JPH0115088B2 - - Google Patents

Description

Detailed Description of the Invention [Technical Field of the Invention] The present invention relates to an input/output system in which input/output devices are connected to a plurality of paths including an input/output control device to an access source such as a processor. This invention relates to a retry control method in which an input/output device performs a retry using a vacant path.

[Technology background]

Input/output devices are connected in order for the processor to perform desired processing. Among input/output devices, especially those requiring mechanical operation such as magnetic disk devices, an input/output control device is provided to control the input/output device in the channel of the processor, and the input/output device is controlled via the input/output control device. ing.

In this way, input/output control devices and input/output devices are connected to form an input/output system, but if multiple input/output devices are connected, one input/output device is connected to the input/output control device. While combining with
An I/O controller cannot access other I/O devices. Therefore, even during mechanical operations of input/output devices that do not require control by the input/output control device, the input/output control device is occupied, making it impossible to access other input/output devices.
In the case of a DASD (Direct Access Storage Device), the processor is kept waiting for a long time. For this reason, input/output systems having an input/output device open control function and a dynamic path selection function have been developed.

[Prior art and problems]

In such an input/output system having a dynamic path selection function, each input/output device is connected to each of a plurality of paths including the input/output control device to the processor that is the access source, and any one of the paths is selected. ,
It is configured so that it can be connected to the input/output control device of that path. After one input/output control device connects and starts up the input/output devices, the input/output device is disconnected to perform mechanical operation, and the input/output control device enables the other input/output devices to start up.
Furthermore, after the input/output device completes its mechanical operation, it utilizes a vacant path to reconnect with the input/output control device on that path to exchange data. In this way, after activation, the I/O control device of a free path is reconnected, so the I/O control device of the activated path may be reconnected to the I/O control device of another path, and for this reason, when disconnected, Path control information must be stored in memory.

On the other hand, if a data error or operation error is detected while an input/output device and an input/output control device are being connected or recombined and data is being transferred between them, a retry is performed. It will be done. This retry occurs during recombination, so the I/O device uses the same path to retry with the I/O controller, so if there is a failure in that path itself, such as a cable failure or an I/O In the case of a failure in the internal circuit of the control device, there is a problem in that even if multiple retries are made, a retryout occurs, resulting in useless retries on that path.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide a retry control method for an input/output system that allows recovery without unnecessary retries even in the case of path-specific failures.

[Structure of the invention]

To achieve the above object, the present invention provides a plurality of paths connected to a common access source and each including an input/output control device, an input/output device connected to each of the plurality of paths, and path control information. After the input/output device is connected to one path and activated, path control information is written to the memory, the occupation of the path is released, and the input/output device is ready. In an input/output system that refers to path control information in the memory and reconnects to a vacant path after completion of data transfer, the input/output control device detects an error during the connection or reconnection. When the I/O device issues a join request for a retry, the I/O control device releases the occupation of the joined or recombined path, and stores retry information in the memory, so that when the I/O device issues a join request for a retry, the The input/output device is characterized in that the input/output control device performs retry control of the input/output device based on retry information in the memory.

[Embodiments of the invention]

Hereinafter, the present invention will be explained in detail with reference to Examples.

FIG. 1 is a block diagram of an input/output system with dynamic path selection according to the present invention, showing the DASD subsystem.

In the figure, 1 is a processor (access source),
It accesses the input/output system to process data, and has two channel sections 1a and 1b.DKC is a magnetic disk (input/output)
2 and 4 are control units (hereinafter referred to as directors) that constitute the control unit of the magnetic disk control unit and are connected to the channel units 1a and 1b of the processor 1, respectively. A device that receives access requests from units 1a and 1b, performs desired processing, and exchanges data. Reference numerals 3 and 5 denote adapters, each of which is composed of an adapter control section 30, 50 and a memory 31, 51.
4, and selectively sends it to the memory 31, 51 or a magnetic disk to be described later, and performs serial-to-parallel conversion; the memory 31, 5
1 stores path control information.

