JP3439906B2 - Computer system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer system equipped with a mirror disk device which is composed of two magnetic disk devices and which can also be used as an IPL disk.

[0002]

2. Description of the Related Art FIG. 12 is a block diagram showing a conventional computer system equipped with a mirror disk device. In the figure, 100 is a host device including a central processing unit, and 30.
0 is a mirror disk device, 400 is a common bus that connects the higher-level device 100 and the mirror disk device 300, and 50
0 is a DMA control unit for controlling DMA transfer of the DMA command chain, 600 is a mirror disk handler, 70
0-1 and 700-2 are single disk handlers,
Reference numerals 800-1 and 800-2 are magnetic disk devices.

In the computer system including the mirror disk device 300, a pair of magnetic disk devices 800-1, 8
When both of the storage devices 00-2 store the same data and one magnetic disk device fails, the reliability is improved by reading the data from the other magnetic disk device.

Next, the operation will be described. FIG. 13 is a diagram showing a DMA command used in a computer system including the conventional mirror disk device shown in FIG. In the data read / write operation with respect to the magnetic disk devices 800-1 and 800-2, first, the DMA for controlling the DMA transfer in the mirror disk device 300 from the upper level device 100
A DMA command chain including the SCSI command shown in FIG. 13 is sent to the control unit 500 via the common bus 400 as a DM.
Sent by A transfer.

Next, the DMA control unit 500 transfers the DMA command chain to the mirror disk handler 600 or the single disk handlers 700-1 and 700-2, and performs exclusive control regarding the right to use the common bus 400. Single disk handler 700-1, 70
Reference numeral 0-2 converts the DMA command chain data transmitted from the DMA control unit 500 or the mirror disk handler 600 into a SCSI command, and instructs the magnetic disk device 80 to execute the converted SCSI command.
The data read / write operation at 0-1, 800-2 is performed.

Further, in the mirror disk handler 600, the magnetic disk devices 800-1 and 800 in which the data transmitted from the DMA controller 500 are mirrored.
-2, single disk handler 700-1,
To the magnetic disk device 800-1, 8
The read / write operation for 00-2 is instructed and the execution result of the read / write operation is received from the single disk handlers 700-1 and 700-2.

Then, the mirror disk handler 600 or the single disk handlers 700-1, 700-
2 transmits the execution result of the SCSI command to the upper device 100 by DMA transfer. The upper-level device 100 performs error processing or S to be executed next based on the received execution result.
The operation of generating a CSI command, that is, the issuing operation is performed.

In a conventional computer system including a mirror disk device, when the host device 100 transmits a SCSI command to the mirror disk device 300 via the common bus 400, the DMA command chain shown in FIG. create. Next, one SCSI command is inserted into one DMA command chain, and this DMA command chain is transmitted to the DMA control unit 500 in the mirror disk device 300 by one DMA transfer. The mirror disk handler 600 or the single disk handlers 700-1 and 700-2 sends the SCSI command in the DMA command chain shown in FIG. 13 to the magnetic disk devices 800-1 and 800-2. The magnetic disk devices 800-1 and 800-2 execute this SCSI command.

Single disk handler 700-1, 7
00-2 is a magnetic disk device 800-1, 800-2
The error information or success information of the read / write operation received from is written in the area of the sense buffer 18 indicated by the pointer in the sense entry 16 of the DMA command chain.
The DMA control unit 500 describes the execution result of data transfer by the DMA command chain in the report entry 20.

Next, the high-level equipment 100 receives the DMA command chain, reads the execution result of the SCSI command written in the sense buffer 18, and if it is data indicating an operation error, performs error processing, and succeeds. In the case of the data shown, a process of creating a new DMA command chain to be executed next and transmitting the SCSI command to the mirror disk device 300 is performed.

As described above, in the computer system including the conventional mirror disk device, the host device 100 shown in FIG. 13 issues the DMA command chain. At the time of issuing, one SCSI command and data is stored in each buffer area such as the SCSI command block 17 and the input / output data buffer 19 in the DMA command chain. After that, the higher-level device 100 activates the DMA transmission operation of the DMA command chain and transmits it to the DMA control unit 500 via the common bus 400.

The DM in the mirror disk device 300 will be described below.
The operations of the A control unit 500, the mirror disk handler 600, the single disk handlers 700-1 and 700-2, and the magnetic disk devices 800-1 and 800-2 will be described in more detail. The DMA controller 500 includes a mirror disk handler 600 or a single disk handler 700-
D transmitted from the higher-level device 100 to 1,700-2
Pass the MA command chain. Single disk handler 700-1, 7 that received the DMA command chain
00-2 or the mirror disk handler 600 instructs the magnetic disk devices 800-1 and 800-2 to execute the SCSI command in the DMA command chain, and the magnetic disk devices 800-1 and 800-2 read and write data. The action is executed.

