JP2767661B2 - Disk controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk controller for asynchronously transferring data between a higher-level device and a disk device, and more particularly to a reduction in processing efficiency when selecting a record by means of key data sent from the higher-level device. And a disk control device for preventing a decrease in memory use efficiency and a decrease in recovery capability.

2. Description of the Related Art A disk device used as an external storage device of a computer system is normally controlled by a disk controller, and performs data transfer with a channel.
Some disk controllers asynchronously transfer data between a channel and a disk device via an internally provided memory, and include a count unit, a key unit, and a data unit. When selecting a record to be used, there is a method that uses data of a key part.

When a record is selected using the data of the key portion, if the search key data to be collated with the data of the key portion provided from the channel is not stored in the disk control device, the processing efficiency is reduced. I do.

In the case where the data of the key portion given from the channel is stored in the internal memory, the use efficiency of the internal memory is reduced due to the size of the memory capacity per unit divided by the memory boundary. During operation, the recovery capability is reduced by the disappearance of the search key data stored in the internal memory. However, such a state needs to be prevented.

[0005]

2. Description of the Related Art FIG. 5 is a block diagram for explaining an example of the prior art, and FIG. 6 is a detailed block diagram of each functional unit.

[0006] The disk control device 2 is composed of a plurality of functional units, and the channel adapters 4 and 5 are arranged as shown in FIG.
The processor 13 reads out the program stored in the control memory 14 and operates, receives the channel command word (CCW) from the channel 1 via the interface circuit 11, and operates the data between the channel 1 and the channel 1 through the interface circuit 11. In addition to performing the transfer, the common bus control circuit 12 is controlled to transfer data to and from other functional units via the common bus 10.

The device adapters 6 and 7 are configured as shown in FIG. 6A. The processor 13 reads out a program stored in the control memory 14 and operates, and sends an instruction to the disk device 3 via the interface circuit 11. The common bus control circuit 12 controls the common bus control circuit 12 to perform data transfer with the other functional units via the common bus 10.

The resource manager 8 has a configuration as shown in FIG. 6 (B). The processor 16 operates by reading a program stored in the control memory 15 and controls the common bus control circuit 17 to control the common bus 10. All CCWs from channel 1 are received and stored in RAM 18 via
In addition to centrally managing the CWs, it instructs the channel adapters 4 and 5 and the device adapters 6 and 7 to perform data transfer processing for each CCW.

The received control information for all CCWs, control information for each channel path, and control information for the disk device 3 are stored in the RAM 18, and other functional units are transferred to the RAM 18 via the RAM access control circuit 19. It is possible to access and read out the information, and the information is centrally managed.

For example, if channel 1 is channel adapter 4
To instruct data read, the next CC
W is chained and transmitted. That is, DX (Define extent) that defines a read or write operation, LR (Locate record) that indicates a read or write operation start position, and SrKE (Search key equal) that instructs comparison of key data,
Each CCW of RD (Read data) instructing data reading
And send them out.

FIG. 7 is a diagram for explaining the track format of the disk device, and FIG. 8 is a diagram for explaining the operation of FIG. The track of the disk device 3 is divided into a plurality of records (0) to (5) as shown in FIG. 7 (A), and each record (0) to (5) is divided as shown in FIG. 7 (B). A key in which key data used for indexing a record designated by channel 1 is compared with a count section in which index information for indexing a record is recorded and search key data transmitted by channel 1. And a data section for recording user data.

[0012] As shown in FIG.
The channel adapter 4 of the disk control device 2 that has received the definition information and the definition information reports the normal reception to the channel 1 when it is normally received.

As shown in FIG. 8, when the channel 1 is notified of normal reception, the channel 1 sends the LR and the position information including the cylinder address, the head address, and the record address to the disk control device 2 and the channel adapter. 4 is
Upon successful reception, it reports successful receipt to channel 1.

The channel adapter 4 requests the resource manager 8 to use the common bus 10, and when permitted, sends out the address of the device adapter 6 and connects with the device adapter 6.

Then, the channel adapter 4 sends the received position information to the device adapter 6 to control the disk device 3, and as shown in the position of FIG. 8, a designated record, for example, record (1) Position the head.

