JP2730209B2 - I / O control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Overview] An I / O control method for reducing the time required to stop an I / O device under a failed channel in the process of resetting the system in response to a channel failure, and relates to a path connected to the failed channel. For an I / O device that has an I / O path other than, the channel failure countermeasure processing unit and the I / O request processing unit recognize the failed I / O path and input a failure to the I / O processor that is the hardware that controls channels. By notifying the selection of the output path, the I / O request is executed even during the countermeasure for the channel failure, and the purpose is to prevent the system from being stopped due to the channel failure. A plurality of input / output paths provided for each of the output devices; and channels on the input / output paths for controlling the input / output devices. The output device selects the input / output path to select which channel to use, and the I / O is the hardware that controls the channel.
In a computer system performed by a processor, a channel failure detection unit that detects a channel failure, a channel failure countermeasure processing unit that analyzes a channel failure and performs countermeasure processing, an input / output request processing unit that executes an input / output request, An I / O request issuance permission mask in which an I / O device and a channel are connected and an I / O path that can accept issuance of I / O requests is set to ON and an I / O path that is not accepted is set to OFF; A device information storage unit that stores a failure occurrence mask in which an I / O path in which a failure has occurred is set to ON and an OFF otherwise is stored as information of an I / O device; The channel failure detection unit detects the failure and notifies the channel failure countermeasure processing unit of the failure, and the channel failure countermeasure processing unit performs a channel failure analysis process. The I / O request processing unit receives the I / O request from the user program, and the I / O request issuance permission mask is turned on from the device information storage unit. Set an I / O path that is off, and
It is configured to have a data processing device for the I / O processor.

[Industrial applications]

The present invention relates to an input / output control method for reducing the time for stopping input / output devices under a failed channel in the course of performing a system reset as a countermeasure for a channel failure.

2. Description of the Related Art In recent computer systems, with an increase in the processing speed of a CPU, an increase in the speed of an input / output device has been required. Further, with the demand for a highly reliable system, early detection of a faulty device and early recovery by the system are required.

[Conventional technology]

Therefore, by offloading the I / O device path selection to the hardware side, a system is provided that releases the I / O path automatic selection function and the I / O request wait management function from software and reduces CPU overhead. But
Since there is no path selection method at the time of a channel failure, all input / output devices under the failed channel have to be stopped until the channel failure countermeasure processing is completed.

 FIG. 2 shows a hardware configuration diagram of the input / output control.

In the figure, 84, 85, 86, and 87 are channels, 90, 91, and 92 are input / output devices, and 93 is an input / output device 90 connected to the channel 85.
Reference numeral 94 denotes an input / output path of the input / output device 90 connected to the channel 84. Reference numeral 95 denotes an input / output path of the input / output device 90 connected to the channel 86, and reference numeral 96 denotes an input / output path of the input / output device 90 connected to the channel 87. 88 is channel 8
An input / output controller that controls 4,85, 89 is channel 8
It is an input / output control device that controls 6,87. Reference numeral 79 denotes a data processing device which controls the entire processing. 80 is a channel failure countermeasure processing unit, 81 is an input / output request processing unit, and 82
Denotes a device information storage unit. Reference numeral 83 denotes an I / O processor, which is hardware for selecting an input / output path. Here, the input / output path is a path through which input / output is performed, and when a channel is interposed therebetween, 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, the connection with the channel is disconnected while the disk or the like is performing a mechanical operation for effective use of the channel resource, so that the channel can perform other work. Further, when a disk or the like is reconnected to a channel, a usable one is selected from all channel paths connected to the system that has started the input / output operation. This is called the dynamic recombination function (DP
R function).

FIG. 6 is a functional block diagram of a conventional technique. First, the normal input / output processing will be described with reference to the hardware configuration of FIG.