The DMA control unit 500 reports the execution result of the DMA transfer to the report entry 20 of the DMA command chain.
Described in The mirror disk handler 600 or the single disk handlers 700-1 and 700-2 are
The execution result of the CSI command is written in the area in the sense buffer 18 indicated by the point of the sense entry 16. The host device 100 receives the DMA command chain and
After referring to the result of the MA transfer, the area of the sense buffer 18 in which the execution result of the SCSI command is stored is referenced.

FIG. 14 is a diagram showing another configuration of a conventional computer system. In the figure, 101 is a boot loader for performing system startup processing, and 102 is access to the magnetic disk devices 800-1 and 800-2. A disk driver to be controlled, and 103 is a main memory. The same components as those shown in FIG. 12 are designated by the same reference numerals, and the duplicate description will be omitted. Magnetic disk device 800-1,800
-2 stores a program of an operating system (hereinafter referred to as OS). Host device 10
0, main memory 103, and magnetic disk devices 800-1, 8
00-2 is connected to the common bus 400. In this computer system, the system is initialized from one magnetic disk unit 800-1 or 800-2 as an initial program.
Load (hereinafter referred to as IPL).

Next, the operation will be described. Boot loader 1 started on the higher-level device 100 at system startup
01 uses the disk driver 102 to load, for example, the program of the OS main body stored in the magnetic disk device 800-1 into the main memory 103, and then the upper level device 100 starts the OS in the main memory 103. .

[0016]

Since the computer system provided with the conventional mirror disk device is configured as described above, the DMA is performed for each access to each SCSI device.
Create a command chain and use the mirror disk device 300
To the target SCSI
Whenever the number of devices increases or the number of accesses increases, the DMA command chain is set to the mirror disk device 30 each time.
There is a problem in that the access operation must be transmitted to 0, which slows down the access operation.

Disk-specific error information detected by the mirror disk handler 600 and the single disk handlers 700-1, 700-2 in the mirror disk device 300 is stored in an area in the sense buffer 18 of the DMA command chain shown in FIG. Is stored. Host device 10
0 reads the error information stored in the area in the sense buffer 18, and if it indicates an operation error, performs error processing. In this way, the mirror disk handler 600 or the single disk handler 700-1, 7
00-2 does not record operation error information. Therefore, for example, if the computer system goes down and reboots for some reason, the operation error information disappears, and there is a problem that the analysis of the cause of the operation error is hindered.

Further, the mirror disk device 300 is IPLed.
When it is used as a magnetic disk for a computer, the magnetic disk devices 800-1 and 800-2 are accessed via the DMA control unit 500, so the file access speed is slow and the time required for the IPL operation is long. There were challenges.

Further, when the computer system abnormally terminates for some reason and the file system of one of the magnetic disk devices 800-1 and 800-2 forming the mirror disk device 300 is destroyed, IP
When the operation of L is started, since the magnetic disk devices 800-1 and 800-2 in the mirror disk device 300 cannot be matched as a mirror disk, they are stored in the magnetic disk devices 800-1 and 800-2. Moreover, there is a problem that the computer system is stopped because the program of the OS main body is not recognized and an IPL error occurs.

Further, when the IPL operation is executed by the magnetic disk devices 800-1 and 800-2 in the mirror disk device 300 and an IPL operation error occurs, information on the operation error occurred during the IPL operation is saved. There is a problem that the cause investigation is difficult because it is not done.

Further, the user sets the magnetic disk device in the mirror disk device as a magnetic disk for IPL, and the identification number of the magnetic disk device set in the mirror disk management information on the backup memory (hereinafter, I
Call D) is not defined, the IPL
There is a problem that an IPL error occurs when the operation of 1 is started.

The present invention has been made to solve the above problems, and in a computer system having a mirror disk device, it is necessary to improve the data access speed to the mirror disk device and to perform the IPL operation. It is an object of the present invention to obtain a highly reliable computer system which minimizes the time spent for access and efficiently provides information regarding an operation error when an access operation error occurs.

[0023]

[0024]

[0025]

[0026]

[0027]

[0028]

According to a first aspect of the present invention, there is provided a computer system in which a status of an upper device is IPL operating or OS is in a DMA command chain transferred between the upper device and a mirror disk device. If a host device is in IPL operation by providing a flag that can determine whether the operation is in operation, a DMA command chain excluding the part used only for the OS in the DMA command chain is created, DMA transfer is performed, and DMA transfer is performed. It saves time.

In the computer system according to the second aspect of the present invention, when the mirror disk device designated by the user as the IPL disk device is not defined as the mirror disk device in the mirror disk management information in the backup memory, the mirror disk device SCS same as disk device
The IPL operation is performed from the magnetic disk device having the ID of I, and the information related to the magnetic disk device of the IPL is stored in the backup memory and the IP due to the user setting error is set.
The number of occurrences of L operation error is reduced.