That is, when the device adapter 6 controls the disk device 3 to read out the index information of the count part of the record (0) shown in FIG. Since the position information does not match the index information, the index information of the count part of the record (1) is read out next as shown in the read count of FIG.
(1) As shown in the count, the index information of the count part of the record (1) is received, and as shown in the ID comparison of FIG. 8, it is checked whether the record is the designated record. Report a match to.

Record specified from channel adapter 4
Channel 1 reported that (1) was detected is shown in FIG.
The SrKE is sent to the channel adapter 4 of the disk controller 2 as shown in the SrKE of FIG.

Accordingly, the channel adapter 4 instructs the device adapter 6 to control the disk device 3, and
As shown by the read key, the data in the key part of the record (1) is read.

At this time, the search key data matching the data of the key portion of the record (5) from the channel 1, ie, Sr
Assuming that the K data has been transmitted, as shown in the key comparison of FIG. 8, the SrK data transmitted by the channel 1 and the key data read from the key portion of the record (1) indicated by the read (1) key Comparisons do not match. Therefore, as shown in FIG. 8, the discrepancy is reported to the channel 1 from the disk controller 2.

In response to this mismatch report, the channel 1 sends out SrKE as shown by SrKE in FIG. Then, the SrK data is compared with the key data of the record (2), and as shown in the key comparison of FIG. 8, the SrK data transmitted by the channel 1 and the key part of the record (2) indicated by the read (2) key. Are compared, the comparison does not match, and the disc controller 2 reports a mismatch to channel 1 as shown by the mismatch in FIG.

Similarly, for the next records (3) and (4), as shown in FIG. 8, a discordance is reported from the disk controller 2 to channel 1, and in FIG.
The disk controller 2 reports a match to channel 1.

When a match is reported, the channel 1 sends an RD to the channel adapter 4 as shown in FIG. 8, and the channel adapter 4 instructs the device adapter 6 to control the disk device 3 and The user data stored in the data section of (5) is read and stored in the shared memory 9.

When user data is stored in shared memory 9, channel adapter 4 reads the user data and transfers it to channel 1. In the above description, the case where the channel adapter 4 and the device adapter 6 execute data transfer between the channel 1 and the disk device 3 has been described.
The same applies to the case where the processing is executed by the channel adapter 5 and the device adapter 7.

[0024]

As described above, conventionally, after the record (1) specified by the channel 1 is detected, the record in which the data of the key portion that matches the SrK data specified by the channel 1 is stored. Until (5) is detected, SrK is applied between the channel adapter 4 and the channel 1.
Since there is transmission and reception of commands and data such as notification of E and SrK data and mismatch, the connection between the channel 1 and the channel adapter 4 is not released.

Therefore, during this time, the disk controller 2 cannot receive another job (CCW chain),
There is a problem that processing efficiency is reduced. To solve this problem, when SrKE is received, the shared memory 9
The rK data is stored, and the inconsistency between the SrK data and the key data is not notified to the channel 1, and the connection between the channel 1 and the channel adapter 4 or 5 is temporarily established until the SrK data and the key data match. You only need to release it. However, even in this case, the following problems exist.

FIG. 9 is a diagram for explaining the storage area allocation of the shared memory. When the SrK data is stored in the shared memory 9 as described above, it is necessary to secure an area for storing the SrK data for each job.

Therefore, channel 1 and channel adapter 4
Or, when the connection between 5 is once released, for example, when three jobs are received and processed in parallel, for example, as shown in jobs 1 to 9 in FIG. The area shown is secured, and the SrK data is stored.

By the way, in the management of a large-capacity memory such as the shared memory 9, a memory boundary exists at the time of allocating a data storage area, and the data is stored only in units of a certain memory capacity. The storage area cannot be increased or decreased.

In the current disk controller 2, this memory boundary has a large value of 16 Kbytes, and if a storage area for SrK data is secured for each job, for example, from the relationship with the storage area for user data, for example,
As in the storage area of SrK data indicated by K in FIG. 9 or the like, it is necessary to cross one memory boundary and store it in the next memory boundary.

That is, a job requires four unit memory capacities, a job requires five unit memory capacities, and a blank portion shown by cannot be used by another job. There is a problem that the memory area is increased and the memory use efficiency is reduced.

The shared memory 9 is also used as a cache memory. However, an increase in the unused memory area decreases the memory area allocated for the cache memory. There is a problem that reduction also occurs.