In the figure, reference numeral 101 denotes a channel failure detection unit which detects the occurrence of a channel failure. Reference numeral 102 denotes a fixed fault detection unit based on the input / output statistical method, which mainly detects a soft fault. 103
Is an input / output request, which is an input / output request of a user program. 104 is a channel failure countermeasure processing unit, 105 is an input / output request processing unit, 106 is a device information storage unit, and 107 is an IO
P (I / O processor). 108 is the general input / output request Q
The uiesce section temporarily stops input / output requests. Ten
Reference numeral 9 denotes an executing I / O request collection unit that stops the executing I / O and places it in a queue. Reference numeral 110 denotes a reserved path switching unit which transfers the input / output information to another input / output path. 111
Is a system reset issuing unit that resets all input / output path information under the failed channel. Reference numeral 112 denotes a system reset analysis unit that analyzes a result of the system reset. Reference numeral 113 denotes a path group re-establishment unit that returns a failed input / output path to a sound path group. Reference numeral 114 denotes a general input / output request quiesce canceling unit which cancels the quiescing of the input / output request under the faulty channel. A restart request unit 115 restarts an input / output request on the waiting queue. An input / output request receiving unit 116 receives an input / output request of a user program. Reference numeral 117 denotes a CCW creation unit that creates an instruction for making an input / output request to a channel. Reference numeral 118 denotes an input / output request issuance request mask determination unit, which makes an input / output request using the path group as a parameter. Reference numeral 121 denotes an input / output request waiting management unit, which waits for the input / output request when there is a preceding request.
Reference numeral 120 denotes a status analysis unit that analyzes an execution result of input / output. Reference numeral 122 denotes a notification unit for notifying the request source, and notifies the user program of the execution result. Reference numeral 123 denotes an input / output request retry unit, which re-executes when the execution result is abnormal. Reference numeral 124 denotes a subchannel number in which the number of an input / output device is stored. Reference numeral 125 denotes Quiesce control information.
The information of the input / output request quiesced in the table is stored. 126
Is reserved information, which is information relating to input / output possessed by a certain input / output device. Reference numeral 128 denotes an input / output request mask, which stores input / output paths that allow input / output. 129
Is path group information, which stores a path group when an I / O request is made to the IOP 107. Reference numeral 131 denotes an activation path selection unit that selects a path to be actually input / output from a path group. Reference numeral 132 denotes an input / output activation unit that issues an input / output request to a channel. Reference numeral 133 denotes a dynamic path reconnection processing unit which performs a DPR function.

It should be noted that some processes in FIG. 6 are the same as those in FIG. 2 and are described below. The channel failure countermeasure processing unit 104 in FIG. 6 and the channel failure countermeasure processing unit 80 in FIG. 2 are the same. This indicates that 104 and 80 are the same. Similarly, 105 and 79, 106 and 82, and 107 and 83 are the same.

With the above configuration, when there is a normal input / output request, the input / output request processing unit 105 receives this input / output request. The CCW creation unit 117 adds the system CCW to the CCW created by the user,
The input / output request issuance request mask determination unit 118 sets an input / output request issuance request mask (LPM), and the input / output request issuance unit 119 issues an input / output request (SSCH instruction). For example, input / output devices
For I / O requests to 90, there are four paths (93,94,95,96) that can issue I / O requests to the device, and if there is no error, all of these I / O paths are online. It is. In this case, in the input / output request mask 128 for the input / output device 90 in the device information storage unit 106, the mask bits corresponding to the above-described four paths are on. In a system in which the path selection of the input / output device is offloaded to the hardware, the input / output device normally holds a path group. The above-mentioned four paths are a path group of the device, and a dynamic path reconnection function (DPR function) may be used to reconnect using any path in this group. The 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 any preceding request to the input / output device 90. If the input / output request is already being executed, the input / output request wait management unit 121 causes the request to wait. If there is no preceding request, an I / O request issuance request is issued by setting an I / O request issuance request mask.

When the IOP receives the input / output request, the IOP selects a boot path for the input / output device. The IOP recognizes which channel the input / output device is connected to, and sequentially activates the channel connected to the requested input / output device. If the channel is executing an I / O request to another I / O device, the channel is in use. In this case, the software is not notified, the next channel is selected in the IOP, and an attempt is made to start again.

