JPS6198437A - Input/output control system - Google Patents

Abstract

PURPOSE:To improve the data transfer efficiency without stopping the drive of a magnetic tape despite the generation of an error and by attaining the retrial just with the control of a magnetic tape controller. CONSTITUTION:A certain area of a main memory contains the 1st table which stores effective table length, the next table address, the retry offset value obtained in an error mode, etc. and a dummy table which stores the designation of propriety for use of the table, the designation of propriety for reference to the next table, the number of blocks which are always set at '0', etc. While a magnetic tape controller 107 contains a secondary table which stores a copy of the table stored in the main memory. A central processor 1 writes the contents necessary for said table every time the data on plural blocks are read out of a magnetic disk device 5 to a data buffer 202. Then a dummy table is used to attain the chained tables, and the data are transferred to a magnetic tape device 3 from the device 5. Thus the retrial is possible just with the control of the controller 107 despite the generation of an error.

Description

Detailed Description of the Invention (Industrial Application Field) The invention is an input/output control method, in particular, operating a microprogram stored in a magnetic tape control device according to instructions output from a central processing unit, The present invention relates to an input/output control method for controlling a logical copy operation of files between a magnetic disk device and a magnetic tape device having a recording mechanism via a main memory.

(Prior art) In general, the input/output control method of microprogram control is
It operates in response to commands output from a central processing unit (hereinafter referred to as CPU). These commands include write commands and read commands. Conventionally, data transfer between an input/output device (hereinafter referred to as an I10 device) and a CPU is performed in 54 rows.
When a command is issued to the input/output control device, the input/output control device decodes this command, sends the necessary command to the device, and performs direct memory access (hereinafter referred to as DMA) data transfer. Performs control and transfers data. When the data transfer is completed, the input/output control device notifies the CPU of the completion of the operation through an interrupt.

Next, a more detailed explanation will be given with reference to the drawings.

FIG. 1 is a seven-system configuration diagram of a data processing system commonly applied to the present invention and a conventional input/output control system. Figure 2 shows the operation of reading data (files) from a magnetic disk unit (hereinafter referred to as DKU) and writing to a magnetic tape unit with streaming function (hereinafter referred to as MTU) in a conventional input/output control method. FIG.

A conventional logical copy of a file between a DKU and an MTU will be described with reference to FIGS. 1 and 2, taking as an example a copy of a file from a DKU to an MTU.

Read command from CPU1 to DKU5 is output
! Then, the magnetic disk controller (hereinafter referred to as DKC) 4 sends a seek command to the DKU 5 to move the head to a position where necessary information (file) is stored. When looking at the beginning of the seek report (file) in DKU5, the information is transferred to DKC4. DKC
4 uses DMA to CP the information sent from DKU5.
Transfer to main memory (not shown) within the UI. When the DKC4 finishes transferring the necessary information, it generates an interrupt to the CPU to notify the end of the read operation.

When the CPU 1 receives the notification of completion of the read operation of the DKUs, it next executes a write command to the MTU 3.

The magnetic tape control device (hereinafter referred to as MTC) 2 starts feeding the magnetic tape, receives information from the MTC 2, and writes the information. When MTC2 finishes transferring the necessary information, the CPU
1, an interrupt is generated to notify the end of the write operation. CPU 1 repeats the above series of operations as many times as necessary.

By the way, in general, D used in data processing systems
As is well known, the configuration of the KU has a plurality of disk plates and a plurality of disks, as shown in FIGS. 3(a) and (b). is divided into multiple circumferences (hereinafter referred to as tracks), and
The same track is divided into multiple sectors. The data handled by the software is shown in Figure 3 (c>, (d)).
, as shown in (e), the minimum recording unit is a record, a set of multiple CIs of records is treated as a block, and a set of multiple blocks is treated as a file. As shown in Figures 4 and 4, they are physically recorded in consecutive or discontinuous positions (
logically followed). Also, in general, the average seek time of DKU5 is 30 tn 3.
U 3 ';4'', when accessing, it takes about 500 seconds on average to run the magnetic tape for 1 no.
Bit m5.

