JPH03105540A - Input/output control system - Google Patents

Abstract

PURPOSE:To carry out an input/output request even while a countermeasure is taken to a channel fault and to prevent the discontinuation of the system operation caused by the channel fault by recognizing a fault input/output path via a channel fault countermeasure processing part and an input/output request processing part and sending the information to an I/O processor not to select the fault input/output path. CONSTITUTION:A channel fault detecting part 11 detects a channel fault and informs a channel fault countermeasure processing part 1 the fault. The part 1 analyzes the channel fault and detects a fault input/output path. At the same time, an input/output request processing part 2 accepts the input/output request from a user program and sets an input/output path where an input/output request issue grant mask is turned on and at the same time a fault production mask is turned off via a device information storage part 3. Then the part 2 gives an input/output request to an I/O processor 10. Thus it is possible to carry out an input/output request even while a countermeasure is taken to a channel fault and to prevent the discontinuation of the system operation caused by the channel fault.

Description

[Detailed Description of the Invention] [Summary] This invention relates to an input/output control method that reduces the time it takes to stop input/output devices under a faulty channel during the process of system reset as a countermeasure against a faulty channel. For input/output devices with other input/output paths, the channel fault countermeasure processing unit and input/output request processing unit recognize the faulty input/output path and prevent the fault input from occurring in the I/O processor, which is the hardware that controls the channel. By notifying the user not to select an output path, the purpose is to allow input/output requests to be executed even when channel failure countermeasures are being taken, and to prevent the system from stopping due to channel failure. a plurality of input/output paths provided for each of the plurality of input/output devices; A channel that is on the input/output path and controls the input/output device, and the input/output device selects which channel to use. O
In a computer system operated by a processor, there is a channel failure detection unit that detects channel failures and an analysis of channel failures. The channel failure countermeasure processing unit that performs the countermeasure processing, the input/output request processing unit that executes the input/output request, the input/output device and the channel are connected, and the input/output path that can accept the issuance of the input/output request is The input/output request issuance permission mask is set to on and off for input/output paths that cannot be accepted. It has a device information storage unit that stores a failure occurrence mask as input/output device information, which is set to ON for the input/output path that is causing the failure, and set to OFF for the other paths, and when a channel failure occurs. , the channel fault detection section detects a fault and notifies the channel fault countermeasure processing section of the fault, and the channel fault countermeasure processing section analyzes the channel fault, detects the faulty input/output path, and The input/output request processing unit receives an input/output request from the user program, and determines that the input/output request issuing permission mask from the device information storage unit is on. At the same time, an input/output path with the failure occurrence mask set to OFF is set, and care is taken to provide a data processing device that makes the input/output request to the ■/○ processor.

[Industrial application field]

The present invention is developed in the process of resetting the system as a measure against channel failure. The present invention relates to an input/output control method that reduces the time it takes to stop input/output devices under a faulty channel.

In recent years, as the processing speed of CPUs has increased in computer systems, there has been a demand for faster input/output processing. In addition, with the demand for highly reliable systems. Early detection of faulty devices and early system recovery are required.

[Prior art] Therefore, by offloading the path selection of input/output devices to the hardware side, the automatic input/output path selection function and the input/output request waiting management function are released from software, and the CPU
Although systems have been developed to reduce overhandling, there is no path selection method in the event of a channel failure, so all input/output devices under the failed channel must be shut down until channel failure countermeasure processing is completed.

In Figure 2. A hardware configuration diagram of input/output control is shown.

In the figure, 84, 85, 86 and 87 are channels, and 90
, 91.92 are input/output devices, and 93 is the channel 85.
94 is an input/output path of the input/output device 90 connected to the channel 84 . 95 is an input/output path of the input/output device 90 connected to the channel 86, and 96 is an input/output path of the input/output device 90 connected to the channel 87. 88 is channel 84.85
89 is an input/output control device that controls channel 8.
This is an input/output control device that controls 6.87.

7 is a data processing device, which controls the entire processing. 80 is a channel failure countermeasure processing section, 8l is an input/output request processing section, and 182 is a device information storage section. 8
3 is a ■/○ processor, which is hardware that selects 5 input/output paths. Here, the input/output path is the path through which input/output is performed. When a channel exists between them, it is called a channel path. Note that the computer system in this figure is a system in which paths between channels and input/output devices are dynamically connected and reconnected.

