JPH02165220A - Disk access control system - Google Patents

Abstract

PURPOSE:To shorten an access time by performing access by connecting access requests whose access areas on a means to queue and a queued disk continue by a means having a data chaining function. CONSTITUTION:Parameters of access requests to the disk are queued in the queue pointer QP in a disk control part DC in the increasing order of disk addresses. At this time, the respective parameters are checked and when the areas on the disk to be accessed continue, the total number of blocks of two parameters from the head of the area is regarded as one parameter and information regarding a transfer area on a main storage memory MM in each area is added as data chain information. Then when the parameters which are made continuous are used to start a disk device MD, the data chain information is set in a DMA control part DH with the data chaining function and the disk device MD is actuated with the parameters of coupled areas through a disk interface control part IF.

Description

[Detailed description of the invention] [Table of contents] Overview Industrial field of application Conventional technology Problems to be solved by the invention Means for solving the problems (Fig. 1) Working examples (Figs. 2 to 4) Effects of the Invention [Summary] Regarding a disk access control method, the present invention aims to shorten the access time when the disk addresses are consecutive among multiple access requests to the disk. In a disk access control method in which queues are rearranged so that disk access addresses are in ascending order when inking, a means for queuing and inking access requests and a means (DH) having a data chain function are provided in the disk control unit. Among the access requests queued in the queue ink means, if the access request areas of the disk are consecutive, the means having the queue ink function (
DH).

[Industrial application field]

The present invention relates to a disk access control method, and more specifically, to a disk access method that can reduce access time when reading or writing data in a magnetic disk device used in an information processing system.

[Conventional technology]

Conventionally, when multiple processes access the same magnetic disk device, these access requests are processed by queuing them in units of requests.

In this case, the access requests are normally queued in the order in which they are issued, but due to the need to process efficiently and shorten the overall access time, the disk access address is set to ``I''+Ill? , : It was also done to rearrange the queues in such a way.

[Problem to be solved by the invention]

The above-mentioned conventional devices had the following drawbacks.

That is, if the queue ink is performed in ascending order as described above, the disk seek wait time can be shortened, but even in this case,
Access operations to disks are performed on a request-by-request basis, and rotational wait time, command activation and collection processing time for disk devices occur on a request-by-request basis.

Therefore, if the disk access addresses are consecutive, the seek time is eliminated, but it is not possible to reduce the rotational waiting time, the activation of commands to the disk device, and the collection processing time.

An object of the present invention is to improve such conventional drawbacks and to shorten the access time when the disk addresses are consecutive among multiple access requests to the disk.

[Means to solve the problem]

FIG. 1 is a diagram showing the principle of a disk access control system according to the present invention.

Now, data is read from the area 1 and area 2 of the disk device MD by the disk controller DC and transferred to the main memory memory MM.
The following describes the case where data is entered into the area 1 data area and area 2 data area.

The disk access request parameters are queued in the queue pointer QP in the disk control unit DC in ascending order of the disk address, as necessary.

At this time, check each parameter and if the areas of the disk to be accessed are continuous (area l and area 2 in the figure).
Suppose that it is continuous. ), one parameter reads the total number of bone blocks of the two parameters from the beginning of the area, and the main memory memo for each area is set as one parameter.
Information regarding the transfer area on the JMM (memory address and data length of area 1 data area and area 2 data area) is added as data chain information.

Next, when starting the disk device MD using this linked parameter, set this data chain information in the DMA control unit with data chain function DI], and use this linked parameter via the disk interface control unit IF. The disk device MD is started using the parameters of the bone area.

Note that the same applies not only when reading data from a disk but also when writing data.

Also, data concatenation is not only done when the disk is started (,
It may be performed when an access request comes.

[Effect]

With the above configuration, among the multiple access request parameters for the disk, the parameters for consecutive areas can be reduced to one parameter. Rotation waiting time between areas can be eliminated.

〔Example〕

Embodiments of the present invention will be described below based on the drawings. Second
The figure is a block diagram of a disk access control system that is an embodiment of the present invention.

The disk control unit includes a request receiving unit 1, a request Q pointer 2, a command execution control unit 4, an execution parameter creation unit 5,
Q extraction unit 6, disk address calculation unit 7, disk address continuity determination unit 8, execution Q pointer 9, data transfer parameter creation unit 11, disk control command creation unit 13
, a disk interface control section 15, a data transfer control section 16, etc. are provided.

Further, the disk control section described above is connected to the disk device 19, as well as the CPU 1B and the main memory 17.