60, 61...6n are magnetic disks (devices), which are respectively connected to adapters 3 and 5 via cables to be described later.
and 7 and 8 are first and second cables, which connect each magnetic disk 60, 61...6n to 2
Control lines 9a and 9b are connected to each of the two magnetic disk control devices DKC, and are used by each magnetic disk 60, 61...6n to notify control signals (interrupt signals) to each magnetic disk control device DKC. It is.

In this way, in an input/output system having a dynamic path selection function, each magnetic disk (input/output device) 60 to 6n is a magnetic disk control device (input/output control device) for an access source such as the processor 1.
It is connected by two paths including DKC, and is configured to be accessed using either path, and the magnetic disk controller DKC of each path controls the magnetic disk according to commands from processor 1, which is the access source. I try to do it.

Next, the dynamic path selection operation of the embodiment configuration of FIG. 1 will be explained with reference to the operation explanatory diagram of FIG. 2 and the time chart of FIG. 4.

When the processor 1 accesses the magnetic disk, the processor 1 accesses the channel section 1a or 1.
b issues an access request (read/write, command address) to the director 2 or 4. To access the magnetic disk 60, the director 2 reads the contents of the memories 31 and 51 via the adapter 3, determines whether the magnetic disk 60 is in use, and if it is not in use, the director 2 reads the contents of the memories 31 and 51 via the adapter 3. A start command is given to the magnetic disk 60, and the magnetic disk 60 is connected via the cable 7 (FIG. 2A).

As a result, the director 2 transfers the read/write command, track, and cylinder address to the magnetic disk 60.

Upon receiving a response signal from the magnetic disk 60 upon completion of this transfer, the director 2 receives a response signal from the memory 3.
1 and 51 via the adapter 3, write path group information (group name of the path used) and device specific information (access status, device address, track, cylinder address, etc.), stop controlling the magnetic disk 60, Release the occupation of the path and push it away (Fig. 2C). The magnetic disk 60 starts a head seek operation based on the above-described given command and address.

Next, when the seek operation of the magnetic disk 60 is completed, a ready signal is given to the adapters 3 and 5 via the control lines 9a and 9b. At this time, assuming that the first path of the director 2, adapter 3, and cable 7 is being coupled to another magnetic disk, and the second path of the director 4, adapter 5, and cable 8 is empty, this completion signal is accepted by the director 4. At this time, director 4
reads the path group information in the memory 51, checks whether the first path combined in step is for the same group as its second path, that is, the same processor 1, accepts this completion signal, and sends the director 4 is reconnected to the magnetic disk 60 via an adapter 5 and a cable 8.
(Figure 2D).

As a result, the director 4 and the magnetic disk 6
Data is transferred between 0 and 0, and data is read from or written to the magnetic disk 60.

When the data transfer is completed, the director 4 erases the contents of the memories 31 and 51 corresponding to the magnetic disk 60 via the adapter 5.

Meanwhile, there is a magnetic disk 6 inside the step.
When access to 0 occurs, for example, via the channel section 1b, the director 4 reads the contents of the memories 31 and 51 via the adapter 5, confirms that the magnetic disk 60 is in use, and prevents the access from occurring. The permission is returned to the channel unit 1b to prevent conflicting activation. Conversely, when the magnetic disk 61 is accessed via the channel section 1 as shown in FIG. 61 is connected to the director 4 via the adapter 5 and cable 8, and simultaneous operations are performed.

In this way, the memories 31 and 51 store path control information used for exclusive control to prevent contention and for connection control at the time of recombination.

During data transfer in these steps, the control circuit of the adapter 5 performs data checks (parity checks) and operation checks to ensure data reliability.

When an error is detected through such a data check or operation check, the present invention performs the following retry control. This will be explained with reference to FIGS. 3 and 4.

(a) When the adapter 5 detects an error, it notifies the director 4. The director 4 stops data transfer with the magnetic disk 60 via the adapter 5, releases the second path occupation of the magnetic disk 60, and separates the magnetic disk 60 to perform a seek operation for retry. Along with this, Director 4
writes the type of error and repositioning information such as error records to the memory 51 of the adapter 5 and the memory 31 of the adapter 3 as retry information, as shown in FIG. 3A.