In this system, in order to make effective use of channel resources, the connection to the channel is disconnected while the disk, etc. is performing mechanical operations. Channels are freed up to do other work. Further, when a disk or the like is reconnected to a channel, an available channel path is selected from among all channel paths connected to the system that initiated the input/output operation.

This is called a dynamic recombination function (DPR function).

FIG. 6 is a functional block diagram of the conventional technology.

First, input/output processing during normal operation will be explained along the hardware configuration shown in FIG.

In the figure. A channel failure detection unit 101 detects the occurrence of a channel failure. Reference numeral 102 denotes a fixed failure detection unit using an input/output statistical method, which mainly detects soft failures. 1
03 is an input/output request, which is an input/output request of the user program. 104 is a channel failure countermeasure processing unit;
1 is an input/output request processing unit, 106 is a device information storage unit, and 107 is a TOP (I/O processor).

A quiesce section 108 for general input/output requests temporarily stops the input/output requests. I09 is an executing input/output request collecting unit, which stops the executing input/output and puts it in a queue. Reference numeral 110 denotes a reserve path switching unit, which transfers input/output information to another input/output path. A system reset issuing unit 111 resets all input/output path information under the faulty channel. A system reset analysis unit 112 analyzes the result of system reset.

A path group reestablishment unit 113 restores a failed input/output path to a healthy path group. 114 is a general input/output request quiesce release unit, which releases the quiescing of the input/output requests under the faulty channel. A restart request unit 115 restarts input/output requests on the waiting queue. Reference numeral 116 denotes an input/output request accepting unit, which accepts input/output requests of user programs. A CCW creation unit 117 creates instructions for making input/output requests to channels.

118 is an input/output request issue request mask determination unit. Make an input/output request using the above path group as parameters. Reference numeral 121 denotes an input/output request waiting management unit, which causes the input/output request to wait if there is a preceding request. 120 is a status analysis unit that analyzes input/output execution results. 122 is a notification section to the request source. The execution result is notified to the user program. Reference numeral 123 denotes an input/output request retry unit, which re-executes the request if the execution result is abnormal. 124 is a subchannel number in which the number of the input/output device is stored. 125 is Quiesc
e control information. Input/output request Quiesce section 10
Information about input/output requests that were stopped at step 8 is stored. 1
Reference numeral 26 is reserve information, which is information related to input/output that a certain input/output device has. 128 is an input/output request mask. Contains input/output paths that allow input/output.

Reference numeral 129 is path group information, in which path groups used when making an input/output request to the IOP 107 are stored.

Reference numeral 131 denotes a startup path selection unit, which selects a path for actual input/output from a path group. Reference numeral 132 denotes an input/output activation unit, which issues input/output requests to the channel. 13
3 is the dynamic path recombination processing section. Performs DPR function.

In addition, in Figure 6. Some processes are the same as those shown in FIG. 2, so they will be described below. The channel fault countermeasure processing unit 104 in FIG. 6 and the channel fault countermeasure processing unit 80 in FIG. 2 are the same. This is expressed as 104 and 80 being the same. Similarly, 105 and 79, 106 and 82, and 107 and 83 are the same.

With the above structure, when there is a normal input/output request, the input/output request processing unit 105 accepts this input/output request. C.C.W.
The system CC is applied to the CCW created by the user in the discipline department 117.
W is added, and the input/output request issuance request mask determination unit 11B sets an input/output request issuance request mask (L P M ). The input/output request issuing unit 11 issues an input/output request (SSCH command).

For example, in the case of an input/output request to the input/output device 90. There are four paths (
93,94.95.96) Exists. If there is no abnormality, all these input/output paths are online. in this case. The input/output request mask 12B regarding the input/output device 90 in the device information storage unit 106 is . The mask bits corresponding to the four paths mentioned above are on. Furthermore, in a system in which path selection for input/output devices is offloaded to the hardware side, the input/output devices usually hold path groups. The four paths mentioned above are the path group of the device. With the dynamic path recombination function (DPR function). Any path within this group can be used to rejoin. Note that this path group information is stored in the input/output control device 88 on the hardware side.

