JPH04245522A - Data transfer control method - Google Patents

Abstract

PURPOSE:To improve the throughput of a system in regard of the data transfer control method by improving the data transfer efficiency between a device and a device controller. CONSTITUTION:A memory 12 is provided in a device 11, which forms a system with a device controller 10. When the controller receives a processing request, the control information (including the data if available) is written into the memory 12. Then the device 11 is thrown off and carries out the due processing with use of the control data, etc., stored in the memory 12. This processing result is written into the memory 12 and the end of the processing is informed to the controller 10 with an interruption.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data transfer control method, and more specifically, to a method for controlling data transfer, and more particularly, to a method for controlling data transfer, which is used in a magnetic disk system, etc., and in particular, for a disk system that improves data transfer between a magnetic disk controller and a magnetic disk device. This invention relates to a data transfer control method.

0002

2. Description of the Related Art FIG. 6 is a block diagram of a conventional magnetic disk system, and FIG. 7 is an explanatory diagram of memory arrangement.

In the figure, 1 and 2 are CPUs, 3 is a channel interface, 4 is a magnetic disk control unit (CU), and 5 is a CPU.
is a magnetic disk device, 6 is an adapter interface,
7 indicates a device interface, and 8 indicates a buffer memory.

Further, CH represents a channel, CHSW represents a channel switch, DIR represents a director, and ADP represents an adapter.

Conventionally, a magnetic disk system as shown in FIG. 6, for example, has been known. This system uses two CPUs, and each channel CH is connected to a channel switch CHSW of a magnetic disk controller 4 through a channel interface 3.

The magnetic disk control device 4 is also provided with two directors DIR, each of which is connected to an adapter ADP.

A plurality of magnetic disk drives 5 (DV1, DV2) are connected to these adapters ADP.

Among such systems, there is a system configured as shown in FIG. 7 in order to improve throughput. In the system shown in Figure 7, there are two director DIs.
A buffer memory 8 is commonly connected to R.

The buffer memory 8 is used as a cache memory or as a memory for asynchronous transfer processing. By using the buffer memory 8 in this way, data transfer can be performed efficiently and throughput can be improved.

0010

SUMMARY OF THE INVENTION The conventional devices as described above have the following drawbacks.

(1) For example, in systems using recent optical communication technology, the channel transfer speed has become more than three times that of conventional systems, but devices (magnetic disk drives, etc.)
At most, it is only 1.5 times as large.

[0012] Therefore, when the above buffer memory is used as a cache memory, if the hit rate of the cache is low, data transfer between the buffer memory and the channel will be delayed.
Data transfer between the buffer memory and the device cannot keep up, and performance cannot be improved any further.

(2) Also, in the case of the asynchronous transfer processing described above, data transfer performance does not improve, as in the above case.

(3) Therefore, in the conventional system, throughput cannot be improved any further.

An object of the present invention is to eliminate such conventional drawbacks and improve data transfer between a device and a device controller, thereby increasing system throughput.

[0016]

[Means for Solving the Problems] FIG. 1 is a diagram showing the principle of the present invention. In the figure, 10 is a device controller, 11 is a device, 12 is a memory, and ADP is an adapter.

[0017] In order to achieve the above object, the present invention is constructed as follows.

(1) In a data transfer control method for a system consisting of a device 11 and a device controller 10 that controls the device, the device 11 includes a control information area for storing control information such as media positioning information. , read/write, and a data area for storing data.
When a processing request is received from the host device, the device 11 is left open after writing control information, etc. to the memory 12, and the device 11 writes the information written by the device controller 10 to the memory 12. After performing the processing using the control information, etc., and writing the processing result to the memory 12, the device controller 10 is notified of the completion of the processing by an interrupt, thereby communicating between the device controller 10 and the device via the memory 12. Data transfer between the two is now possible.