(b) When the magnetic disk 60 is repositioned, an interrupt (repositioning completion) signal is generated again via the control lines 9a and 9b, and is applied to the adapters 3 and 5.
At this time, if the first path (director 2, adapter 3) is empty as shown in FIG. 3B, this interrupt signal is accepted by adapter 3, director 2 reads the path group information from memory 31, Refer to this to perform recombination.

(c) Furthermore, the director 2 reads the retry information in the memory 31, determines the type of error (data check error or seek error), searches for the error record, and records the data and status signals from the magnetic disk 60 ( (signal obtained from a control line (not shown)) to check whether the error occurs again.

(d) If no error is detected by this check, it is determined to be normal and these data are transferred to the processor 1.

In this case, there is an abnormality in the second path or a temporary malfunction of the magnetic disk 60, and the processor 1 is notified of this.

On the other hand, if the error is detected again, the process returns to step (a), and if the error is not recovered even after repeating this process a predetermined number of times, the processor 1 is notified that the magnetic disk 60 itself is defective and appropriate measures are taken. .

In this way, errors in transferred data, magnetic disk operation errors (seek misses), etc. are detected and a retry is performed.

Although the example in FIG. 4 shows an error occurring during transfer of read data, the same applies when an error occurs during transfer after startup.

In the above example, the memories 31 and 51 are provided in both the adapters 3 and 5, but it is also possible to make this one memory so that it can be accessed by both control devices DKC.

Although the present invention has been described above using one embodiment, the present invention can be modified in various ways according to the gist of the present invention, and these modifications are not excluded from the present invention.

[Effects of the Invention] As explained above, according to the present invention, a plurality of paths connected to a common access source and each including an input/output control device, and an input/output device connected to each of the plurality of paths, , a memory that holds path control information, and after the input/output device is connected to one path and started, the path control information is written to the memory, the occupation of the path is released, and the input/output device is activated. In an input/output system that refers to the path control information of the memory to reconnect to a vacant path after the input/output device is ready and transfers data, the input/output control device performs the connection or recombination process. When an error is detected, the combined or recombined path is released, the input/output control device stores retry information in the memory, and the input/output device issues a connection request for retry. The input/output device is characterized in that the input/output device is reconnected with the path that is vacant when the input/output device is issued, and the input/output control device performs retry control of the input/output device based on the retry information in the memory. First, since the path is switched at the time of retry, recovery from path-specific failures can be expected, and a highly reliable system configuration is possible.

Second, since the paths are separated at some point during retry, the path utilization rate during error retry can be improved, and system throughput can also be improved.

Thirdly, since a dynamic path selection function using memory is used for retrying, control is easy and it can be easily implemented.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is an explanatory diagram of the dynamic path selection function of the configuration in FIG. 1, FIG. 3 is an explanatory diagram of retry control of the configuration of FIG. 1
An operation time chart in the diagram configuration is shown. In the figure, 1...Processor (access source), DKC
...Magnetic disk (input/output) control device, 60-6
n...Magnetic disk (input/output device), 31, 51
……memory.

Claims (1)

[Scope of Claims] 1. A plurality of paths connected to a common access source and each including an input/output control device, an input/output device connected to each of the plurality of paths, and a memory holding path control information. After connecting the input/output device to one path and starting it, write path control information to the memory, release the occupation of the path, and after the input/output device is ready, it becomes free. In an input/output system that refers to the path control information of the memory to reconnect and transfer data to a path connected to the memory, when an error is detected by the input/output control device during the connection or recombination, the In addition to de-occupying the combined or recombined paths, the I/O control device stores retry information in the memory, so that when the I/O device issues a connection request for a retry, the paths are free and the paths are unoccupied. 1. A retry control method for an input/output system, characterized in that the input/output device is recombined with the input/output device, and the input/output control device performs retry control of the input/output device based on retry information in the memory.