The input/output request processing unit 105 checks whether there is a preceding request for the input/output device 90, and if the input/output request is already being executed, the input/output request waiting management unit 121 checks whether there is a preceding request. Make the request wait. Unless there is a prior request. Set the I/O request issue request mask and issue the I/O request.

When 10P accepts this input/output request, 10P selects a boot path for the relevant input/output device.

The IOP recognizes which channel the input/output device is connected to, and sequentially activates the channels connected to the requested input/output device.

If the channel is executing an input/output request to another input/output device, the channel is in use, but in this case, the software is not notified and the next channel is selected within the IOP.
Try booting again.

When the input/output processing is completed, an I/O interrupt is notified to the software. The input/output request processing unit reaps the status (TSCH command) in order to obtain the end status of the input/output request and the device status. Perform analysis. If the input/output request ends normally, the request source is notified of the normal end. On the other hand, if the input/output request terminates abnormally, after notifying the fixed failure detection unit 102 using the input/output statistical method of this abnormal state. Skip this input/output request retry (ERP). If it does not complete successfully even after retrying a certain number of times. For input/output request sources. Notifies that input/output processing has ended abnormally.

next. Processing in the conventional technology when a channel failure occurs will be explained.

There are two types of failure detection for input/output devices.

One is a severe channel failure signaled by the hardware via a machine check interrupt; It statistically accumulates the input/output errors that occur within the system and pinpoints the location of the failure.

When an input/output machine check interrupt is reported. The channel fault detection unit 101 issues an STCRW command, records the status, and analyzes which channel the fault is in. Also. A fixed failure detection unit using an input/output statistical method determines that a particular channel is a failure if the frequency of occurrence of input/output errors is high in that particular channel. In this way, the faulty channel is identified and the channel fault countermeasure processing unit 104 is scheduled. for example. Assume that a machine check interrupt indicates that channel 84 is faulty.

The channel fault countermeasure processing unit 104 first sets Quiesc so that all input/output devices 90, 91, and 92 under the faulty channel 84 do not execute general input/output requests even if they subsequently receive a general input/output request.
The e unit 108 performs static processing of the input/output request and makes it wait on the waiting queue.

Next, the executing input/output request collection unit 109 issues the HSC}l command to the input/output device currently being executed to stop the input/output request. This is because if a system reset (RCHP command) is issued during data transfer, data cannot be guaranteed. The input/output request that was being executed is forcibly terminated. It will wait for a retry in the form of a pseudo I/O error.

In addition, the HSC issued by the channel failure countermeasure processing unit 104
The H command is controlled so that it can be executed even when the input/output device is not making any input/output requests. In other words, while I/O requests are quiescent, general I/O requests are made to wait. The input/output requests of the channel failure countermeasure processing unit 104 are controlled by level control so that they can be executed.

Furthermore. If the input/output device in question holds a reserve, the reserve path switching unit 110 transfers the reserve information to another input/output path so that the reserve information is not lost due to system reset. This is done by executing the UR command from a different path. In addition, if a path group exists. A system reset automatically causes the system reset issued path to leave the group.
Since the reserve information is transferred to the remaining path groups, there is no need to issue a UR command in this case.

After the above preparations are completed, the system reset issuing unit 11
1 issues a system reset to the faulty channel. This command resets the channel paths of the faulty channel and all subordinate input/output devices connected to the channel.

The system reset analysis unit 11 completes the system reset.
2 is notified by machine change interrupt. When I practiced the system resent, the faulty channel path that was removed by issuing the system resent became a healthy path. In order to incorporate it into the path group again, the path group re-establishment unit 113 specifies this failed path and sends it to ESTABL.
Issue ISH I/O. In this case, if the input/output request processing unit 105 has specified a path, This path is used as the input/output request issue request mask 128, and dynamic recombination (D
PR) Add the SMPR command to disable the function.

After the path group re-establishment process is completed. General input/output request Q
The uiesce release unit 114 releases the quiescence of input/output requests of all input/output devices under the faulty channel. Make general I/O requests executable. Next, the restart request unit 115
Schedule restart processing for all input/output devices. Executes I/O requests on the waiting queue. At this point, the path group has returned to its pre-failure state, so input/output processing is the same as normal.

[Problems to be Solved by the Invention] As described above, in the conventional technology, important and frequently accessed systems such as magnetic disks that store data sets to be paged in input/output devices under failed channels are If an expensive device exists, input/output requests to this input/output device will have to wait until the troubleshooting process for the faulty channel is completed, creating a problem that the system may stop during this time.