[0030]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below. Embodiment 1. FIG. 1 is a block diagram showing a computer system including a mirror disk device according to a first embodiment of the present invention. In the figure, 1 is a host device (host device means), 2 is a backup memory (memory means), 3 Is a mirror disk device, and 4 is a common bus that connects the higher-level device 1 and the mirror disk device 3. Mirror disk device 3
Includes a DMA controller 5 (DMA transfer control means), a mirror disk handler 6 (mirror disk control means), single disk handlers 7-1 and 7-2 (single disk control means), and magnetic disk devices 8-1 and 8-2. Have

In the computer system according to the first embodiment,
An SC that specifies access operations to multiple SCSI devices
The higher-level device 1 creates a DMA command chain including the SI command, transmits this DMA command chain to the DMA control unit 5 in the mirror disk device 3 by one DMA transfer, and the magnetic disk devices 8-1, 8 -SC such as -2
The I-device access operation is performed.

Next, the operation will be described. FIG. 2 shows D used in a computer system having a mirror disk device.
It is a figure which shows the template of MA command chain.
When a SCSI command is transmitted from the host device 1 to the mirror disk device 3 via the common bus 4, a DMA command chain as shown in FIG.
A plurality of SCSI commands are written in the A command chain, and D in the mirror disk device 3 is executed by one DMA transfer.
This DMA command chain is transmitted to the MA control unit 5.

Sense entries 16-1, .., corresponding to each SCSI command in the DMA command chain. ． ． 16
-3 point to the sense buffers 18-1 ,. ． ． ，
In the area of 18-3, the magnetic disk devices 8-1, 8-2
The mirror disk handler 6 or the single disk handlers 7-1 and 7-2 writes the access error information or the access success information received from DMA controller 5
Describes the execution result of data transfer by the DMA command chain in the report entry 20.

The host device 1 uses the DMA control unit 5 to perform DM
Upon receiving the A command chain, it knows that the DMA transfer is completed. The result of the DMA transfer is judged from the report entry 20 of the DMA command chain, and whether the execution of the SCSI command is correct or not is judged by each sense buffer 18-.
1 ,. ． ． , 18-3 to obtain the execution result information of the SCSI command.

The mirror disk handler 6 and the single disk handlers 7-1 and 7-2 use the S command based on all the SCSI commands included in the DMA command chain.
Magnetic disk devices 8-1, 8 corresponding to CSI command
-2 until the execution result is received from the magnetic disk devices 8-1 and 8-2.
The A control unit 5 is not notified of the completion of the DMA transfer.

As described above, according to the computer system of the first embodiment, when the DMA command chain is transmitted to the mirror disk device 3 by the DMA transfer, the host device 1 is the access target in one DMA command chain. Insert a plurality of SCSI commands for a plurality of SCSI devices, and D to the mirror disk device 3 via the common bus 4.
The MA command chain is transmitted, and the DMA control unit 5, the mirror disk handler 6, and the single disk handlers 7-1 and 7-2 receive this DMA command chain and access each SCSI device.
A plurality of SCSI devices can be accessed by transfer, and a high-speed access operation can be realized.

Embodiment 2. 3 is a block diagram showing a computer system including a mirror disk device according to a second embodiment of the present invention. In the figure, 21 is a SCSI command area in the backup memory 2, 27 is an access control flag in the backup memory 2. Is. The same components as those shown in FIG. 1 are designated by the same reference numerals, and the duplicate description will be omitted. The backup memory 2 is
It is a memory area that can be accessed in both directions from the higher-level device 1, the mirror disk handler 6, and the single disk handlers 7-1 and 7-2 via the common bus 4.

In the computer system according to the second embodiment,
DM of the DMA command chain via the DMA control unit 5
The backup memory 2 transmits / receives the SCSI command and the execution result of the SCSI command between the host device 1 and the mirror disk handler 6 and the single disk handlers 7-1 and 7-2 in the mirror disk device 3 without executing the A transfer. It is done at high speed through.

Next, the operation will be described. In the computer system according to the second embodiment, the host system 1 writes the SCSI command in the SCSI command area 21 reserved for the SCSI command processing on the backup memory 2. After the writing is completed, the higher-level device 1 sets the access control flag 27 to the inaccessible state from the higher-level device 1 and the accessible state from the mirror disk device 3 in order to exclusively control the SCSI command area 21 in the backup memory 2.