When a storage area for SrK data is secured for each job, as shown in FIG. 9, the SrK data exists on the shared memory 9 allocated in the same manner as the user data to be read or written.

By the way, if the disk device 3 which is a target of a certain job is a target of a cache operation, and if the job instructs a read operation, a staging operation is required in the operation of the job. That is, an operation of storing data read from the disk device 3 in the cache area of the shared memory 9 is required.

In this staging operation, in order to enhance the cache effect, the records which have become unmatched in the search operation,
For example, the records (1) to (4) shown in FIG. 7 are also stored in the cache area.
Will be erased before a match with the key data is detected.

That is, at the time shown in FIG. 8, the SrK data is sent out and stored in the shared memory 9. Thereafter, the data of the records (1) to (4) read out is stored in the shared memory 9. You.

Since the cache data needs to match the data read from the disk device 3, the SrK data previously stored in the shared memory 9 may exist in the cache data. Can not.

Therefore, since the SrK data storage area is ignored and the cache area is allocated, the SrK data in the shared
The K data will be erased.

Here, for example, if a failure occurs during the operation for detecting the coincidence between the SrK data and the key data by connecting the device adapter 6 and the disk device 3, the device adapter 7 and the disk device 3 are connected. Thus, a recovery operation for continuing this operation is performed.

However, since the device adapter 7 cannot directly obtain the internal information of the device adapter 6, the Sr stored in the internal register of the device adapter 6 cannot be obtained.
K data cannot be obtained directly.

Therefore, the device adapter 7 performs an operation of reading the SrK data from the shared memory 9. At this time, since the SrK data on the shared memory 9 has been erased as described above, the SrK data and the key data are not read. However, there is a problem that a recovery operation that continues the operation by detecting the coincidence of the data cannot be performed, and the recovery capability is reduced.

The present invention has been made in view of such a problem, and has been developed in consideration of SrK.
E, a plurality of S in the specific area of the shared memory 9.
By centrally storing and managing rK data,
The loss of the SrK data is prevented, the mismatch between the SrK data and the key data is not notified to the channel 1, and the connection between the channel 1 and the channel adapter 4 or 5 is temporarily established until the SrK data and the key data match. By canceling,
An object of the present invention is to prevent a reduction in processing efficiency when selecting a record, a reduction in memory use efficiency, and a reduction in recovery capability.

[0042]

FIG. 1 is a block diagram for explaining the principle of the present invention. The disk control device 21 includes a disk device 3 having a track formatted by a plurality of records each including a count portion, a key portion, and a data portion, and a desired record group based on the index information recorded in the count portion. Is selected, and then, based on the key data recorded by the key unit, intervenes with the higher-level device 20 that sends out a command to select a desired record in the selected record group. Receiving the command via the interface control means 22 coupled with
After detecting the leading record of the desired record group, the desired record is detected, and data transfer between the disk device 3 and the host device 20 is performed.

A storage means 24 for centrally storing a plurality of search key data transmitted by the higher-level device 20 to select the desired record, and a method for detecting the first record of the desired record group When the desired record is detected by using the search key data stored in the storage unit 24, the interface control unit 22 again releases the connection of the interface control unit 22 with the host device 20. Control means 23 for controlling the connection with the device 20 is provided.

After the first record of the desired record group is detected, until the desired record is detected, the interface control means 22 can be connected to the host device 20 at any time. State.

[0045]

With the above configuration, the interface control means 22 selects a record desired by the higher-level device 20 from the record group after the first record of the record group desired by the higher-level device 20 is detected. Until,
Since it is in a free state that can be connected to the host device 20 at any time, the host device 20 can execute other jobs during this time,
The interface control unit 22 can receive and process another job.

Further, the storage means 24 stores a specific area which is not an object of the cache data storage area allocation,
Since the search key data sent by the is centrally stored, it is not erased even by the cache operation.

Further, since it is not necessary to allocate a storage area for search key data for each job, it is possible to effectively use a unit memory area divided by a memory boundary, and to reduce the occurrence of an unused area. Can be prevented.

Therefore, it is possible to prevent a reduction in processing efficiency, a reduction in memory use efficiency, and a reduction in recovery capability.

[0049]

FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention, and FIG. 3 is a diagram for explaining the operation of FIG.