The present invention is. For input/output devices that have an input/output path other than the path connected to the failed channel, the channel failure countermeasure processing unit and input/output request processing unit are hardware that recognizes the failed input/output path and controls the channel. By notifying the I/O processor not to select the failing I/O path. The purpose is to execute input/output requests even when channel failure countermeasures are being taken to prevent the system from stopping due to channel failure.

[Means to solve the problem]

FIG. 1 is a diagram explaining the principle of the present invention.

In the figure. l is a channel failure countermeasure processing unit. channel 7
Investigate failures and take countermeasures.

2 is an input/output request processing unit, which is used when a channel failure occurs. Make input/output requests. Reference numeral 3 denotes a device information storage unit, which displays an input/output request issuing permission mask 4 that displays input/output paths for which an input/output request can be issued and input/output paths that are not suitable, and an input/output path in which a failure has occurred. It has a failure occurrence mask 5 and stores information on input/output devices. Reference numeral 6 denotes a data processing device that performs the entire processing described above. 8 is an input/output device, 9
is an input/output path, which is the path through which the channel 7 controls the input/output device 8. 10 is an I/O processor, which is hardware for selecting an input/output path. Reference numeral 11 denotes a channel failure detection unit, which detects channel failures.

Note that some parts of Fig. 1 are the same as those of Fig. 2, so they are shown below. The channel failure countermeasure processing section 1 in FIG. 1 and the channel failure countermeasure processing section 80 in FIG. 2 are the same. This is abbreviated and 1 and 80 are expressed as the same. Similarly, 2, 81.3, 82.6, 79, 10 and 83 are the same.

With the above configuration, when a channel failure occurs, the -y- or channel failure detection unit ll in the data processing device 6 notifies the channel failure countermeasure processing unit 1 of the failure.

The channel fault countermeasure processing unit 1 temporarily stops the input/output device 8 under the faulty channel. The faulty input/output path is stored in the fault occurrence mask 5 for each input/output device 8 in the device information storage unit 3, and a message indicating that fault countermeasures are being taken is displayed. Thereafter, the input/output requests being executed for all input/output devices 8 under the faulty channel are recovered.

Thereafter, the stoppage of the input/output request is canceled and the input/output request processing section 2 is requested to start the input/output request.

The input/output request processing unit 2 detects a path that allows input/output requests, and issues an input/output request to the input/output device using the path that allows input/output as a parameter. Next, the channel failure countermeasure processing unit 1 issues a system reset.

Next, reset the error occurrence mask and turn off the error countermeasure processing display.

[Operation] As described above, in the present invention, an input/output path in which a failure has occurred is stored in the device information storage unit as a failure occurrence mask, and input/output is possible using the input/output request issuance permission mask and the failure occurrence mask. By determining the path and using this input/output enabled path as an argument to issue an input/output request when issuing a system reset, even when the channel failure countermeasure processing unit issues a system reset note to the faulty channel and executes countermeasure processing,
Allows I/O requests to be performed using other I/O paths.

〔Example〕

FIG. 3 shows a functional block diagram of an embodiment of the present invention. however. The hatched boxes indicate parts that are different from the conventional technology.

Figure 3 is. Since it is almost the same as FIG. 6, the changes made by the present invention compared to the conventional technology will be explained.

Reference numeral 45 is a section for displaying the fault countermeasures in progress and setting the fault path mask. This is set when a channel failure occurs. Reference numeral 53 denotes a fault countermeasure cancellation and fault occurrence path mask reset section, which is a process performed when the fault is cleared. 56 is S
This is an MPR command addition section that returns data to the channel that received the input/output request. 58 is a processing unit that performs exclusive OR with the failure path mask. Set a path that allows input/output requests during a failure. 7l is a fault countermeasure processing display, indicating that a channel fault has occurred.

Reference numeral 72 is a failure path mask, which indicates the path causing the failure.