The mirror disk handler 6 or the single disk handlers 7-1 and 7-2, when the access control flag 27 indicates an accessible state from the mirror disk device 3, sends the SC in the SCSI command area 21.
The SI command is input and executed, and the execution result is written to a predetermined area in the backup memory 2. Furthermore, the access control flag 27 is set to the accessible state from the higher-level device 1,
The mirror disk device 3 sets the access inaccessible state.

The host computer 1 refers to the access control flag 27, and if the access control flag 27 is accessible from the host computer 1, the execution result of the SCSI command is read from a predetermined area in the backup memory 2 and The SCSI command of No. 2 is written in the backup memory 2, and the access control flag 27 is set to the inaccessible state from the higher-level device 1 and the accessible state from the mirror disk device 3.

Host device 1, mirror disk handler 6,
The single disk handlers 7-1 and 7-2 have a function of constantly monitoring the access control flag 27 and repeatedly performing the series of operations described above for all SCSI commands.

As described above, according to the second embodiment, the higher-level device 1, the mirror disk handler 6, and the single disk handler 7-1 are executed via the backup memory 2 without performing the DMA transfer using the DMA command chain. , 7-2 are directly transferred between the SCSI commands, the access speed of the computer system can be increased.

Embodiment 3. FIG. 4 is a block diagram showing a computer system including a mirror disk device according to a third embodiment of the present invention. In the figure, 9 is a SCSI controller for controlling the operations of the magnetic disk devices 8-1, 8-2, and 25. Is a disk error information area for storing errors unique to the magnetic disk devices 8-1 and 8-2. The same components as those shown in FIG. 1 are designated by the same reference numerals, and the duplicate description will be omitted.

In the computer system according to the third embodiment,
The mirror disk handler 6 and the single disk handlers 7-1 and 7-2 are magnetic disk devices 8-1 and 8-
2 is detected, the DMA shown in FIG.
Information indicating the occurrence of an operation error of the magnetic disk devices 8-1, 8-2 is stored in the report entry 20 of the command chain, and the magnetic disk devices 8-1, 8 are stored in the disk error information area 25 of the backup memory 2. By writing the information indicating that the operation error has occurred, the operation error information is prevented from being erased.

Specifically, in a state where the host device 1 cannot refer to the error information, for example, when the DMA controller 5 controlling the common bus 4 fails, the host device 1 is notified of the operation error information by DMA transfer. If not possible, or if the computer system goes down and reboots, the host system 1
Disk error information area 25 in backup memory 2
The operation error can be analyzed by looking at the information stored in the disk which indicates the occurrence of the operation error of the magnetic disk devices 8-1, 8-2.

Next, the operation will be described. SCS in the DMA command chain DMA-transferred from the higher-level device 1
The I command is received by the mirror disk handler 6 and the single disk handlers 7-1 and 7-2. Mirror disk handler 6 and single disk handler 7-1,
7-2 is the SC received by the SCSI controller 9
Instruct execution of SI command. Magnetic disk device 8-
The SCSI controller 9 or the magnetic disk devices 8-1, 8-2 themselves detect the operation errors of 1, 8-2.

When an operation error is notified from the SCSI controller 9, the mirror disk handler 6 or the single disk handlers 7-1 and 7-2 are operated by the sense buffers 18-1 ,. ． ． , 1
The operation error information is written in the area 8-3, a flag indicating that a device error has occurred is set in the report entry 20, and the host device 1 is set via the DMA controller 5.
Notify that an operation error has occurred.

If the device error flag indicating the occurrence of the operation error is set in the report entry 20 of the DMA command chain, the host device 1 detects the sense buffer 1
8-1 ,. ． ． , 18-3, and executes error processing such as notifying an application program (not shown) that an operation error has occurred.

Further, the mirror disk handler 6 or the single disk handlers 7-1 and 7-2, register information in the SCSI controller 9 when an operation error occurs, or operation from the magnetic disk devices 8-1 and 8-2. The error information is written in the disk error information area 25 in the backup memory 2.

FIG. 5 shows the sense buffers 18-1 ,. ． ． , 18-3 is a diagram showing operation error information for the number of retries in the area. Mirror disk handler 6 or single disk handler 7-1,
When 7-2 notifies the occurrence of an operation error, first SCSI
The controller 9 is initialized and a specified number of retries are executed. If the operation error occurs even after the specified number of retries, the sense buffers 18-1 ,. ． ． , 18-3, the operation error information for the number of retries is written in the area of FIG. 5, and an error flag is set in the area of the report entry 20 via the DMA control unit 5, and also in the backup memory 2. In the disk error information area 25, the sense buffer 18-
1 ,. ． ． , 18-3, the time stamp 26 and the ID of the SCSI device are further added and written.