In FIG. 2, the same reference numerals as those in FIGS. 5 and 6 indicate those having the same functions. The resource manager 25 shown in FIG. 2A has a configuration as shown in FIG. 2B, and differs from the resource manager 8 in FIG. 6 in the programs stored in the control memory 28. The difference from the shared memory 9 is that an SrK data area 27 is provided.

Similarly to the above, when instructing the channel adapter 4 of the disk controller 21 to read data, the channel 1, the DX, the LR, the SrKE, and the RD are chained and transmitted.

As shown in FIG.
The channel adapter 4 of the disk controller 21 that has received the definition information and the definition information reports the normal reception to the channel 1 when it is normally received.

As shown in FIG. 3, when the normal reception is reported for the channel 1, the LR and the position information including the cylinder address, the head address, and the record address are sent to the disk controller 21 and the channel adapter is sent. 4
Reports successful receipt to channel 1 upon successful receipt.

Then, the channel adapter 4 requests the resource manager 25 to use the common bus 10, and when permitted, sends out the address of the device adapter 6 and connects with the device adapter 6.

Then, the channel adapter 4 sends the received position information to the device adapter 6 to control the disk device 3, and as shown in the positioning of FIG. 3, a designated record, for example, FIG. Position the head on record (1) shown in.

That is, when the device adapter 6 controls the disk device 3 to read out the index information of the count part of the record (0) shown in FIG. Since the position information does not match the index information, the index information of the count part of the record (1) is read next as shown in the read count of FIG.
(1) As shown in the count, the index information of the count part of the record (1) is received, and as shown in the ID comparison of FIG. 3, it is checked whether the record is the designated record. Report a match to.

Record specified from channel adapter 4
Channel 1 reported that (1) was detected is shown in FIG.
The SrKE is sent to the channel adapter 4 of the disk controller 21 as shown in the SrKE of FIG.

When the channel adapter 4 receives the SrK data transmitted following the SrKE as shown in the SrK data of FIG. 3 according to the instruction of the resource manager 25, the channel adapter 4 transmits the SrK data to the SrK data of the shared memory 26.
When the data is stored in the data area 27, the connection with the channel 1 is temporarily released as shown in the release of the connection in FIG.

The resource manager 25 has a configuration as shown in FIG. 2B, and the processor 16 reads and operates the program stored in the control storage 28, and sends the above instruction to the channel adapter 4.

Next, the resource manager 25 instructs the device adapter 6 to control the disk device 3 as shown by the read key in FIG. 3 and to switch from the disk device 3 to the read (1) key in FIG. As shown, the data of the key part of the record (1) is read.

Assuming that search key data matching the data of the key portion of the record (5), that is, SrK data, is transmitted from channel 1, channel 1 transmits data as shown in the key comparison of FIG. The comparison between the SrK data and the key data read from the key portion of the record (1) indicated by the read (1) key does not match.

Since the disk controller 21 has released the connection with the channel 1, the disk controller 21 does not report the mismatch detected by the device adapter 6 to the channel 1. The device adapter 6 receives an instruction from the resource manager 25,
Then, as shown by a read key in FIG. 3, the disk device 3 is instructed to read data of a key portion of the record (2) from the disk device 3 as shown by a read (2) key in FIG. As shown in the comparison, the data is compared with the SrK data stored in the SrK data area 27.

In this case as well, the comparison is inconsistent, but the disk control unit 21 does not report the inconsistency in the channel 1 and the device adapter 6 responds to the instruction from the resource manager 25 as shown by the read key in FIG. , And instructs the disk device 3 to read the data of the key portion of the record (2) from the disk device 3 as shown by the read (2) key in FIG.
7 is compared with the SrK data stored in.

In this case as well, the comparison results in a mismatch, but the disk controller 21 does not report the mismatch to the channel 1, and the device adapter 6 subsequently instructs the disk device 3 as shown by the read key in FIG. And lead (3) in FIG.
As shown by the key, the data of the key part of the record (3) is read out from the disk device 3, and the SrK
This is compared with the SrK data stored in the data area 27.

In this case as well, the comparison results in a mismatch, but the disk controller 21 does not report the mismatch to the channel 1 and the device adapter 6 subsequently instructs the disk device 3 as shown by the read key in FIG. And lead (4) in FIG.
As shown by the key, the data of the key portion of the record (4) is read from the disk device 3, and the SrK
This is compared with the SrK data stored in the data area 27.

