JPH04365152A - Sharing controlling system for disk cache device - Google Patents

Abstract

PURPOSE:To ensure the rightness of data not written yet in a magnetic disk at the time of the restoration of power-off by making a system capable of operating at a write back mode only when the setting of the write back mode is finished before and the POWER ON READY signal levels of its own and an opposite systems are '1'. CONSTITUTION:In the case that a magnetic disk device 1K is shared, the system can operate at the write back mode only when the setting of the write back mode is finished before and in addition, the POWER ON READY signal levels of both its own and the opposite systems are '1'. In other conditions, the system operates at a write through mode. Thus, the data in a buffer (not shown in figure) and the state of this buffer can be held even in the case of the power-off of the disk cache device 1d of its own system, and the state of the disk cache device 1h of the opposite system can be always detected. Thus, in the case that the stoppage of power supply is restored after the disk cache device 1d of its own system falls into the power-off, the rightness of the data not written yet in the magnetic disk 1k can be ensured.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shared control system for disk cache devices.

0002

2. Description of the Related Art A disk cache device is constructed using the principle of a cache memory for a main storage device in order to shorten access time to a magnetic disk device. FIG. 6 is a block diagram showing a conventional shared control method for a disk cache device, in which information processing devices of the 0 system and 1 system share control of the magnetic disk device 4i. The 0-system central processing unit (hereinafter referred to as CPU) 4b, the 0-system main storage device 4c, and the 0-system disk cache device 4d are connected to the 0-system bus 4a, and the 1-system system bus 4e is connected to , 1st system CPU 4f, 1st system main storage device 4g, and 1st system disk cache device 4h
is connected. Furthermore, the 0-system disk cache device 4
d, the 1-system disk cache device 4h, and the magnetic disk device 4i are connected by a disk interface 4j. The two disk cache devices 4d and 4h have disk cache memories 4k and 4l, respectively, and a copy of part of the data stored in the magnetic disk device 4i is stored in these disk cache memories 4k and 4l. There is.

[0003] The operation of the disk cache device is classified into a write-through method and a write-back method depending on the method of writing back to the magnetic disk device. The write-through method is a method in which when the CPU performs write access to the magnetic disk device, the disk cache device writes to the disk cache memory and immediately writes (writes back) to the magnetic disk device.

On the other hand, the write-back method means that when the CPU makes a write access to the magnetic disk device, the disk cache device writes to the disk cache memory, but does not immediately write to the magnetic disk device.
This is a method of writing back immediately or sequentially, little by little, according to an algorithm (LRU, FIFO, etc.) written in a microprogram inside the disk cache device.

Therefore, when the disk cache device is operating in write-back mode, in response to a write access from one CPU to a predetermined address on the magnetic disk, write data is written to the disk cache memory, and the magnetic disk device If the data is written only to the disk cache memory of one system, when the CPU of the other system reads the address on the magnetic disk, there will be an inconvenience that the data before the update will be read. Therefore, when writing only to the disk cache memory without writing to the magnetic disk device, the disk cache device of one system writes to the disk cache memory of the other system, thereby writing to the disk cache memory of both systems. In this way, the magnetic disk devices are shared and controlled while maintaining the validity of data that has not been written to the magnetic disk devices.

[0006]

[Problems to be Solved by the Invention] However, in the shared control system with the above configuration, data that has not been written to the magnetic disk device remains on the disk cache device and is
Or, if the power is restored after both systems have been sequentially turned off, the following will occur. For example, as shown in the write operation explanatory diagrams and state diagrams in Figures 7 to 9, (1) While both systems are in operation, unwritten data a is left in the magnetic disk device in the disk cache memory of both systems. Mama (Figure 7), (2)
The 0-system disk cache device 4d is powered off, and only the remaining 1 system continues to operate in write-back mode, and after updating the unwritten data in the magnetic disk device 4i to b (Fig. 8), (3) 0 system disk cache device 4d
When the power is restored, data at the same address in both systems becomes inconsistent (FIG. 9), which poses a problem in that the validity of data that has not been written to the magnetic disk device cannot be guaranteed.

[0007] The present invention solves the above problems, and when the power is restored after the disk cache device is powered off,
An object of the present invention is to provide a shared control method for a disk cache device that can guarantee the validity of unwritten data in a magnetic disk device.

[0008]

[Means for Solving the Problems] In order to solve the above-mentioned problems, the present invention provides a shared control method for a disk cache device in which a magnetic disk device is sharedly controlled by a plurality of information processing devices each having a disk cache device.
The disk cache devices of both systems include a buffer that is an area for storing unwritten data in the magnetic disk device, a buffer backup means, a means for generating a buffer flag indicating the presence or absence of data in the buffer, and a POWER
ON READY signal level from '1' to '0'
means for generating a relay that saves the contents of the buffer flag when the content of the buffer flag changes, means for detecting the POWER ON READY signal of the local disk cache device and the opposite disk cache device, and means for detecting the buffer flag of the opposite disk cache device. , a means for detecting the relay of the disk cache device of the opposite system, and a dedicated interface between the disk cache devices of both systems, the write-back mode is set, and the POWER ONREADY signal level of the disk cache devices of both systems is set. '1
It was configured so that it operates in write-back mode only when ', and in write-through mode otherwise.