(Problem to be solved by the invention) In such a conventional method, from DKU to Δ■MT
For logical copying of a file to U, seek operation,
1) Since DMA data transfer from KU to CPU, interrupt processing, magnetic tape running, DMA data transfer from CPU to MTU and interrupt processing are repeated in this order, magnetic tape running is There are problems in that it requires a lot of time and the load on the CPU becomes very heavy due to interrupt processing and the like.

In addition, in an MTU with a streaming function, it is necessary to start data transfer of the next block while passing through the IRQ (gap between blocks) of the magnetic tape, and if this start is delayed, the magnetic tape will stop running. M
prepared a buffer in the main memory to enable asynchronous data transfer, but if the seek time of DKU is long, data transfer will still be delayed, so this is not a sufficient countermeasure.
However, in the data processing system, DKU
Since it is often used as a logically structured area, logical copying of files takes time and also requires C
There is a problem that the load on the PU becomes heavy.Therefore, the purpose of the present invention is to reduce the load on the CPU and to provide a streaming function for logically copying files between the DKU and MTU via the main memory. Our goal is to provide a human output control method that can be executed at high speed without any loss.

(Defined Means for Solving the Problems) For this purpose, the system of the present invention provides a system for controlling logical copying operations via the main memory between a magnetic tape device having a streaming function and a magnetic disk device. In the output control method, the main memory includes a plurality of data buffers each storing one block of transfer data, each buffer storing an effective table length, table usage specification, next table reference permission specification, number of blocks, At least one table that stores the current block number, table length of the secondary table, block information, next table address, and retry offset value in the event of an error; effective table length; specification of whether the table can be used; specification of whether the next table can be referenced; , a dummy table for storing the number of blocks of % Q II, the next table address, and the table length of the next table are provided, and the magnetic tape control device is provided with a dummy table for storing a copy of one of the tables in the main memory. A central processing unit stores the contents in the table each time a plurality of blocks of data are read from the magnetic disk device into the data buffer, and chains the table using the dummy table as necessary. read from the magnetic disk device and issue a write command to the magnetic tape controller, and the magnetic tape controller responds to the write command by writing the contents of the table in the main memory into the sub table. By writing the contents of the data buffer to the magnetic tape device based on the contents of the writing sub-cheagle, it is possible to retry without stopping the running of the magnetic tape and only by controlling the magnetic tape control device when an error occurs. It is characterized by being made possible.

(Example) Next, the present invention will be explained in detail with reference to the drawings.

An embodiment of the present invention is shown in FIGS. 1 and 5. FIG.

In Figure 1, 1 is CPU, 2 is MTC, 3 is MTU, 4
is DKC, 5 is DKU, 101 is address register, 1
02 is a data register, 103 is a bus control circuit, 104
is a microprocessor.

105 is a ROM, 106 is a RAM, and 107 is a device control circuit 7, respectively. In FIG. 5, 201 is an address counter, 202 is a data buffer, 203 is a data counter, 204 is a first-in first-out control buffer (hereinafter referred to as FIFO), and 205 is a sub-control circuit.

FIG. 6 shows the relationship between n tables expanded in the main memory on the CPUI, and FIG. 7 is a diagram showing the organization of this table. In FIG. 7, the SPM tamper area pointer indicates the location of the RAM 106 at the time of error.
Indicates the start address of main memory when transferring to own memory, effective table length indicates the size of the table itself, control information area indicates the control status of the table itself, and number of blocks indicates the number of blocks included in the table. , the offset value indicates the number of operations to be skipped from block 1, the current block number indicates the block number that MTC2 is processing, and the next table address indicates the next table to be referenced when processing of one table is completed. 7 address, the table length of the next table indicates the size of the table to be referenced next, the buffer address indicates the start address of main memory to be accessed in that block, and the range indicates the size of the table to be accessed in that block. The remaining range indicates the remaining range at the end of the operation of the block, and the status stores the status at the end of the operation of the block.

FIG. 8 is a diagram showing an example of details of the control information area in the table, and FIG. 8g9 is a diagram showing an example of details of the control information area in the table.
Indicates the block information of the table written to the FIFO 204 by 04.

In FIG. 8, the reference symbol vTb-1 indicates that this table can be used, and the reference symbol TM indicates that it cannot be used. When the reference symbol TM is -1, the table chain ends with this table. When the reference symbol CP is 1〃, it indicates that this table has been processed normally, and when the reference symbol ER is 11〃, it indicates that an error occurred while processing this table, When the reference symbol IT is SS I II, it indicates that the value set in this table is inappropriate.