When the input / output processing is completed, an I / O interrupt is notified to software. The input / output request processing unit cuts the status (TSCH instruction) and analyzes it to obtain the end state of the input / output request and the device status. If the I / O request has been completed normally, the request source is notified of the normal completion. On the other hand, if the I / O request ends abnormally, this abnormal state is notified to the fixed failure detection unit 102 using the I / O statistical method, and then
Retry (ERP) this I / O request. If the retry is not completed normally even after a certain number of retries, the I / O request source is notified that the I / O processing has ended abnormally.

Next, a description will be given of processing by a conventional technique when a channel failure occurs.

There are two types of fault detection for input / output devices. One is a severe channel failure notified from hardware by a machine check interrupt, and the other is to statistically accumulate input / output errors that have occurred in the system and identify the location of the failure.

When an I / O machine check interrupt is reported,
The channel failure detection unit 101 issues an STCRW instruction, cuts the status, and analyzes which channel has a failure. In addition, in the fixed failure detection unit using the input / output statistical method,
If I / O errors occur frequently only for specific channels,
The channel is determined to be faulty. In this way, the failed channel is specified, and the channel failure countermeasure processing unit 104 is scheduled. For example, a machine check interrupt
Assume that channel 84 has been notified of a failure.

In the channel failure countermeasure processing unit 104, first, the failure channel 8
4. The Quiesce unit 108 performs a quiesce process on the input / output request so that all the input / output devices 90, 91, and 92 under the control do not execute the general input / output request thereafter, and waits on the waiting queue.

Next, the executing I / O request collection unit 109 issues an HSCH instruction to the currently executing I / O device to stop the I / O request. This is a system reset (RC
Issuing the HP instruction does not guarantee the data. The I / O request that was being executed will be forcibly terminated and will wait for a retry in the form of a pseudo I / O error.

Further, the HSCH instruction issued by the channel failure countermeasure processing unit 104 is controlled so that the HSCH instruction can be executed even while the input / output device is in the input / output request state. That is, while the I / O request is quiesced, the general I / O request is waited, but the channel failure countermeasure processing unit
The level control is performed so that the input / output request of 104 can be executed.

Furthermore, when the input / output device holds the 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 the system reset. This is done by executing the UR command from another path. If a path group exists, the system reset automatically causes the system reset issue path to be withdrawn from the group, and reserve information is transferred to the remaining path groups. In this case, issue a UR command. do not have to.

After the above preparations are completed, the system reset issuing unit 11
1 issues a system reset to the failed channel. With this instruction, the channel paths of the faulty channel and all the subordinate I / O devices connected to the channel are reset.

Completion of system reset is performed by the system reset analysis unit 11.
It is notified by a machine check interrupt at 2. If the system reset succeeds, the failed channel path that has been withdrawn by issuing the system reset has returned to a healthy path. To incorporate the failed channel path into the path group again, the path group reestablishing unit 113 specifies this failed path and sets ESTABLISH. Issue I / O. In this case, if the path is specified, the input / output request processing unit 105 sets the path as the input / output request issue request mask 128 and adds the SMPR command to prohibit the dynamic reconnection (DPR) function. I do.

After the path group re-establishment process is completed,
The quiesce canceling unit 114 cancels the quiesce of the I / O requests of all the I / O devices under the failed channel, and makes the general I / O requests executable. Next, the restart request unit 115 schedules a restart process for all the input / output devices, and executes the input / output requests on the waiting queue. At this point, since the path group has also returned to the state before the failure, the input / output processing is the same as the normal input / output processing.

[Problems to be solved by the invention]

As described above, according to the conventional technology, if there is an important and frequently accessed object in a system such as a magnetic disk storing a data set to be paging in an input / output device under a failure channel, An input / output request to the device is waited until the countermeasure processing for the failed channel is completed, and during this time, there is a problem that the system may be stopped.