FIG. 6 is a flowchart showing the operation of writing data to the disc error information area 25. Disk error information area 25 in backup memory 2
Is a cyclic storage area. First, the operation error information obtained by the retry operation for the specified number of times is set as one set, and it is checked whether or not N sets (N ≧ 1 integer) of error information are already stored (step ST1). If N
Timestamp 2 if the set information is not stored
6 is used to select the area next to the latest error information, and the data is stored in the area next to the latest error information (step ST
2). Conversely, if N sets of information have already been stored,
The oldest error information area is deleted using the time stamp 26 and data is stored in the obtained area (step S
T3).

The size of the operation error information of one retry written in the disk error information area 25 is fixed, and as shown in FIG. 5, the operation error information of N sets of magnetic disks can be stored n times each. To have.

As described above, according to the third embodiment, the mirror disk handler 6 and the single disk handlers 7-1 and 7-2 are the magnetic disk device 8-1,
When the occurrence of the operation error of 8-2 is detected, the information indicating the occurrence of the operation error of the magnetic disk devices 8-1, 8-2 is stored in the report entry 20 of the DMA command chain and the disk error in the backup memory 2 is stored. In the information area 25, the history information of the operation error occurrence of the magnetic disk devices 8-1, 8-2 is written. Host device 1 is DMA
Even when the report entry 20 in the command chain cannot be referred to, the information in the disk error information area 25 can be referred to and the operation error analysis can be efficiently performed without erasing the operation error history information. .

Fourth Embodiment FIG. 7 is a block diagram showing a computer system provided with a mirror disk device according to Embodiment 4 of the present invention. In the figure, 10 is a main memory, 11 is a boot loader (boot loader means), and 13 is a mirror disk driver. . The same components as those shown in FIG. 1 are designated by the same reference numerals, and the duplicate description will be omitted.

In the computer system according to the fourth embodiment, the boot loader 11 issues an instruction to the mirror disk driver 13 and the magnetic disk device 8-in the mirror disk device 3.
When 1 and 8-2 are used as a magnetic disk device for IPL and an operation error occurs during the IPL operation, the operation error information is stored in a predetermined area in the backup memory 2 to facilitate the clarification of the IPL error. It is a thing.

The host device 1 is a mirror disk driver 1
3 has a boot loader 11 for loading the OS main body programs stored in the magnetic disk devices 8-1, 8-2 into the main memory 10. The host device 1, the backup memory 2, the main memory 10 and the mirror disk device 3 are connected to a common bus 4, respectively. Mirror disk device 3
Includes a DMA controller 5 for controlling the common bus 4, single disk handlers 7-1 and 7-2 for controlling the operations of the magnetic disk devices 8-1 and 8-2, and single disk handlers 7-1 and 7-2. It has a mirror disk handler 6 for controlling the above operation. Magnetic disk device 8-1,
8-2 is mirrored by the mirror disk handler 6, and a program of the OS main body is stored in each of the magnetic disk devices 8-1 and 8-2.

The boot loader 11 has a function of controlling the procedure of system startup and the operation of the mirror disk driver 13. The backup memory 2 is a memory that can be read and written by both the higher-level device 1 and the mirror disk device 3, and mirror disk management information is arranged in the backup memory 2. Mirror disk device 3
Mirrors the magnetic disk devices 8-1, 8-2 based on this mirror disk management information.

Next, the operation will be described. FIG. 8 is a flowchart showing the IPL operation of the boot loader 11 in the computer system. When the system is started up, the higher-level device 1 activates the boot loader 11, and the boot loader 11 issues an instruction to the mirror disk driver 13. The mirror disk driver 13 issues an instruction to the mirror disk handler 6 via the DMA controller 5 and the magnetic disk devices 8-1, 8-2.
Instructing the read operation of the program of the OS main body stored in the OS. The read program is loaded into the main memory 10.

Next, the operation of the mirror disk driver 13 activated by the boot loader 11 will be described with reference to FIG. The mirror disk driver 13, which has received the read request from the boot loader 11, refers to the mirror disk information defined in the backup memory 2, and the single disks constituting the mirror disk are the magnetic disk devices 8-1 and 8-2. Is recognized (step ST
4). Next, the mirror disk driver 13 issues a read request to the single disk handler 7-1 which controls the operation of the magnetic disk device 8-1 (step ST
5). The single disk handler 7-1 receives this request via the DMA control unit 5 and executes a read process for the magnetic disk device 8-1. The execution result is transmitted from the single disk handler 7-1 to the mirror disk driver 13 via the DMA controller 5.

The mirror disk driver 13 determines whether or not the execution result is an error (step ST6), and if the execution result is normal, notifies the boot loader 11 of normal end (step ST11). A read request is issued to the single disk handler 7-2 that controls the disk device 8-2 (step ST
7). The single disk handler 7-2 receives this request via the DMA control unit 5, and executes the read processing again for the magnetic disk device 8-2. The execution result is transmitted from the single disk handler 7-2 to the mirror disk driver 13 via the DMA controller 5. The mirror disk driver 13 determines whether the execution result is an error (step ST8), and if the execution result is normal, notifies the boot loader 11 of normal termination (step ST11). If the execution result is an error, the backup memory 2 is stored. Set the IPL error information to (step ST
9) The boot loader 11 is notified of the abnormal end (step ST10).

