JPS58171724A - Copy controller of magnetic disc tape - Google Patents

Abstract

PURPOSE:To make continuous read/write possible only with a buffer having 2-track capacity. CONSTITUTION:When a copy instruction is issued from a CPU10 to a copy controller 11, the CPU connects two disc devices 16 and 17 through disc controllers 12 and 13 and disc connecting devices 14 and 15 on a basis of copy control information. A read instruction is issued to one device; and when data is inputted to a buffer of the CPU, a write instruction is issued to the other device. As the result, the copy processing is executed without a disc latency time and is completed in the about half time of the conventional processing time.

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention The present invention relates to a control device that makes it possible to copy magnetic disks and tapes in a computer system at a higher speed than ever before.

Prior Art In conventional copying, the CPU issues a READ command for each track to the device to be copied, and reads the memory of the CPU. Normally, the memory area is set to have a capacity of two tracks, and when the data has entered the capacity of one track or more, a WRITE command is issued to the copying device to output the data. In this process, the I10 command is issued only in units of one track, so (when issuing Ilo for two or more consecutive tracks, a buffer corresponding to the number of tracks is required), the next I10 command is issued once the disk has rotated once. will be executed after consecutive REA
D/WRITI is not possible.

OBJECTS OF THE INVENTION It is an object of the present invention to provide a device that allows continuous READ/WRITE using only a 2-track capacity buffer.

In conventional Cinesystems, the I10 instruction is issued in one unit, and recognition of completion of memory read/write is performed by I10 instructions.
This is a problem with the method of reporting the completion of LO.
As a countermeasure for this, depending on the status of the data in the buffer (memory is used), REA can be sent directly to the device control unit.
To invent a device that has the function of issuing a D/WRITE command, performing this command continuously (equivalent to copying all data using the I10 command), and controlling data in a buffer. It is possible to achieve your goals.

Examples of the invention An embodiment of the present invention will be explained with reference to FIG. 1.2. .

FIG. 1 is a system configuration diagram having the present invention. C.P.
A copy command is sent from Ulo to the copy control device 11 (hereinafter referred to as C).
When the CPC is issued to a PC (abbreviated as PC), the CPC uses copy control information to copy two disk units (hereinafter abbreviated as DKU) 16.
．． 17 is a disk control device (hereinafter abbreviated as DKC) 12.
15 and disk connection device (hereinafter abbreviated as DSC) 14.
15, and for one device, RE
When the AD command is issued and the data is stored in the buffer within the CPC, a WRITE command is issued to the other device.

FIG. 2 is a diagram showing the configuration of a buffer inside the CPC.
3.24.27.28.29 and buffer point 425
．． Clear 26 to 0. Issue a command to read one track to the device α to be copied, and
If the value of the flag 23 is 0, the AD data is distributed to the buffer 20.1 and is input into the buffer 21.

When inputting data starts, the flag 22 or 24 corresponding to each buffer is set to 1, and when the R and EAD of 1 tiger is completed, the flag 27t or 29 is similarly set to 1,
Invert the value of flag 25 for the next operation.

The input pointer 25 is cleared. For the device β to be copied, depending on the value of the flag 28, if it is 0, the flag 22. If it is 1, the value of the flag 24 is determined.
If so, it is recognized that the buffer input operation has started, and a WRITE command is issued to transfer data from the buffer corresponding to the flag. 1 tiger, ri WRITE
When finished, set flag 27 or 29 corresponding to each buffer.
After confirming that the flag is set to 1 (indicating that the data in the buffer is correctly entered), set the flag to 22 or 2.
4. Reset each flag 27 or 29 to 0, invert the value of flag 28 for the next operation, and clear the buffer output pointer.

Furthermore, in order to prevent the reading from overtaking the writing to the buffer, the values of the input pointer 25 and the output point 426 are compared,
If the output pointer becomes the same value or higher, WRITE data transfer is stopped, and the WRITE is restarted from the beginning of the transfer.
Perform TE.

Hereinafter, the READ system continuously issues READ commands based on the values of flags 25, 22, or 24, while checking the buffer free status, and the WRITE system issues flags 28, 22, or 24.
With the value of 4, check the input status to the buffer and write
Issue instructions sequentially.

The command operation for the READ system DKC is to first issue a 5EEK command to the target track, then issue a command to position to the Baum address at the beginning of that track, and then issue a command to read everything from record 1 onwards. After one track is read, a command is issued to locate the home address at the beginning of the next track, and the READ operation is similarly performed in a command chain until the end of the cylinder.

When one cylinder is completed, the head is moved to the beginning of the next cylinder using the 5EEK command, and the same process is repeated.

For WRITE and related command operations, the head position is R.
Similar to the EAD system, when writing from record 1 onward, a WRITE command for each record is created and issued based on the data in the buffer.

Since the above commands are executed by a command chain, it is possible to eliminate rotation waiting losses.

Depending on the buffer status, RE A D/WRITE
If there is a delay in issuing a command, the original command retry function will return the orientation data.

reestablish the connection.

Effects of the Invention Copy processing can be executed without delay in waiting for disk rotation, and the processing is completed in about half the time of the conventional method.

When incorporating into a conventional system, only program copy command support is required, and no modification is required to other devices.

In the CPU, there is no need to use memory as a buffer, and it can be used for other processing.

There is no data transfer on the channel and the load is almost zero.

[Brief explanation of drawings]

FIG. 1 is a system configuration diagram including a copy control device, and FIG. 2 is a buffer configuration diagram within the copy control device. 10...Central processing unit, 11...Copy control device, 12.13...Disk control device, 14.15...Disk connection device, one 17...Disk device, 20.21...・Buffer, 22.23.24.27.2 29...Flag, 25
...c, buffer input pointer, 26...buffer output pointer.

Claims (1)

[Claims] 1. In computer systems using magnetic disks and tapes, for high-speed copying between disks and between disks and tapes, it receives copy commands from the CPU via a channel interface and controls the device being copied. A control device of a device that issues a READ command to the device via the channel interface, stores the read data in a buffer, and copies the data.
A copy control device for magnetic disks and tapes, characterized in that it issues a command to write data in a buffer via the channel interface.