclaim (taking back pages) and page-in.

Processing can be simplified. Furthermore, when there is an acquisition request for an extended region, a real page is simply allocated, so this page is a pageable area, and unlike acquiring a page fixed area, the real page is occupied until the page is returned. Therefore, real memory can be used effectively.

〔Example〕

FIG. 2 shows an embodiment of the invention.

In FIG. 2, 13 is an area acquisition macro instruction.

14 is an input/output request macro instruction, 15 is an area release macro instruction, 20 is a program fetch processing unit, and 21 is a CETM
AIN processing section, 22 is input/output request processing section, 23 is FR
Represents the EEMAIN processing unit.

Similar to the conventional example shown in FIG. It issues an area release macro instruction 15 to return the output buffer.

GETMA activated by area acquisition macro instruction 13
If the requested area is an extended region, the IN processing unit 21 immediately allocates a real page.
This real page allocation is done in the process.

Since the acquired area is always a new area, there is no need to check whether page retrieval processing is necessary or whether page-in is necessary. Page allocation takes fewer steps than processing.

The process can be completed. Note that this actual page layout is processed as follows.
Except for the above check being unnecessary.

The processing may be similar to the processing conventionally performed in page load processing and the like.

Issuance of the input/output request macro instruction 14 causes the input/output request processing unit 22 to operate, but! ! In the page fixing process executed for the first time, since a real page has already been allocated, no error occurs and input/output can be activated by the star) Ilo (310) instruction without performing page load processing. Thereafter, if the channel is unfixed by page free processing, that input/output buffer becomes a paging target.

The program fetch processing unit 20 transfers the text data on the input/output buffer to the program area in the region, and calls the FREEMAIN processing unit 23 using the area release macro instruction 15 to return the human output buffer that is no longer needed. do.

For example, according to the conventional method shown in FIG. 4, the approximate number of dynamic steps in the part related to the present invention is as follows.

GETMARN processing... 800 steps page fixing processing (error)... 400 steps page load processing... 4.000 steps page fixing processing (normal)... 500 steps page free processing... ... 500 steps In contrast, in the case of the present invention, the total number of dynamic steps is 2, as shown in FIG.

It will be as follows.

GETMA IN processing ... 1,500 steps Page fixed processing (normal) ... 500 steps Page free processing ... 500 steps, approximately 53.6% more dynamic than before The number of steps will be reduced.

〔Effect of the invention〕

As explained above, (1) according to the present invention, real pages are allocated when acquiring an extended region, which is a pager pull area, (2) the number of dynamic steps in related processing is reduced by more than 50%, for example, in program fetch processing, etc. It becomes possible.

System processing performance is improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is an embodiment of the present invention, FIG. 3 is an explanatory diagram of program fetch processing, and FIG. 4 is an example of a conventional system. In the figure, 10 is an input/output buffer area requesting unit, 11 is an area acquisition processing unit, 12 is an extended region real page allocation processing unit, and 13 is an area acquisition macro instruction. 14 represents an input/output request macro instruction.

Claims (1)

[Scope of Claims] In a data processing system having a processing function of acquiring a free area in a logical space by an area acquisition macro instruction (13), a request to acquire an input/output buffer area by the area acquisition macro instruction (13) is provided. , while securing a free area in the requested logical space, and in a series of processing by the area acquisition macro instruction (13), when no real page is allocated to the secured free area, Means for immediately allocating real pages (11)
, 12).