(2) In the above configuration (1), a plurality of CPs
When accessed from U, path identification information (CPUID) is written as control information on the memory 12, and when the device is reconnected after being released, the device controller 10 reads out the path identification information. The path decision is now made.

[0020]

[Operations] Since the present invention is constructed as described above, it has the following functions.

In FIG. 1, a device controller 10
When receiving a data write processing request from a host device, do the following:

[0022] The device controller 10 controls the device 1
Control information such as medium positioning information (seek address, record address, etc.) is written in the control information area of a memory 12 provided in the memory 12, and at the same time,
Write write data.

Thereafter, the device controller 10 releases the device when the device 11 starts operating.

The device 11 reads control information from the memory 12 to position the medium, retrieves data from the data area on the memory 12, and writes the data to the medium.

When this process is completed, the device 11 issues a process completion notification to the device controller 10 by interrupt.

Further, when the device controller 10 receives a data read processing request from a higher-level device, the following steps are performed.

First, the device controller 10
writing control information into the control information area of the memory 12;
Leave the device 11 alone (in this case, there is no writing to the data area).

After positioning to the medium, the device 11 reads data from the medium and stores it in the data area of the memory 12.

Thereafter, the device 11 notifies the device controller 10 of the completion of processing by an interrupt. The device controller 10 that has received the failure notification reads the read data from the memory 12.

[0030] In this way, the time required to connect the device controller 10 and the device 10 can be shortened, and
The data transfer speed between the two can be performed at the memory transfer speed rather than the device transfer speed, and the data transfer speed can be increased.

[0031] Also, when accessing from multiple CPUs,
By writing the path identification information (CPUID) in the memory 12, the path can be determined quickly when reconnecting after the device is released.

Therefore, the overhead (processing time for internal processing associated with data processing) of the device controller 10 can be reduced during the initial recombination.

[0033]

Embodiments Hereinafter, embodiments of the present invention will be explained based on the drawings. 2 to 5 are diagrams showing one embodiment of the present invention, in which FIG. 2 is a configuration diagram of a magnetic disk system, FIG. 3 is a configuration diagram of a magnetic disk device, and FIG. 4 is an explanatory diagram of a memory. 5
is a time chart during multiple operation.

In the drawings, the same reference numerals as in FIGS. 1, 6, and 7 indicate the same parts. Also, 5-1 and 5-2 are magnetic disk devices, 13 is an interface (IF), 14 and 15 are automatic data transfer circuits (ADT), 16 is an MPU, 17 is a control storage (CS), and 18 is a serial/parallel converter. circuit, 19 is a variable frequency oscillator, 20 is an access controller, 21 is a voice coil motor (VCM), 22 is a DAS
D (mechanical), 23 indicates a clock generation circuit.

This embodiment uses the device 11 shown in FIG.
In this example, a magnetic disk device is used as the device controller 10, and a magnetic disk control device is used as the device controller 10.

As shown in FIG. 2, the magnetic disk system includes a magnetic disk controller 4 and a magnetic disk device 5.
-1, 5-2, etc. Also, two CPUs 1 and 2 are connected to this magnetic disk system.
The magnetic disk control device 4 is provided with a buffer memory 8 .

Each magnetic disk device 5-1, 5-2 includes:
A memory 12 is provided for each, and this memory 12
Data is transferred between the magnetic disk device and the magnetic disk control device via the .

Each of the magnetic disk drives 5-1 and 5-2 is constructed as shown in FIG.

As shown in FIG. 3, each magnetic disk drive includes a memory 12 and an interface circuit (IF) 13.
, automatic data transfer circuit (ADT) 14, 15, MPU1
6, control storage (CS) 17, serial/parallel conversion circuit 18, variable frequency oscillator (VFO) 19, access controller 20, voice coil motor (VCM) 21,
A DASD (mechanical) 22 and a clock generation circuit 23 are provided.

[0040] The control storage (CS) 17
is for storing control programs, etc., and M
The PU 16 performs various processes using this program.