Then, the request parameter 3-1 of the above request Q pointer 2
~3-n, and the execution parameter to of the execution Q pointer 9
-1-10-n is the command (R/W), disk area address, data length, and memory address, and the disk startup parameter 14 is the command (R/W).
, disk area address and total data length.

Furthermore, the data transfer information 12 includes the number of data chains, transfer direction, data length, memory address, etc.

A disk access request from the CPU 18 is received by the request receiving section 1 of the disk control section, and queued in the request Q pointer 2 so that the disk area addresses are in ascending order.

Next, the command execution control unit B 4 creates parameters using the execution parameter creation unit 5.

The execution parameter creation unit 5 takes out the parameters queued in the request Q pointer by the Q extraction unit 6 and inks them into the execution Q pointer 9.

At this time, the disk address calculation unit 7 calculates the final address of the disk access area, and the disk address continuity determination unit 8 compares it with the disk address of the next retrieved parameter, and executes that parameter until it becomes discontinuous. Queue to Q pointer 9.

Next, from these extracted execution parameters, the disk control command creation unit 13 creates disk drive parameters 14 consisting of commands, disk area addresses, and total data length, and the data transfer parameter creation unit 11 creates
Data chain information 12 consisting of the number of data chains and each data transfer information (data length/memory address)
is created.

Based on the disk drive parameters created in this way, the disk device 19 is activated through the disk interface control section 15, and data transfer is started.

The data transfer control unit 16 extracts information on each transfer unit (data length/memory address) from the data transfer information and performs pig transfer with the main memory 17 .

The above embodiment will be further explained in detail with reference to FIGS. 3 and 4.

FIG. 3 is an explanatory diagram when access requests are queued in ascending order of disk addresses in the above embodiment.

The disk access request parameters are the command (R
EAD/WRITE), disk area address (cylinder, head, sector number, etc.), data length of transfer data, address of main memory 17, etc., and is passed from the CPU 18 to the disk control unit at the time of an access request.

The disk control unit uses this parameter to control the disk drive 19 through the disk interface control unit 15.
Start.

However, in a state where access requests from multiple processes etc. are frequently generated, the disk device 19 may already be in an operating state, and in this case, the queue is inked in the waiting queue pointer in the disk control unit. Ru.

During this queue ink, by performing queue ink in which the disk area addresses are in ascending order as described in the prior art, the disk seek time is shortened, the overall access time is shortened, and the concatenation process is This is intended to simplify the process.

FIG. 4 is an explanatory diagram when access requests arranged in ascending order are compared and connected in the above embodiment.

Parameters currently being executed are managed by being queued in the executing Q pointer 9. When the operation based on the parameters queued in this execution Q pointer is completed, the next process is performed.

(1) Dequeue the parameters queued in the request Q pointer 2 and queue them in the execution Q pointer 9.

(2) Add the data length of the parameter to the total data length of the data chain information, and set the data length and memory address as the data chain information.

Furthermore, the end address of the area is determined from the area address and data length of the disk to be accessed.

(3) Check the command and disk area address of the next parameter queued in the request Q pointer, and if the command is the same and the area is continuous from the area end address obtained in (2) above, perform the process in (1) above. Perform processing.

(1) above until a parameter that makes the area discontinuous appears.
, (2), and (3) are repeated.

(4) Set each data length of data chain information and memory address information in the DMA control unit with data chain function, and transfer data for the total data length from the disk area address (disk area address of the first parameter of the execution queue) This command is activated on the disk device 19.

〔Effect of the invention〕

As explained above, the present invention has the following effects.

In other words, among multiple access requests to the disk,
Since disk addresses with consecutive addresses can be executed by starting a single command, the rotation waiting time, command starting, and gathering time can be significantly reduced, and the overall access time can be shortened. .

[Brief explanation of the drawing]

FIG. 1 is a principle diagram of a disk access control method according to the present invention, FIG. 2 is a block diagram of a disk access control method that is an embodiment of the present invention, and FIG. 3 is a diagram of the embodiment shown in FIG. FIG. 4 is an explanatory diagram when access requests are queued in ascending order. FIG. 4 is an explanatory diagram when access requests arranged in ascending order are compared and concatenated in the embodiment shown in FIG. M M-m=main memory QP-queue pointer D[I-DMA control unit with data chain function IF-disk interface control unit MD---disk device.

Claims (1)

[Claims] (1) In a disk access control method that rearranges the queue so that disk access addresses are in ascending order when queuing access requests to the disk, a means (QP) for queuing access requests is provided in the disk control unit. , means (DH) having a data chain function, and among the access requests queued in the queue inking means (QP), when the access request area of the disk is continuous, these access requests are A disk access control method characterized in that access is performed in a connected manner by means (DH) having a data chain function.