As described above, according to the fourth embodiment, the boot loader 11 issues an instruction to the mirror disk driver 13 to set the magnetic disk devices 8-1 and 8-2 in the mirror disk device 3 to the magnetic disk for IPL. Since it is used as a device, the IPL operation can be speeded up. Further, even if one of the magnetic disk devices as the mirror disk causes an operation error, the other magnetic disk device executes the IPL operation, so that the occurrence of the IPL operation error can be reduced. Further, when an operation error occurs during the IPL operation, the operation error information is stored in a predetermined area in the backup memory 2, and the host device 1 sees this operation error information and I
The cause of the PL error can be easily analyzed.

Embodiment 5. FIG. 9 shows the mirror disk driver 13 and the single disk handlers 7-1 and 7-2.
It is a figure which shows the DMA command chain transferred between. The same components as those shown in FIG. 7 are designated by the same reference numerals, and duplicate description will be omitted.

In the computer system according to the fifth embodiment, whether the status of the higher-level device 1 is the IPL operation state or the OS in the command entry 15 which constitutes the DMA command chain for DMA transfer between the higher-level device 1 and the mirror disk device 3. The system status flag 28 indicating whether the operation is being performed is created, a DMA command is created by omitting the DMA command used only for the operation of the OS during the IPL operation, and the DMA transfer is executed to perform the IPL.
The operating time of is shortened.

The DMA command chain is used in the IPL operation after the IPL operation is finished and the OS is started.
It is also used during DMA transfer between the OS and the mirror disk handler 6 and the single disk handlers 7-1 and 7-2. In FIG. 9, 14 is a setup entry for DMA control, 15 is a command entry for disk handler control, 17 is a SCSI command block for mirror disk or single disk control, 19 is an input / output data buffer, 18 is a sense buffer, and 20 is The report entry indicating the end of the DMA command and the sense entry 16 are command entries dedicated to the sense buffer 18.

The command entry 15 stores information for controlling the operation of the mirror disk handler 6 or the single disk handlers 7-1 and 7-2 when the mirror disk driver 13 or the OS requests the execution of the SCSI command. In the sense entry 16, the mirror disk handler 6 or the single disk handler 7-
It stores the SCSI command execution result set by 1 and 7-2. 28 is a system status flag, which is provided only in the command entry 15 immediately after the setup entry 14. When the execution result of the SCSI command results in an error, the single disk handlers 7-1 and 7-2 store the sense data of the magnetic disk devices 8-1 and 8-2 in the sense buffer 18. This sense data is used only by the OS. That is, the sense entry 16 and the sense buffer 18 are DMs used only by the OS.
It is an A command.

Next, the operation will be described. FIG. 10 is a flowchart showing the operation of the single disk handlers 7-1 and 7-2. Mirror disk driver 1 started by boot loader 11 at system startup
3 is an SCS for the magnetic disk devices 8-1, 8-2
In order to request the execution of the I command, the DMA command chain shown in FIG. 9 is transmitted to the single disk handlers 7-1 and 7-2. At this time, the mirror disk driver 13 sets the system status flag 28 of the command entry 15 in the DMA command chain (for example, sets it to a high level) to clearly indicate that the system is in IPL.

Single disk handlers 7-1 and 7-2
Analyzes the DMA command chain received via the DMA control unit 5 (step ST12), and the single disk handlers 7-1 and 7-2 respectively send SCSI commands to the magnetic disk devices 8-1 and 8-2 to be controlled. Execute the SCSI command in block 17 (step ST1
3). When the SCSI command ends in error (step ST14), the single disk handlers 7-1 and 7-2
Is the system status flag 28 in the command entry 15.
Is referred to (step ST15). If it is not during the IPL as a result of the reference, the sense data extracted from the magnetic disk devices 8-1, 8-2 is stored in the sense buffer 18 (step ST16). If retrying (step ST17), the same SCSI command is executed again (step ST13).

When the host device 1 is in the IPL operation and retries (step ST17), the SCSI command is executed again (step ST13). When not retrying, the SCSI command execution result is set in the sense entry 16 in the DMA command chain, and the DM
The A command chain is transmitted to the mirror disk driver 13 or the OS (step ST18). During the IPL operation, since the DMA commands of the sense entry 16 and the sense buffer 18 in the DMA command chain are not used, the boot loader 11 generates a DMA command chain omitting this DMA command, performs DMA transfer, and performs the magnetic disk drive 8-. O stored in 1, 8-2
Boot the S program.