The clock generation circuit 23 supplies clocks to various parts within the magnetic disk device, and the access controller 20 receives instructions from the MPU 16, and
It sends a control signal to the voice coil motor 21.

[0042] The voice coil motor (VCM) 21 is
This is a motor that drives the ASD (mechanical) 22. The serial-to-parallel conversion circuit 18 performs serial-to-parallel exchange of data written to a medium and data read from a medium.

The automatic data transfer circuit 14 transfers data to and from the magnetic disk control device and internal data transfer, and the automatic data transfer circuit 15 transfers data between the memory 12 and the magnetic disk controller.
It is used to transfer data to.

The memory 12 is a buffer memory that stores positioning information of the medium, other control information, data written to the medium, data read from the medium, and the like. The configuration of the memory 12 will be described below based on FIG. 4.

FIG. 4A shows the internal structure of the memory 12, and FIG. 4B shows the structure of the task control block. The memory 12 is provided with a control information area and a data area. This data area is an area for storing data written to the medium and data read from the medium.

Further, positioning information for the medium, etc. is written in the control information area, but normally, one task control block (referred to as TCB) is generated for one process, and the control information area is write to

The configuration of the task control block (TCB) is as shown in FIG. 4B, and a data area for necessary read/write data is allocated to each task control block. If not, respond device busy).

The task control block has a "front link" and a "back link", which order the arrangement of a plurality of TCBs.

[0049] "CPU (channel path) ID" is path identification information, and holds which channel path should be recombined in order to recombine the channel and the magnetic disk control device.

In addition to read and write operations, the "operation code" specifies the relationship between positioning records and read/write records, format writing, and the like.

"Data length/number of records" represents the key length and data length of the record to be processed, and the number of records to be processed. For example, when processing three records with a data length of 100 bytes, a memory area (buffer area) of 300 bytes is required.

[0052] "Error information" includes read/write processing,
This is used to notify the magnetic disk control device of an error that has occurred during positioning processing, and is written by the MPU 16 of the magnetic disk device.

"R/W buffer address 1...N" is the address of the data area allocated for read/write.

The data transfer control method of the above embodiment will be explained below based on the time chart of FIG.

In the magnetic disk system of this embodiment,
During the seek operation of the magnetic disk device and while waiting for rotation,
Acceptance of the next read or write operation, data transfer of the previous read operation, and completion notification of the previous write operation (
multiplexed operations can be performed.

Furthermore, since the magnetic disk control device 4 can read/write data on all memories 12,
Writing and reading of the above task control block (TCB)
CPU ID when reading/writing write data and interrupts
(channel path), etc.

The magnetic disk devices 5-1 and 5-2 accept the operation start from the magnetic disk control device 4 as a TCB address, then allocate an area on the memory 12, and
Performs error retry, information writing, medium positioning and read/write, and processing completion interrupt.

In FIG. 5, one of the channels of the host device is shown as channel A (CHA), and the other channel is shown as channel B (CHB). Further, the horizontal direction of the figure is shown as a time axis.

Assume that there is now a data write request from channel A. In this case, first, a write command and write data are transferred from channel A to the magnetic disk control device 4. It is assumed that the data transfer rate at this time is, for example, 9 MB/sec.

Upon receiving this processing request, the magnetic disk control device 4 sends these commands and data to the magnetic disk device and writes them onto the memory 12.

[0061] The data transfer rate in this case is 9MB/se.
c. Assuming that these processes take from time t1 to time t2, at t2 the magnetic disk control device 4 notifies channel A of the end of the channel.

The magnetic disk device reads control information from the memory 12 and performs positioning for data writing (during this time, the mechanism is stopped).

[0063] When the operation of the magnetic disk device starts, the magnetic disk control device 4 releases the magnetic disk device.

Next, assume that there is a read processing request from channel B. In this case, first start I/O (SLO
) will be sent. The magnetic disk control device 4 creates a read TCB and stores it in the control information area on the memory 12.