The operation of logically copying files from to MTU3 will be explained.

First, CPU1 selects the file you want to copy from DKUs.
! are transferred to main memory in sequential order. After reading an appropriate number of blocks from the DKU 5, a table for activating the MTU is generated on the main memory. When the table generation is completed, the VT flag in the control information area of this table is sectored to 11, and the start address of the table (XXX in FIG. 6) is designated to the MTC 2 to start it.

When the MTC 2 receives the starting address XXX of the table, the microprocessor 104 inputs the starting address XX.
Takes the table specified by X from main memory and stores it in RAM1.
Store in 06. Next, after confirming that the VT flag in this table is 1, the block number S% 1 is written in the current block number area of the table in RAM 106 and the current block number area of the table in main memory.
/f, but if an offset value is specified in the table at this time, the block number will be the offset value.

Next, the microprocessor 104 stores the block information shown in FIG. 9 in the FIFO 204 and gives an instruction to the control circuit 205 to start the transfer. The control circuit 205 is an FI
Take out the buffer address and range from FO204,
address counter 201 and data counter 2, respectively.
03 and sends a write command to MITJ3.

The MTU 3 starts running the magnetic tape, and when it becomes ready to receive data, makes a data request to the control circuit 205. When the control @ circuit 205 receives this data request,
With the assistance of the address register 101 that holds the address from the address counter 201 and the path control circuit 103, a path from main memory to data register 102 to data buffer 202 to control circuit 205 to MTU 3 is created to perform take transfer. conduct. At this time, address counter 2
01 is incremented by 1 every time 1 byte is transferred, and the data counter 203 is incremented by 1.

When the data counter 203 reaches Th Q //, the same 1
The wholesale circuit 205 generates an interrupt for the end of one block transfer to the microprocessor 104, and the data counter 203 indicates N.
Even if O, if an error does not occur, interrupt the microprogram. ) If there is no error, take out the back address and range of the next block from FLF 0204, set them in the address counter 201 and data counter 203, and start transferring the next block. If the error is M, do not perform.

On the other hand, when the microprocessor 104 receives the block end interrupt h, it receives the end status from the control circuit 205 and writes it into the block status area of the table in the main memory. Further, if there is no error in the status, the current block number area of the table in the RAM 106 and main memory is incremented by one. If there is an error in the status, the current block number will not be incremented by 1. Control circuit 20
5 repeats the above operation until an error occurs or all blocks are transferred.

After writing one block, MTU3 writes iRG, which is the gap between plotters, so this IRQ
If the next write command is received while the magnetic tape is being written, the magnetic tape will continue to run without stopping, and the data will be written. When the microprocessor y 104 finishes processing all blocks of the table, it sets the CP flag in the control information area of the table in the main memory (
If an error occurs, set the ER flag or the ER flag and the IT flag. ) cPuiVc interrupt.

Next, referring to the next table address area of the RAM 106 and the table length of the arrow table, the second table is retrieved from the main memory and stored in the RA8 106 (if an error occurs, or the control information area (If the TM flag is set to 1, the missing table is not referenced.) The operation of the intermediary is repeated.

On the other hand, when the CPUI interrupts 8 (Te3), it checks the ER flag and CP flag in the control information area of the existing table, and when the CP flag is set to -1 at BR; y lag -10'', the existing table is updated ( When the ER flag = 11, set the current block number of the existing table to the offset value of the existing table, set the current block number, ER7 ring, and IT flag to %O1/, and delete the existing table. Retry processing can be performed by simply instructing the MTC2 to the start address on the main memory and starting it.

When an error occurs, MTC2kt issues a two-block back command to MTU3 from the block where the error occurred, then issues a one-block forward command to position the magnetic tape at the beginning of the error block, and then
Wait for command from I. In addition, the CPU can determine the processing status of the table by looking at the current block number area of the table, so if necessary, it will chain the read speed of the file from DKU5 and the speed of write to MTU3 from then on. This is possible by changing the table size and number of tables. What has been explained above is based on FIG. 10, which is a conceptual diagram of the logical copy operation of files from DKU5 to MTU3, and
FIG. 11 is a flowchart showing an example of processing between MTU3, and FIG. 1 is a flowchart showing an example of processing within MTCZ.
The understanding will be further deepened by referring to Figure 2.