As described above, according to the fifth embodiment, the state of the higher-level device 1 is IPLed in the DMA command chain transferred between the higher-level device 1 and the mirror disk device 3.
System status flag 28 that can be used to determine whether the OS is operating or operating
If the boot loader 11 in the higher-level device 1 is in the IPL operation, the DMA transfer of the DMA command chain excluding the part used for the OS in the DMA command chain can be executed at high speed to shorten the IPL operation time. .

Sixth Embodiment FIG. 11 is a flow chart showing the operation of the computer system including the mirror disk device according to the sixth embodiment of the present invention. The same components as those shown in FIG. 7 are designated by the same reference numerals, and duplicate description will be omitted.

In the computer system according to the sixth embodiment,
When the magnetic disk devices 8-1 and 8-2 in the mirror disk designated by the user as the IPL disk device do not match the mirror disk management information defined in the backup memory 2, the ID designated by the user is used. The magnetic disk device having the same ID is regarded as an IPL disk device to avoid the occurrence of an IPL operation error and reduce the number of occurrences of the IPL operation error, and the information of the IPL disk is stored in the backup memory 2. IPL
It is easy to analyze the cause of the occurrence of the IPL error due to the setting error of the disk device.

Next, the operation will be described. In the mirror disk driver 13 that has received the read request from the boot loader 11, the magnetic disk device 8-1 or 8-2 that is the mirror disk that has received the request is the backup memory 2
It is determined whether or not it is defined as a mirror disk in the mirror disk information defined in (step ST
19 and step ST20). Figure 8 if defined
The processing after step ST7 shown in is executed.

If it is found that the ID is not defined, the same SCSI identification number (I as the magnetic disk unit 8-1 or 8-2 that received the read request from the boot loader 11).
The magnetic disk device having D) is recognized as a single disk (step ST21). Here, for convenience, the single disk is referred to as a magnetic disk device 8-1.

Next, the mirror disk driver 13 issues a read request to the single disk handler 7-1 which controls the magnetic disk device 8-1 (step ST2).
2). The single disk handler 7-1 receives this request via the DMA controller 5 and executes a read operation to the magnetic disk device 8-1. The execution result is transmitted from the single disk handler 7-1 to the mirror disk driver 13 via the DMA controller 5.

The mirror disk driver 13 sets data read from the magnetic disk device 8-1 as IPL disk information (step ST23), and then determines whether the execution result is an error (step ST2).
4). If the execution result is normal, the boot loader 11 is notified of the normal end (step ST25). If the execution result is an error, IPL error information is set in the backup memory 2 (step ST26), and the boot loader 11 ends abnormally. Is notified (step ST27).

As described above, according to the sixth embodiment, the magnetic disk device in the mirror disk designated by the user as the IPL disk device matches the mirror disk management information defined in the backup memory 2. If not, the magnetic disk device having the same ID as the ID specified by the user is regarded as an IPL disk device and I
Since the PL operation is performed, the occurrence of an IPL operation error can be avoided. Moreover, since the information of the IPL disk is stored in the backup memory 2, the cause of the IPL operation error can be easily analyzed, and the user can prevent the occurrence of the IPL error due to the setting error of the IPL disk device.

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

As is evident from the foregoing description, according to the first aspect of the invention, the DMA command in the chain to be DMA transferred between host device and the mirror disk device, the state of the host system I
A system state flag is provided to determine whether PL is in operation or the OS is in operation, and if the host device is in IPL operation, DMA transfer is performed except for the portion of the DMA command used exclusively for the OS in the DMA command chain. So I
This has the effect of shortening the PL operation time.

According to the second aspect of the present invention, when the mirror disk designated by the user as the magnetic disk device for IPL operation is not defined as the mirror disk in the mirror disk management information in the backup memory, the mirror disk The IPL operation is performed from the single disk device having the same SCSI device ID as the disk, and the information of the magnetic disk device in the IPL operation is stored in the backup memory.
It is possible to prevent the occurrence of an IPL operation error due to a setting error. In addition, the execution of the IPL operation from the single magnetic disk device is stored in the backup memory as IPL disk information, and after the IPL operation is completed, the application dependent on the mirror disk refers to the information, so that the application program There is an effect that it is possible to prevent an operation error caused by accessing a mirror disk that does not exist.

[Brief description of drawings]

FIG. 1 is a block diagram showing a computer system including a mirror disk device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a template of a DMA command chain used in a computer system including a mirror disk device.

FIG. 3 is a block diagram showing a computer system including a mirror disk device according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a computer system including a mirror disk device according to a third embodiment of the present invention.

FIG. 5 is a diagram showing operation error information for the number of retries in the sense buffer area of the DMA command chain.