At this time, the magnetic disk device is left empty with no data, and a retry request is issued to the CHB.

At time t3, when the seek operation for write positioning is completed, the write data (request data from the CHA) is read from the memory 12 and written to the medium.

The data writing speed at this time is, for example, 4.
5 MB/sec. This data writing is at time t4.
When the process is completed, the magnetic disk device generates an interrupt to the magnetic disk control device 4 and notifies the magnetic disk control device 4 of the completion of the process.

[0068] Upon receiving this interrupt notification, the magnetic disk control device reads the CPU ID from the memory 12 and notifies the requested channel of device termination.

Further, the magnetic disk device that has completed the data writing starts processing the read request from channel B, and performs read positioning.

Thereafter, at time t5, when the read seek operation is completed, predetermined data is immediately read from the medium (4.5 MB/sec) and stored in the data area of the memory 12.

When this data read is completed, the magnetic disk controller 4 is notified of the completion of the process by an interrupt. After that, the magnetic disk control device 4
2 and notifies channel B of the retry completion, and reads read data from memory 12 and transfers it to channel B (9 MB/sec).

In this way, data can be transferred between the magnetic disk and the magnetic disk control device via the memory 12. Note that if a plurality of magnetic disk devices are connected, the same process is performed for each device.

In this example, the data transfer rate of the magnetic disk device (4.5 MB/sec) is the same as the data transfer rate of the magnetic disk controller and the memory 12 (9 MB/sec).
It is half of c). Therefore, by performing these data transfers separately, it is possible to improve the throughput of the entire system.

Although the embodiments have been described above, the present invention can also be implemented as follows.

(1) Any number of magnetic disk devices may be connected to the magnetic disk control device. In this case, each magnetic disk device is configured as shown in FIG.

(2) In addition to magnetic disk systems, the present invention can be applied to equivalent systems.

(3) The number of host CPUs connected to the magnetic disk control device is not limited to two, but may be any number.

[0078]

[Effects of the Invention] As explained above, the present invention has the following effects.

(1) Data transfer between the device and the device controller can be performed at the transfer speed of the memory, so the data transfer speed becomes faster.

(2) Also, by leaving the device on,
The connection time between the device and the device controller can be shortened.

(3) By using path identification information (CPUID), paths can be easily determined during recombination, and the overhead of the device controller can be reduced.

(4) For the above reasons, the busy rate of the path between the device controller and the device can be reduced, and the system throughput can be improved.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the principle of the present invention.

FIG. 2 is a configuration diagram of a magnetic disk system in an embodiment of the present invention.

FIG. 3 is a configuration diagram of a magnetic disk device.

FIG. 4 is an explanatory diagram of a memory.

FIG. 5 is a time chart during multiple operations.

FIG. 6 is a configuration diagram of a conventional magnetic disk system.

FIG. 7 is an explanatory diagram of a conventional memory arrangement.

[Explanation of symbols]

10 Device controller 11 Device 12 Memory ADP adapter 5-1, 5-2 Magnetic disk device

Claims (2)

[Claims] 1. A data transfer control method for a system comprising a device (11) and a device controller (10) that controls the device, comprising:
) is provided with a memory (12) having a control information area for storing control information such as media positioning information and a data area for storing read/write data. When a processing request is received from the device (11), the device (11) is left open after writing control information etc. to the memory (12).
) performs processing using the control information, etc. on the memory (12) written by the device controller (10), writes the processing result to the memory (12), and then sends the data to the device controller (10). . A data transfer control method, characterized in that data is transferred between a device controller (10) and a device (11) via the memory (12) by notifying processing completion using an interrupt. [Claim 2] When accessed from multiple CPUs,
Path identification information (CPUID) is written as control information on the memory (12), and when recombining after the device is released, the device controller (10) reads the path identification information and determines the path. 2. The data transfer control method according to claim 1, further comprising: performing the following steps.