In addition, in Figure 3. Some items are the same as those in Figure 6, so they are shown below. For example, the channel failure detection unit 37 in FIG. 3 and the channel failure detection unit 101 in FIG. 6 are the same. This will be abbreviated and 37 and 101 will be expressed as the same. 38 and 10 similarly below
2.39 and 103.44 and 108.46 and 109.47
and 110.48 and 114 49 and 115.50 and 11
1.51 and 112.52 and 113.54 and 116.55
and 117.57 and 118.59 and 119.61 and 121
．． 60 and 120.62 and 122.63 and 123.64 and 124.65 and 125.66 and l26.68 and 128.
69, 129.73, 131.74 and 132 75 and 133 are the same. As a whole function, 40 and 104
．． 41 and 105, 42 and 106, 43 and 107 correspond to each other as having the same function.

here. Based on the hardware configuration diagram of FIG. 2, the parts improved by the present invention will be mainly explained.

Note that some of the processes in Figure 3 are the same as those in Figure 2, so they will be described below to avoid confusion. The parts 80 are the same.This will be abbreviated as 40 and 80.In the same way, 41 and 81.42 and 82.4
3 and 83 are the same.

Under the configuration shown in FIG.
．． The Qui
The esce unit 44 performs static processing of input/output requests. next. The fault countermeasure display and fault occurrence path mask setting unit 45 sets the fault countermeasure display and fault occurrence path mask (4 2. 7 1', 7) for all devices under the fault channel.
Make the settings for 2). next. The running input/output request collecting unit 46 stops the input/output request to the input/output device currently being executed. The I/O request that was in progress is forcibly terminated, and a pseudo I/O error awaits retry. Next, the reserve path switching unit 47 transfers the reserve information 66 held by the input/output device to another input/output path.

Once the above preparations are completed, it is possible to execute input/output requests from online paths other than the faulty channel path. This is because the range of influence of a system reset is the channel path of the faulty channel and all subordinate input/output devices connected to the channel.

Here, while the system resent is being issued to the faulty channel 84, the input/output device 90 under the faulty channel is being issued.
Consider executing an input/output request using a path 95 that uses a channel different from the failed channel. The general input/output request Quie Sce cancellation unit 48 cancels the quiescence of the input/output requests of all input/output devices under the faulty channel, makes the general input/output requests executable, and the restart request unit 49 releases all input/output requests. Schedule restart processing for the output device. This causes the input/output requests on the waiting queue to be executed.

In the input/output request processing section 41. If the input/output device is in troubleshooting process, the SMPR command addition unit 56 sends the SMP
Add R command. Prohibit dynamic recombination. Next, the processing unit 58 that takes the exclusive OR with the failure path mask selects the input/output request mask that is currently considered to be able to issue an input/output request (at this point, the input/output request mask 68 of the input/output device 90 is set to 4). bits corresponding to all input/output paths 93, 94, 95, and 96 are turned on) and the fault occurrence path mask set by the channel fault countermeasure processing unit 40, and Do not select.

In this way, the input/output device 90 is requested to issue an input/output request to the IOP with the input/output issue request mask (the focus corresponding to the input/output paths 93, 95, and 96 is on). In rOP. Select one path from these three paths and activate it. Also. Since the recombination 51 also has an SMPR condo added, it is recombined using the same path as the path used for activation. Here 2 If you request an input/output request with three paths,
If the SMPR command was not added, the input/output controller 88 still believes that four paths form a path group. Fault channel path 9
There is a possibility of recombination in 4. in this case. If a system reset occurs on this path, the I/O request completion interrupt will be lost.

In this manner, the system reset issuing unit 50 asynchronously issues a system reset to the faulty channel while allowing execution of the 1C input/output request to all devices under the faulty channel. It works great for resetting the system. The path group reestablishment unit 52 issues a system reset, and the failed channel path that was withdrawn has returned to a healthy path. In order to incorporate it into the path group again, issue ESTABLISH 1/0 by specifying this failed node 94. After completing the re-establishment of the path group, cancel the display indicating that the troubleshooting process is in progress and cancel the troubleshooting process in the failure path mask Nori Nanano 1/Part 53. Reset the failed path mask.

Later. The input/output request processing section 41. Since failure countermeasure processing is not in progress, normal input/output processing will be performed.

Next, FIG. 4 shows a detailed diagram of the device information storage section.

In FIG. 4, 151 is a device information storage unit regarding information on one input/output device. Regarding the information on one input/output device below, 152 is a fault countermeasure processing display section, and when this is on, fault processing is in progress. When it is off, it means that no fault countermeasures are being taken.