FIG. 6 is a flowchart showing a data write operation to a disc error information area.

FIG. 7 is a block diagram showing a computer system including a mirror disk device according to a fourth embodiment of the present invention.

FIG. 8 is a flowchart showing an IPL operation of a boot loader in the computer system.

FIG. 9 is a diagram showing a DMA command chain transferred between a mirror disk driver and a single disk handler.

FIG. 10 is a flowchart showing the operation of a single disk handler.

FIG. 11 is a flowchart showing the operation of a computer system including a mirror disk device according to the sixth embodiment of the present invention.

FIG. 12 is a block diagram showing a conventional computer system including a mirror disk device.

FIG. 13 is a diagram showing a DMA command used in a conventional computer system including a mirror disk device.

FIG. 14 is a block diagram showing another conventional computer system including a mirror disk device.

[Explanation of symbols]

1 upper device (upper device means), 2 backup memory (memory means), 5 DMA control section (DMA transfer control means), 6 mirror disk handler (mirror disk control means), 7-1, 7-2 single disk handler (single disk control) Means), 8-1, 8-2 magnetic disk device, 11 boot loader (boot loader means).

Continuation of front page (51) Int.Cl. ⁷ Identification code FI G06F 13/12 340 G06F 13/12 340D (56) References JP-A-8-44652 (JP, A) JP-A-6-161931 (JP, A) JP-A-57-150016 (JP, A) JP-A-5-307445 (JP, A) Actual Kaihei 2-108144 (JP, U) (58) Fields investigated (Int.Cl. ⁷ , DB name) ) G06F 3/06-3/08 G06F 13/10-13/14 G06F 13/28

Claims (2)

(57) [Claims] 1. A plurality of magnetic disk devices, An access instruction to the plurality of magnetic disk devices
DMA command chain including multiple SCSI commands
Upper-level device means for generatingTwo magnetic disks of the plurality of magnetic disk devices
Set the device as a mirror disk device and perform mirror control.
Lardisk control means, A plurality of systems for controlling the operation of the plurality of magnetic disk devices.
Toggle disk control means, The DMA command chain with the host device means.
Controlling the DMA transfer operation of the
According to the data inThe mirror disk control means and
AndDM for controlling the operation of the single disk control means
A transfer control means, Memory means accessible from the host device means, One of the mirror disk devices, a magnetic disk device
Performs Initial Program Load (IPL) operation
Set as a magnetic disk device to run and control its operation
In the magnetic disk device that executes the IPL operation.
If an error occurs, the mirror disk device
On the other hand, the other magnetic disk device is magnetic
A boot loader that controls switching with the disk device
Has steps and The single disk control means performs an IPL operation
When an operation error occurs in the magnetic disk device,
Writing error information into the memory means, The computer system in the host device means performs an IPL operation.
If the IPL operation is in progress, it is determined whether the
Indicates that the IPL is in operation in a predetermined area in the chain
Write data to the flag, and further write the DMA command
Only commands needed for IPL operations in the chain
A DMA command chain consisting of
DM of the DMA command chain with the transmission control means
A transfer is executed, and the mirror disk control means and system
The toggle disk control means refers to the flag and moves the IPL.
Do the work, With the DMA transfer operation of one DMA command chain,
Multiple SCSI commands Transfer Total
Computer system. 2. A plurality of magnetic disk devices, An access instruction to the plurality of magnetic disk devices
DMA command chain including multiple SCSI commands
Upper-level device means for generating Two magnetic disks of the plurality of magnetic disk devices
Set the device as a mirror disk device and perform mirror control.
Lardisk control means, A plurality of systems for controlling the operation of the plurality of magnetic disk devices.
Toggle disk control means, The DMA command chain with the host device means.
Controlling the DMA transfer operation of the
The mirror disk control means and the thin disk
DMA transfer control for controlling operation of glue disk control means
Means and, Memory means accessible from the host device means, One of the mirror disk devices, a magnetic disk device
Performs Initial Program Load (IPL) operation
Set as a magnetic disk device to run and control its operation
In the magnetic disk device that executes the IPL operation.
If an error occurs, the mirror disk device
On the other hand, the other magnetic disk device is magnetic
A boot loader that controls switching with the disk device
Has steps and The single disk control means performs an IPL operation
When an operation error occurs in the magnetic disk device,
Writing error information into the memory means, In the host device unit, the boot loader unit is an IPL.
When starting the operation, the magnetic disk that is the target of the IPL operation
Device is defined as the mirror disk device
Check if it is not defined
The magnetism that is the target of the IPL operation with respect to the troder means
Define a disk device as the mirror disk device, With the DMA transfer operation of one DMA command chain,
Transfer multiple SCSI commandsCharacterized byTotal
Computer system.