JPH0786842B2 - Block write-back processor - Google Patents

Description

Description: TECHNICAL FIELD OF THE INVENTION The present invention relates to a block writing for writing back a data block read from a direct access storage device (hereinafter referred to as DASD) to an original position in a data set updating process or the like. With respect to a return processing device, an efficient block write back processing device is provided, in which a DASD address of a read data block is held in the data block and can be used at the time of write back.

[Background of technology]

Conventionally, a certain block is read from the data set in the DASD volume and updated, and the updated block is written back to the original position.
E) is a very ordinary process, but in this case, it is necessary to hold the position after reading the block.

If there is no access to the data set between the read process and the re-write process, the block address is held by the control program, but once the access process to the data set occurs, the block address is Not retained. In this case, the following macro instruction sequence is executed.

READ NOTE POINT WRITE FIG. 1 is a conceptual diagram of the processing by the above macro instruction sequence, where 1 is main storage area and 2 is DASD.

The READ macro reads a block from DASD to main memory.
The READ macro reads from the top block of DASD by sequential processing. Since reading is done in order, R
The read position after the EAD macro is issued is the next block position. To specify the READ position with a relative address,
Just before the READ macro for that purpose, the relative address is pointed (specified) using the POINT macro. The WRITE macro also writes in a sequential process, so when specifying a relative address, specify it with the POINT macro immediately before the WRITE macro.
The NOTE macro is a macro that holds the address of the read block executed immediately before in main memory.

Next, the writing process after reading will be described along the macro instruction sequence.

The READ macro reads a block from DASD to the main memory, and the NOTE macro reads a read operation by the READ macro or WRI.
After confirming the end of the write operation by the TE macro, the relative address in the data set of the block is held in the main memory. The POINT macro points (specifies) the held relative address, and performs a WRITE operation or R from that position.
Request the EAD operation from the control program. Then, the WRITE macro executes write-back from the relative address pointed to.

For the relative address, the relative block address indicating the block position from the start position of the data set, and the TTR indicated by combining the relative track position from the start track of the data set and the block (record) position in it
Format of relative track address etc. Hereinafter, these will be represented by relative block addresses.

As described above, the NOTE macro needs to be executed in order to fetch and store the relative block address in the data set used at the time of READ every time the write-back processing is performed, which has a drawback that processing efficiency deteriorates.

Furthermore, for example, in the case of compiler processing, processing is performed by overlapping about 10 to 30 paths for multiple processing phases, but the generated intermediate text is initially processed sequentially. However, it is necessary to read blocks at a plurality of random positions, and after processing, write those blocks back to their original positions. Often becomes. However, in such a case, the NOTE macro described above can hold only the relative block address of the last read block. This is because the holding of the relative block address is limited to a fixed area called a special "register 1". Because of this, other inefficient measures have to be taken, leading to larger compiler sizes and
There is a problem that the processing time becomes long.

[Object and Structure of Invention]

An object of the present invention is to provide a means for simplifying the instruction processing for holding the write-back position in the write-back processing of a block to a data set, and further for executing it in multiplex for arbitrary plural blocks. Therefore, the present invention allows the block itself to hold the address information, and as a configuration for realizing it, the block read from the data set on the DASD volume is written back in the write-back processing device of each block. A block address table that records an address on a data set corresponding to a block, a block address processing unit that registers the address that writes the block in the block address table when writing the block to the data set, and an arbitrary block from the data set When read, refer to the address table above A read processing unit that embeds an address on the data set corresponding to the block in the block, and a write back processing unit that writes back the block by designating a write back destination to the data set using the address embedded in the block It is characterized by having and.

Example of Invention

 The details of the present invention will be described below with reference to examples.

According to the present invention, in order to retain its own relative block address information in each block of the data set, the first step of acquiring the relative track address of the block in the writing process when creating the data set and reading the block from the data set. Preprocessing is performed by embedding the previously acquired relative block address in the block, that is, by updating the block, and by writing the updated block back to the data set.

FIGS. 2 (a), (b), and (c) are example diagrams for explaining the processing of each of the above steps. In the figure, 1 is a main storage area, 2 is DASD, 3 is a block, 4 is a block address table, 5 is a data set creation processing section, and 6 is a block address processing section. Also, A, B, and C represent block names.

Block addresses in DASD volumes are represented in TTR format. TT is the relative track number and R is the block number within the track.

FIG. 2A shows the first block address acquisition processing. The block A including the code created by the data set creation processing unit is written to DASD2 by the WRITE macro. Since the block A is the first writing, it is written from the beginning of DASD2 and then sequentially. After this, the block address processing unit issues a NOTE macro to obtain the relative block address. Since block A is the first block, the relative block address is TTR-000. The acquired relative block address “000” is registered in the block address table in association with the block name “A”.

With respect to the other blocks B, C, ... Since the blocks B, C, ... Are sequentially written after the block A, the relative block address TTR = 001,002. This gives the relative block address for each block in the dataset,
It is registered in the block address table 4.

With respect to the other blocks B, C, ... As a result, the relative block address of each block in the data set is obtained and registered in the block address table 4.

FIG. 2B shows an address embedding process in the block. That is, the block address processing unit 6 uses the POINT macro to locate the first block A, and then sequentially performs RE.
Update process of reading blocks A, B, C ... of data set using AD macro, then referring to block address table 4 to extract block address information corresponding to each block and embedding the block address information in a predetermined position of the block. Do.

FIG. 2 (c) shows the process of writing back the block with the embedded block address and updated to DASD. That is, the block address processing unit 6 sequentially writes back the blocks A, B, C ... In which the block addresses are embedded to the original positions of the DASD.

As described above, DASD stores only the block holding its own block address, and when any block is read for update processing and then written back, the block address is immediately set. It becomes possible to specify.

〔The invention's effect〕

From the above, the idea is to use the NOTE when updating the data set.
Since it is not necessary to use macros, the processing procedure can be simplified and the processing time can be shortened.

[Brief description of drawings]

FIG. 1 is a conceptual diagram of a conventional writeback processing apparatus, and FIGS. 2A to 2C are explanatory diagrams showing a sequential operation of an embodiment of a writeback processing method according to the present invention. In the figure, 1 is a main storage area, 2 is a DASD, 3 is a block, 4 is a block address table, 5 is a data set creation processing section, and 6 is a block address processing section.

Claims (1)

[Claims] 1. A block address table for recording an address on a data set corresponding to each block in a block write-back processing device for writing back a block read from a data set on a DASD volume, and a block to the data set. A block address processing unit for registering an address at which a block is written in the block address table at the time of writing, and when an arbitrary block is read from a data set, the address table is referred to and an address on a data set corresponding to the block is set in the block. A write-back processing unit that embeds a read processing unit, and a write-back processing unit that writes back the block by designating a write-back destination to a data set using an address embedded in the block Processing equipment.