According to the present invention, for an I / O device having an I / O path other than the path connected to the failed channel, the channel failure countermeasure processing unit and the I / O request processing unit recognize the failed I / O path and control the channel. The purpose is to notify the hardware I / O processor not to select a failed I / O path so that I / O requests can be executed even during countermeasures against channel failures, thereby preventing the system from being stopped due to channel failures. .

[Means for Solving the Problems] FIG. 1 is a diagram illustrating the principle of the present invention.

In the figure, reference numeral 1 denotes a channel failure countermeasure processing unit,
Investigate any failures and take countermeasures. Reference numeral 2 denotes an input / output request processing unit that makes an input / output request when a channel fails.
Reference numeral 3 denotes an apparatus information storage unit, which displays an I / O request issuance permission mask 4 for displaying an I / O path which may or may not issue an I / O request and an I / O path in which a failure has occurred. It has a fault 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. Reference numeral 8 denotes an input / output device, 9 denotes an input / output path, and serves as a path when the channel 7 controls the input / output device 8. Reference numeral 10 denotes hardware for selecting an input / output path by an I / O processor. Reference numeral 11 denotes a channel failure detection unit that detects a channel failure.

1 are the same as those in FIG. 2 and are shown below. The channel failure countermeasure processing unit 1 in FIG. 1 and the channel failure countermeasure processing unit 80 in FIG. 2 are the same. For simplicity, 1 and 80 are the same. Similarly, 2 and 81, 3 and 82, 6
79, 10 and 83 are identical.

With the above configuration, when a channel failure occurs, the channel failure detection unit 11 in the data processing device 6 notifies the channel failure countermeasure processing unit 1 of the failure. The channel failure countermeasure processing unit 1 temporarily stops the input / output device 8 under the fault channel, stores the fault input / output path in the device information storage unit 3 in the fault occurrence mask 5 for each input / output device 8, Is displayed. Thereafter, the input / output requests being executed for all the input / output devices 8 under the failed channel are collected. Thereafter, the suspension of the I / O request is released, and the I / O request processing unit 2 is requested to restart the I / O request. The input / output request processing unit 2 detects an input / output requestable path and performs an input / output request to the input / output device using the input / output path as a parameter. Next, the channel failure countermeasure processing unit 1 issues a system reset. Next, the fault occurrence mask is reset, and the display during the fault countermeasure processing is turned off.

[Action]

As described above, according to the present invention, a failure input / output path is stored in the device information storage unit as a failure occurrence mask.
The I / O request issue permission mask and the fault occurrence mask determine a path through which I / O can be performed, and an I / O request is issued when a system reset is issued using the I / O-enabled path as an argument. Can issue an I / O request using another I / O path even while issuing a system reset to the failed channel and executing countermeasure processing.

〔Example〕

FIG. 3 shows a functional block diagram of one embodiment of the present invention.
However, the hatched boxes indicate different parts from the conventional technology.

FIG. 3 is substantially the same as FIG. 6, and therefore, a description will be given of a modified example of the present invention as compared with the prior art.

Numeral 45 is a setting section for the display of the in-progress countermeasure and the path mask of the fault occurrence, which is set when a channel fault occurs. Numeral 53 denotes a reset unit for resetting during failure countermeasures and for resetting a faulty path mask, and is performed when the fault is cleared. 56 is SMPR
A command addition unit that returns data to the channel for which input / output has been requested. Reference numeral 58 denotes a processing unit that performs an exclusive OR operation with the faulty path mask, and sets a path that can be input / output requested during the fault. A failure countermeasure processing display 71 indicates that a channel failure has occurred. Reference numeral 72 denotes a failure path mask, which indicates a path in which a failure has occurred.