Now, if a table chain cannot be directly connected temporarily due to DKU5's seek time wait or the processing overhead of other devices in the system, a dummy table whose area is reserved in advance on the main memory is used. Ru. 13th
The figure shows the contents to be stored in the dummy table, and the 14th
The figure shows how tables can be indirectly chained even in the case described above, rather than using a dummy table.

First, as shown in FIG.
I want to create a and start write to MTC2, but
If the size of table 2 cannot be determined, table 1
It is not possible to set values in the whisper next table address and table length areas of the next table. In this case, set the table address and table length of the dummy table to the respective X-th table address of table 1.
and set it in the table length area of the Nth table,
The VT flag in the control information area of the dummy table is 10.
and start up MTC2. Then, when the size of table 2 is determined while MTC2 is processing table 1,
Write the table address and table length of table 2 in the dummy table's next table address and table length of the next table, and write the table address and table length of table 2 in the VT of the control information area.
Set the flag to negative 1. The number of blocks in the dummy table is always SSO1/. This results in table 1
and the chains in Table 2 are removed without impairing the streaming functionality of MTU3.

(Effects of the Invention) According to the present invention, by employing the simple structure described above, it is possible to reduce the number of interrupts to the CPU by the number of blocks set in the table, thereby reducing the load on the CPU and , it is possible to perform logical copying of files at high speed, and also to use magnetic tape running in cases where table chaining cannot be done directly due to waiting DKU seek time or processing overhead of other devices in the system. (This has the effect of making it possible to maximize the functionality of the MTU, which has a streaming function.)

[Brief explanation of the drawing]

Fig. 1 is a system diagram common to the prior art and the present invention, and Fig. 2 is a diagram showing the ethical copying of files in the prior art. Cpl Intent, 3rd U(a)(b) FL D
K U (1) Memorial diagram of head, track, sector, funeral: Figure 3 (C), d) <e> is the 4th composition of Qua f 11
Figure 34 and f) A diagram showing logical copying to IT from KU, and the fifth factor is a diagram showing an embodiment of the present invention, Figures 6, 17, 8, and 13. The figure is a configuration diagram of a table and a dummy table in an embodiment of the present invention.
The figure is a conceptual diagram of block information stored in flF'0, and FIGS. 10, 11, 12, and 14 are conceptual diagrams of the operation of an embodiment of the present invention. l...CPU, 2...MTC, 3...
...Δ4TU. 4...DKC, 5...1) KU, 10
1...Ato Vs register, 102...
Data register, 1o3, pa... bus system? i1 time L
l 04...Microprocessor, 105.
...ROM, 106...RAM. 107...Device control circuit, 201...
... Address counter, 202 ... Data backup, 203 1 ... Data counter, 204 ... FIFo. 205...Control circuit. Agent Patent Attorney Uchihara On No.S Figure 7 Figure 7 Figure 8 Figure 77 Fuho I3 times

Claims (1)

[Claims] In an input/output control method for controlling a logical copying operation via a main memory performed between a magnetic tape device and a magnetic disk device having a streaming function, each of the main memories includes: Multiple data buffers that store one block of transfer data, each with an effective table length, table usability specification,
At least one table that stores the next table referenceability specification, number of blocks, current block number, table length of the next table, block information, next table address, retry offset value in case of error, effective table length, and table usability. A dummy table is provided for storing the specification, whether or not the next table can be referenced, the number of blocks that are always "0", the next table address, and the table length of the next table. A sub-table is provided for storing one copy of the dummy data, and each time the central processing unit reads a plurality of blocks of data from the magnetic disk device to the data buffer, the central processing unit writes the content to the table and stores one copy of the dummy data as necessary. The table is chained using a table to continue reading from the magnetic disk device and a write command is issued to the magnetic tape control device, and the magnetic tape control device responds to the write command to write the main memory. By writing the contents of the table to the sub-table and writing the contents of the data buffer to the magnetic tape device based on the contents of the sub-table, the magnetic tape can be stored without stopping the running of the magnetic tape and when an error occurs. An input/output control method characterized by being able to perform retrials only by controlling the control device.