153 is an input/output request issuance permission mask. Indicates channel paths that are allowed to issue input/output requests and channel paths that are not allowed. A path with a mask value of 1 indicates that an input/output request can be issued, and a path with a mask value of 0 indicates that an input/output request cannot be issued. 1
54 is a failure occurrence mask, which indicates a channel path in which a failure has occurred. 155 is mask A, which is an example of an input/output request issuance permission mask. 156 is mask B, which is an example of a failure occurrence mask. By calculating the exclusive OR of mask A and mask B, it is possible to control so that an input/output request is not issued from a faulty channel path.

Finally, the differences between the processing of the present invention and the processing of the prior art will be shown using a time chart.

Figure 5 is. FIG. 3 is a time chart of failure processing.

The upper part of FIG. 5 is a time chart of troubleshooting according to the prior art, and the lower part is a timechart of troubleshooting according to the present invention.

The failure handling diagram of FIG. 5 shows the operation of channels 84 and 85 in the configuration diagram of FIG.

The time chart shows the processing from when a failure occurs in the channel 84 until the failure is recovered.

The details of each processing step are omitted as they have already been explained.
As shown in the figure, in the conventional technology. The period from when a failure occurs in the channel 84 until it starts after the failure is recovered.
Although the processing is suspended, in the present invention, normal processing can be resumed after the input/output processing unit issues an I/O request and the healthy channel 85 starts input/output processing. That is, in the present invention, restart of input/output processing is significantly faster than in the prior art.

〔Effect of the invention〕

As described above, according to the present invention, even if a channel failure occurs, Execute I/O requests to I/O devices under the faulty channel. This can be done even while recovery processing is being performed on a faulty channel, which greatly contributes to speeding up input/output processing in the system.

[Brief explanation of drawings]

Figure ■ is. This is a diagram explaining the principle of the present invention. Figure 2 is a hardware configuration diagram for input/output control, and Figure 3 is. FIG. 4 is a functional block diagram of an embodiment of the present invention, and FIG. 4 is a detailed diagram of the device information storage section. Figure 5 shows
FIG. 6 is a time chart of failure processing, and FIG. 6 is a functional block diagram of a conventional technique. In Figure 1. l is a channel failure countermeasure processing unit, 2 is an input/output request processing unit, 3 is a device information storage unit, is an input/output request issuance permission mask, and is a failure occurrence mask. is a data processing device. is a channel, is an input/output device, is an input/output path, 0 is a ■/○ processor (IOP), and 1 is a channel failure detection unit. Motomei's original burial explanation prisoner Ward 1 -332-

Claims (2)

[Claims] (1) A plurality of input/output devices (8) and a plurality of input/output paths (9) provided for each of the plurality of input/output devices (8).
) and the input/output device (8) on the input/output path (9).
and a channel (7) for controlling the input/output device (8).
) detects a channel failure in a computer system in which the I/O processor (10), which is the hardware that controls the channel (7), performs input/output path selection to select which channel (7) to use. A channel fault detection unit (11), a channel fault countermeasure processing unit (1) that analyzes channel faults and takes countermeasures, an input/output request processing unit (2) that executes input/output requests, and an input/output device. (8) and channel (7) are connected,
The I/O request issuance permission mask (4) is set to ON for I/O paths that can accept I/O requests and OFF for I/O paths that cannot be accepted, and for the I/O paths that are causing problems. and a device information storage unit (3) that stores a failure occurrence mask (5) that is set to be on for the input/output device (8) and turned off for the others, and when a channel failure occurs, , the channel failure detection unit (11
) detects the failure, and the channel failure countermeasure processing unit (1
), the channel fault countermeasure processing unit (1) analyzes the channel fault and detects the faulty input/output path, and the input/output request processing unit (2) handles the input/output from the user program. Upon receiving the request, the device information storage unit (3) sets an input/output path in which the input/output request issuance permission mask is on and at the same time the failure occurrence mask is off, and sends the input/output request to the I/O processor. An input/output control method characterized by comprising a data processing device (6) that performs processing for (10). (2) When the input/output request processing unit (2) makes an input/output request to a predetermined input/output device (8), the input/output is always performed from the channel that issued the input/output request. 1
Input/output control method described in section.