Note that, in FIG. 3, the same components as those in FIG. 6 are shown below. For example, the channel failure detection unit 37 in FIG. 3 is the same as the channel failure detection unit 101 in FIG. For simplicity, 37 and 101 are the same. Similarly, 38 and 102, 39 and 103, 4
4 and 108, 46 and 109, 47 and 110, 48 and 114, 49 and 115, 50 and 111, 51
And 112,52 and 113,54 and 116,55 and 117,57 and 118,59 and 119,61
121,60 and 120,62 and 122,63 and 123,64 and 124,65 and 125,66 and 12
6,68 and 128, 69 and 129, 73 and 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 have the same function.

Here, a description will be given mainly of a portion improved by the present invention with reference to the hardware configuration diagram of FIG.

Note that, in FIG. 3, there is the same processing as that in FIG. 2, so that it is described below to avoid confusion. The channel failure countermeasure processing unit 40 of FIG. 3 and the channel failure countermeasure processing unit 80 of FIG.
Are the same. For short, 40 and 80 are written as the same. Similarly, 41 and 81, 42 and 82, and 43 and 83 are the same.

In the configuration of FIG. 3, the channel failure countermeasure processing unit 40 first waits in a wait queue so that all the input / output devices 90, 91, and 92 under the failure channel 84 will not execute even if they receive a general input / output request thereafter. The quiesce part 44 of the general input / output request
Then, the input / output request is quiesced. Next, for all the devices under the failure channel, the display for failure recovery and the setting of the failure path mask 45 are set for the display for failure recovery and the failure path masks (42, 71, 72). Next, the executing input / output request collection unit 46 stops the input / output request for the currently executing input / output device. The I / O request that was being executed will be forcibly terminated and will wait for a retry in the form of a pseudo I / O error. Next, the reserved path switching unit 47 transfers the reserved information 66 held by the input / output device to another input / output path.

When the above preparation is completed, it is possible to execute an input / output request from an online path other than the failed channel path. This is because the range affected by the system reset is the channel path of all the input / output devices under the failure channel and the subordinate channels. Here, while a system reset is issued to the failed channel 84, an I / O request is executed to the I / O device 90 under the failed channel using the path 95 using a channel different from the failed channel. Think about it. The quiesce release unit 48 for general I / O requests releases the quiesce of I / O requests of all I / O devices under the failed channel,
The general input / output request is also made executable, and the restart request unit 49 schedules restart processing for all input / output devices. As a result, an input / output request on the waiting queue is executed.

In the input / output request processing unit 41, if the input / output device is in the process of taking measures against a failure, the SMPR command adding unit 56 adds an SMPR command to prohibit dynamic reconnection. Next, an I / O request mask that is currently considered to be able to issue an I / O request in the processing unit 58 that takes an exclusive OR with the faulty path mask (at this time, there are four I / O request masks 68 of the I / O device 90). The bits corresponding to all of the input / output paths 93, 94, 95, and 96 are turned on.) And the failure path mask set by the channel failure countermeasure processing unit 40 are exclusive-ORed, and the failure channel path is taken. Do not select.

Thus, the I / O device 90 is requested to make an I / O request to the IOP with the I / O issuance request mask (the bits corresponding to the I / O paths 93, 95, and 96 are on). In the IOP, one of the three paths is selected and activated. Also,
At the time of reconnection, the SMPR command is added, so reconnection is performed using the same path as the path that was started. Here, if the I / O request is requested through three paths and the SMPR command is not added, the I / O controller 88 still believes that the path group is formed by the four paths.
There is a possibility of rejoining on the failed channel path 94. In this case, if the path encounters a system reset, the completion interrupt of the input / output request is lost.

As described above, the system reset issuance unit 50 asynchronously issues a system reset to the failed channel in a state where execution of the general input / output request is permitted to all devices under the failed channel. If the system reset is successful, the failed channel path that has been withdrawn by issuing the system reset in the path group reestablishment unit 52 has returned to a healthy path.
ESTABLISH I / O is issued by specifying this faulty path 94 so that it can be included in the path group again. After the reestablishment of the path group is completed, the display during the troubleshooting is canceled by the resetting unit 53 and the faulty path mask is reset. Hereafter, the I / O request processing unit
No. 41 performs normal input / output processing since the failure countermeasure processing is not being performed.