[0009]

[Operation] According to the present invention, since the disk cache device sharing control method is configured as described above, when operating in write-back mode, data that has not been written to the magnetic disk device is stored in the buffer. The presence or absence of data within is indicated by the buffer flag. And P.O.
When the level of the WER ON READY signal changes from '1' to '0', the contents of the buffer flag are stored in the relay, and the data in the buffer is retained by the backup means. In addition, the POWER of the own system and opposite system
Detects the level of the ON READY signal and confirms that it is '
If it is 1', it operates in write-back mode, otherwise it operates in write-through mode.

0010

Embodiments Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the basic configuration of a disk cache device sharing control method according to an embodiment of the present invention, FIG. 2 is a block diagram showing a disk cache device sharing control method according to an embodiment of the present invention, and FIGS.
FIG. 5 is a flowchart of processing when power is cut off and restored in the embodiment of the present invention.

In FIG. 1, 1a is the 0 system CPU, 1b is the 0 system main storage device, 1c is the 0 system bus, 1d is the 0 system disk cache device, 1e is the 1 system CPU, and 1f is the 1 system bus.
1g is a 1-system system bus, 1h is a 1-system disk cache device, 1i is a dedicated interface between both disk cache devices, 1j is a disk interface, and 1k is a magnetic disk device.

In FIG. 2, 2a is a 0-system disk cache device, 2b is a 0-system disk cache device control unit,
2c is a 0-system disk cache memory section, 2d is a 0-system buffer, 2e is a 0-system backup means, 2f is a 0-system buffer flag generation circuit, 2g is a 0-system flash flag generation circuit, 2h is a 0-system relay, and 2i is a 0-system POWER O
N READY detection means, 2j is 0 system buffer flag detection means, 2k is 0 system flash flag detection means, 2l
0 system relay detection means, 2m 1 system disk cache device, 2n 1 system disk cache device control unit, 2o
is the 1-system disk cache memory section, 2p is the 1-system buffer, 2q is the 1-system backup means, 2r is the 1-system buffer flag generation circuit, 2s is the 1-system flash flag generation circuit, 2t is the 1-system relay, and 2u is the 1-system POWER. ON
READY detection means, 2v is 1 system buffer flag detection means, 2w is 1 system flash flag detection means, 2x is 1
The system relay detection means 2y is a dedicated interface between the disk cache devices of both systems.

The disk cache device according to the embodiment of the present invention operates in write-back mode when the magnetic disk devices are shared and controlled if the write-back mode is set and the disk cache devices of both systems operate in write-back mode. POWER ON READY signal level is '1'
' is the only time. Also, POWER ON RE
When the signal level of the ADY signal is '1', it indicates that the disk cache device is ready for operation.After the power of the disk cache device is turned on, it rises after a short time and the POWER ON READY signal is activated. After starting up, turn off the power after a while.
It is supposed to be done.

[0014] Since the buffer is backed up by the backup means, the internal data is retained even if the power to the disk cache device is turned off. Furthermore, the buffer flag is POWER ON REA
When the DY signal level is '1', it indicates '1' if there is data in the buffer, otherwise it indicates '0'. And P
When the OWER ON READY signal level changes from '1' to '0', immediately before the power is turned off, if the content of the buffer flag is '1', the relay is turned on, and if it is '0', the relay is turned off. set to the state. Also, when the POWER ON READY signal level of the opposing system changes from '1' to '0', the power is turned off.
If the content of the buffer flag is '1', the relay is set to the ON state, and if it is '0', the relay is set to the OFF state. mode.

The flush flag indicates '1' when the disk cache device is flushing, and indicates '0' at other times. Note that the disk cache device is powered on by the POWER ON of the opposing disk cache device.
It can detect the READY signal level, the presence or absence of data in the buffer, and whether it is being flushed.
When the ON READY signal level changes, the operations shown in the flowcharts shown in FIGS. 3 to 5 are performed.

The flowchart in FIG. 3 shows the process when the POWER ONREADY signal level of the local disk cache device changes from '1' to '0' (step 11).
In this case, as mentioned above, the contents of the buffer flag are transferred to the relay, and then the power is turned off (
Steps 12, 13). The flowchart in FIG. 4 shows the POWER ONREA of the opposing disk cache device.
This shows the operation when the DY signal level changes from '1' to '0' (step 21), and the operation differs depending on the contents of the buffer flag. That is, if the buffer flag is '0', there is no unwritten data in the magnetic disk device in the buffer, so the operation is in write-through mode (steps 22 and 24), and the buffer flag is '0'.
If it is 1', since there is unwritten data in the magnetic disk device in the buffer, the data is flushed and then operates in write-through mode (steps 22, 23, and 24). As a result, the buffer flag becomes '0'.