In this case as well, the comparison results in a mismatch, but the disk controller 21 does not report the mismatch to the channel 1, and the device adapter 6 subsequently instructs the disk device 3 as shown by the read key in FIG. And lead (5) in FIG.
As shown by the key, the data of the key portion of the record (5) is read from the disk device 3, and the SrK
This is compared with the SrK data stored in the data area 27.

In this case, since the comparisons match, the channel adapter 4 of the disk controller 21 rejoins with the channel 1 based on the instruction of the resource manager 25, and
Notify that the K data matches the key data.

When a match is reported to channel 1, in FIG. 3, channel 1 sends an RD to channel adapter 4, and channel adapter 4 instructs device adapter 6 to control disk device 3, record
The user data stored in the data section of (5) is read and stored in the shared memory 9.

When user data is stored in shared memory 9, channel adapter 4 reads the user data and transfers it to channel 1. FIG. 4 is a diagram for explaining storage area allocation of the shared memory according to the present invention.

FIG. 4 shows a case where an SrK data area 27 for storing SrK data in a centralized manner in the shared memory 26 is provided. As shown in FIG. 9K, an area for storing SrK data is allocated for each job. In order to perform centralized management, a unit memory area divided by a memory boundary is allocated, and as shown in the SrK data area of FIG.
Each SrK data of the job is stored in this unit memory area.

Therefore, unlike the case of FIG. 9, the blank memory area indicated by the job is prevented from becoming an unused area. Further, by defining the SrK data area in FIG. 4 so as not to be a target area of the cache data storage area allocation, the SrK data is not erased by the cache operation.

[0072]

As described above, according to the present invention, after the channel adapter detects the first record of the desired record group of the channel, the time between when the channel in the record group selects the desired record, However, since the channel is in a free state that can be combined with the channel, the channel can perform another job during this time, and the channel adapter can receive and process another job.

In the shared memory, search key data transmitted by the channel is intensively stored in a specific area which is not a cache data storage area allocation target, so that it is not erased by the cache operation.

Further, since it is not necessary to allocate a search key data storage area for each job, it is possible to effectively use a unit memory area divided by a memory boundary, and to reduce the occurrence of an unused area. Can be prevented.

Therefore, it is possible to prevent a reduction in processing efficiency, a reduction in memory use efficiency, and a reduction in recovery capability.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating the principle of the present invention.

FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention.

FIG. 3 is a view for explaining the operation of FIG. 2;

FIG. 4 is a view for explaining storage area allocation of a shared memory according to the present invention;

FIG. 5 is a block diagram illustrating an example of a conventional technique.

FIG. 6 is a detailed block diagram of each functional unit.

FIG. 7 is a diagram illustrating a track format of a disk device.

FIG. 8 is a view for explaining the operation of FIG. 5;

FIG. 9 is a view for explaining storage area allocation of a shared memory;

[Explanation of symbols]

 1 channel 2, 21 disk controller 3 disk device 4, 5 channel adapter 6, 7 device adapter 8, 25 resource manager 9, 26 shared memory 10 bus 11, interface circuit 12, 17 common bus control circuit 13, 16 processor 14, 15, 28 Control memory 18 RAM 19 RAM access control circuit 20 Host device 22 Interface control means 23 Control means 24 Storage means 27 SrK data area

Claims (1)

(57) [Claims] 1. A disk device (3) having a track formatted by a plurality of records comprising a count section, a key section and a data section, and a desired record based on index information recorded in the count section. After selecting a group, the key unit is interposed between a higher-level device (20) that sends out a command to select a desired record in the selected record group based on the key data recorded by the key unit. Host device
The command is received via the interface control means (22) coupled to (20), and after detecting the first record of the desired record group, the desired record is detected and the disk device is detected.
A disk controller (21) for performing data transfer between (3) and a higher-level device (20), wherein the higher-level device (2) is used to select the desired record.
(0) and a storage means (24) for centrally storing a plurality of search key data to be transmitted, and when the first record of the desired record group is detected,
The connection between the interface control means (22) and the higher-level device (20) is released, and when the desired record is detected by using the search key data stored in the storage means (24), again. The interface control means (22) is
(20) and control means (23) for performing control so as to combine with the interface control means after the head record of the desired record group is detected until the desired record is detected. A disk control device wherein (22) is in a free state.