 Next, FIG. 4 shows a detailed view of the device information storage unit.

In FIG. 4, reference numeral 151 denotes a device information storage unit relating to information of one input / output device. In the following, regarding one I / O device information, 1
Reference numeral 52 denotes a failure countermeasurement processing display section. When this is on, the fault is being processed, and when it is off, no failure countermeasure is being taken. Reference numeral 153 denotes an I / O request issuance permission mask, which indicates a channel path that is not allowed to issue an I / O request. 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 the path cannot be issued. Reference numeral 154 denotes a failure occurrence mask, which indicates a failed channel path. Reference numeral 155 denotes a mask A, which is an example of an input / output request issuance permission mask. Reference numeral 156 denotes a mask B, which is an example of a failure occurrence mask. By taking the exclusive OR of the mask A and the mask B, it is possible to control not to issue an input / output request from the failed channel path.

Finally, the difference between the processing of the present invention and the processing of the prior art is shown by a time chart.

FIG. 5 is a time chart of the failure processing. Fifth
The upper part of the figure is a time chart of the failure processing according to the conventional technique, and the lower part is a time chart of the failure processing according to the present invention.

The fault processing diagram of FIG. 5 shows the operation of the channels 84 and 85 in the configuration diagram of FIG. The time chart shows processing from the occurrence of a failure in the channel 84 to the recovery from the failure. The details of each processing step have already been described, and a description thereof will be omitted. As shown in the figure,
The processing is suspended from the occurrence of a failure in the channel 84 until the failure is recovered and restarted. However, in the present invention, the I / O processing unit issues an I / O request and the After the start of the input / output processing, the normal processing can be resumed. That is, in the present invention, the resumption of the input / output processing is remarkably quicker than in the related art.

〔The invention's effect〕

As described above, according to the present invention, even if a channel failure occurs, it is possible to execute an I / O request to an I / O device under the failed channel even during the recovery process of the failed channel. This greatly contributes to speeding up the input / output processing in the system.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of the present invention, FIG. 2 is a hardware configuration diagram of input / output control, FIG. 3 is a functional block diagram of an embodiment of the present invention, and FIG. FIG. 5 is a detailed diagram of a device information storage unit. FIG. 5 is a time chart of a failure process. FIG. 6 is a functional block diagram of a conventional technique. In FIG. 1, 1 is a channel failure countermeasure processing unit, 2 is an I / O request processing unit, 3 is a device information storage unit, 4 is an I / O request issuance permission mask, and 5 is a failure occurrence mask. , 6 is a data processing device, 7 is a channel, 8 is an input / output device, 9 is an input / output path, and 10 is an I / O processor (IOP).
And 11 is a channel failure detection unit.

Claims (1)

(57) [Claims] An input / output request processing unit for executing an input / output request; a plurality of input / output devices; a channel for controlling the input / output device; An input / output path (9) provided for each of the input / output device (8) and the channel (7); a device information storage unit (3) for storing paths for which the input / output requests are allowed and disabled; In a computer system having an I / O processor (10) for selecting which I / O path to use for an I / O request, a channel failure detector (11) for detecting a channel failure
And a channel failure countermeasure processing unit (1) for analyzing channel faults and performing countermeasure processing. When an input / output fault occurs, the channel fault detection unit (1) is used.
1) detects the failure and notifies the channel failure countermeasure processing unit (1) of the failure. The channel failure countermeasure processing unit (1) analyzes the channel failure and identifies the failed input / output path as described above. The input / output request processing unit (2) receives the input / output request from the user program, and continues the input / output processing by referring to the device information storage unit (3). At the same time, the channel failure countermeasure processing unit (1) interrupts the faulty path, releases the I / O command on the failed path, executes a restart request process, and stores the recovered path in the device information storage unit (3). ), The input / output processing instruction is continued.