The flowchart in FIG. 5 shows that when the POWER ONREADY signal level of the own disk cache device changes from '0' to '1' (step 25)
The operation differs depending on the POWER ON READY signal level of the opposing disk cache device. In other words, the POWER ON READY signal level of the opposing disk cache device is '
In the case of 0', if the buffer flag is 0', it operates in write-through mode (steps 27, 31).

On the other hand, when the buffer flag is '1', the state of the relay of the opposite system disk cache device is sensed, and if the relay is in the ON state, as is clear from FIGS. 3 and 4, the self system disk Since the cache device and the opposing disk cache device are powered off at almost the same time, after flushing, they operate in write-through mode (
Steps 27, 28, 29, 31), if the relay is in the OFF state, the opposing disk cache device has been powered off since then, and therefore, as shown in FIG.
Since the data in the buffer of the opposing disk cache device has been flushed, the data in the local disk cache device is invalidated and then operates in write-through mode (steps 27, 28, 30, and 31).

[0019] Also, P of the opposing system disk cache device
When the OWER ON READY signal level is '1' and the buffer flag is '0', it operates in write back mode (steps 26, 32, 36). on the other hand,
When the buffer flag is '1', if the buffer flag of the opposing disk cache device is '0', it means that the opposing disk cache device has been powered off, and therefore the buffer of the opposing disk cache device is Since the data in the internal buffer has been flushed, the data in the local buffer is invalidated and then operates in write-back mode (steps 32, 33, 35, and 36). If the flush flag of the opposing disk cache device is '1', the opposing disk cache device is flushing, so it similarly operates in write-back mode after invalidating the data in the buffer (steps 33 and 34). ,35,
36). Then, if the opposing system buffer flag is '1' and the opposing system flush flag is '1', there is unwritten data in the buffer, so it operates in write-back mode (steps 33, 34, 36). .

With the above processing, in a disk cache device operating in write-back mode, when a magnetic disk device is shared and controlled, one system or both systems can be restored after power is cut off while leaving data in the buffer. It is possible to guarantee the validity of the data in the buffer even if the data is Note that the present invention is not limited to the above embodiments, and various modifications can be made based on the spirit of the present invention.
They are not excluded from the scope of the present invention.

[0021]

As described above in detail, according to the present invention, when a magnetic disk device is sharedly controlled by a disk cache device operating in write-back mode, even if the power of the disk cache device is turned off, Since the data in the buffer and the state of the buffer are maintained, and the state of the opposing disk cache device can always be detected, the validity of the data can be guaranteed.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the basic configuration of a shared control method for a disk cache device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a sharing control method for a disk cache device in an embodiment of the present invention.

FIG. 3 is a flowchart showing a first process when the power is turned off in the embodiment of the present invention.

FIG. 4 is a flowchart showing a second process when the power is turned off in the embodiment of the present invention.

FIG. 5 is a flowchart showing processing at the time of power restoration in the embodiment of the present invention.

FIG. 6 is a block diagram showing a sharing control method of a conventional disk cache device.

FIG. 7 is an explanatory diagram of a first write operation in a shared control method of a conventional disk cache device.

FIG. 8 is a diagram illustrating a second write operation in a conventional disk cache device sharing control method.

FIG. 9 is an explanatory diagram of a state at the time of power recovery in a conventional shared control method of a disk cache device.

[Explanation of symbols]

1a, 1e CPU 1b, 1f Main storage device 1c, 1g System bus 1d, 1h Disk cache device 1i
Dedicated interface 1j
disk interface 1k
Magnetic disk devices 2a, 2m Disk cache devices 2b, 2n Disk cache device control units 2c, 2o Disk cache memory 2d
, 2p Buffers 2e, 2q Backup means 2f, 2r Buffer flag generation circuits 2g, 2s
Flash flag generation circuit 2h, 2t
Relay 2i, 2u POWER ON READY
Detection means 2j, 2v Buffer flag detection means 2
k, 2w Flash flag detection means 2l, 2x
Relay detection means

Claims (2)

[Claims] 1. A shared control method for a disk cache device in which a magnetic disk device is shared and controlled by two systems of information processing devices each having a disk cache device, comprising:
The disk cache devices of both systems include (a) a buffer that is an area for storing unwritten data in the magnetic disk device; (b) backup means for the buffer;
(c) means for generating a buffer flag indicating the presence or absence of data in the buffer; and (d) POWER ON.
means for generating a relay that stores the contents of the buffer flag when the READY signal level changes from '1' to '0'; and (e) detecting the POWER ON READY signals of the own system and opposite system disk cache devices. (f) means for detecting a buffer flag of the opposing disk cache device; (g) means for detecting a relay of the opposing disk cache device;
(h) A dedicated interface between the disk cache devices of both systems, a write-back mode is set, and the POWER O of the disk cache devices of both systems is provided.
A shared control method for a disk cache device, characterized in that it operates in a write-back mode only when the NREADY signal level is '1', and operates in a write-through mode otherwise. 2. The method is characterized by comprising means for generating a flush flag indicating that all data in the buffer is being saved to the magnetic disk device, and means for detecting the flush flag of the opposing disk cache device. 2. A shared control method for a disk cache device according to claim 1.