JPH10275116A - Computer system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow the computer system to operate adaptively to a fault state without stopping the computer system even if a fault occurs in an input/output device by determining whether or not an input/output request of the input/output device is outputted by referring to a device management table. SOLUTION: A RAS process part 2 receives error detection signal from input/output devices 6a to 6c and performs necessary processes. The device management table 3 holds the states of the input/output devices 6a to 6c which can be changed according to a signal from the RAS process part 2. Device drivers 4a to 4c determine whether or not input/output requests are outputted according to the states of the input/output devices 6a to 6c by referring to the device management table 3. Consequently, even if a fault occurs in one of the input/output devices 6a to 6c, the computer system 1 can be placed in operation according to the fault state without being stopped.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a computer system having an input / output device, and more particularly to a computer system having a processing mechanism when a failure occurs in an input / output device.

[0002]

2. Description of the Related Art Conventionally, when a computer system detects the occurrence of a failure in an input / output device, the computer system specifies the cause and performs a corresponding process. As this processing method, the following two methods are mainly used.

As a first processing method, for example, “PCI
SYSTEM ARCHITECTURE (PCI System Architecture) "(February 1995: Addi)
son-Wesley Publishing Com
pany (Addison Wesley Publishing Company) ISBN 0-
201-40993-3) Chapter 10 (first
Chapter 0) "Error Detection and H
and (error detection and processing).

FIG. 18 is a block diagram showing a configuration of a method of processing an error notification from an input / output device as disclosed in this document. In FIG. 18, reference numeral 181 denotes a computer system;
185 is a PCI bus of the computer system 181, and 6d and 6e are input / output devices such as a hard disk and a display connected to the PCI bus 185. Computer system 1
81 has an operation system (hereinafter referred to as OS) 184 that detects an interrupt signal for notifying an error from the input / output devices 6d and 6e and performs a corresponding process.

Next, the operation will be described. According to this method, when an error interrupt signal is input from the input / output device 6d or 6e connected to the PCI bus 185, the OS 184 specifies in each computer system 181 to cope with the error interrupt signal. The specified processing is performed. In the specification of the PCI bus of this system, unified processing for an error interrupt via the PCI bus 185 is not specified, but is left to the predetermined processing for each computer system 181. In some cases, the system is stopped.

As a second processing method when a failure occurs in an input / output device, for example, there is a method described in Japanese Patent Application Laid-Open No. 8-171520. FIG. 19 is a flowchart showing the second processing method.

According to this method, an application program is started (step A1), and it is determined whether the device is an I / O device or a memory, or to which address the package number corresponds to the I / O device or the memory. The indicated parameters are prepared (step A2) and passed to the common subroutine (step A3). The common subroutine stores the passed parameters in the failure index table and activates a function for accessing the I / O device and the memory (step A4). The function goes to access the corresponding I / O device or memory (step A5), but if there is a failure in the accessed package, the NMI (Non Maskable) is used.
Xack Tim that generates (Interrupt)
e Out occurs (step A6).

When the Xack Time Out occurs, the common subroutine detects this (step A).
7) Retrieve the failure index table held in the common subroutine (step A8), and check the Xack Time
The corresponding package number is detected from the address of the access destination when Out occurs, a common subroutine call is made using the detected package number as parameter information (step A9), and the application program is passed to the application program.

[0009]

Since the conventional computer system is configured as described above, in the first processing method, the OS 184 receives an error interrupt indicating that a failure has occurred in the input / output device 6d or 6e. In such a case, there is a problem that the system is unconditionally stopped, and subsequent processing cannot be performed.

On the other hand, in the second processing method, when a failure occurs, the failure package is detected by using the failure index table, so that the failure package can be easily confirmed, and the failure measures can be taken without stopping the computer system. May be possible. However, the failure index table must be created at the computer system design stage, and it is extremely difficult to predict the operation status of the computer system and the configuration of the input / output devices in advance and create the failure index table at the design stage. There is a problem that it is difficult. Further, when the design of the computer system is changed, the contents of the failure index table often become stale, and there is a problem that the design of the computer system is not flexibly changed.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and it is an object of the present invention to operate a computer system in response to a failure situation without stopping the computer system even if a failure occurs in an input / output device. It is another object of the present invention to provide a computer system capable of easily changing the design of the computer system by comprehensively managing the operation of the entire computer system.

Another object of the present invention is to provide a computer system capable of easily configuring a device driver.

Further, according to the present invention, when a failure occurs in an input / output device, it is possible to prevent a process waiting for a result of an input / output request from continuing to wait for a response forever,
It is an object of the present invention to obtain a computer system capable of continuing the subsequent processing.

Another object of the present invention is to provide a computer system which can prevent a failure from spreading to other parts of the computer system when a failure occurs in an input / output device.

A further object of the present invention is to provide a computer system which can avoid an input / output request for a function having a high possibility of occurrence of a failure in advance and has a high operation stability.

Further, according to the present invention, it is possible to reconfirm the normal operation of the input / output device having a high risk of occurrence of a failure, and to continue the operation of the computer system stably without unnecessarily deteriorating the functions. The purpose is to obtain.

Another object of the present invention is to provide a computer system capable of preventing a failure from spreading to other parts of the computer system when a failure occurs in a PCI-PCI bridge.

Still another object of the present invention is to provide a computer system capable of executing an optimal method for keeping the computer system operating after a failure occurs in an input / output device.

Still another object of the present invention is to provide a computer system which can easily access information for management processing, create and maintain a program.

A further object of the present invention is to provide a computer system capable of performing a reset process in a format according to a target of the reset process and performing the reset process without giving a problem to the operation of the entire computer system. .

[0021]

According to a first aspect of the present invention, there is provided a computer system which receives an error detection signal from an input / output device and performs necessary processing, and a signal from the RAS processing unit. Device management table that holds the status of each input / output device that can be changed according to the status of the input / output device, and a device driver that determines whether to output an input / output request according to the status of the input / output device with reference to the device management table It is provided with.

In the computer system according to the present invention, a column for recording information for identifying a process issuing a request to the input / output device is provided in the device management table.

According to a third aspect of the present invention, there is provided a computer system provided with a device use table which holds a relation between a specific input / output device and a process which operates while using the input / output device.

According to a fourth aspect of the present invention, there is provided a computer system in which a column for recording information of a physical connection position of each input / output device is provided in a device management table.

According to a fifth aspect of the present invention, in the computer system, the RAS processing unit has a function of confirming whether or not another function at the same physical connection position as the failed device operates normally. Have

According to a sixth aspect of the present invention, there is provided a computer system including a bus configuration table for holding configuration information indicating a relationship between a plurality of buses of the computer system.

According to a seventh aspect of the present invention, there is provided a computer system in which a field for recording information for designating a process to be executed when a failure occurs in an input / output device is provided in a device management table.

[0028] In the computer system according to the present invention, the input / output devices connected to the device management table in a data format having a tree structure according to a configuration of a PCI bus connected to the computer system. Is stored.

In the computer system according to the ninth aspect of the present invention, the target of the reset processing is an input / output device or a PCI-PCI
It is provided with a discriminating means for discriminating a bridge, and a reset means for performing a reset process in a format according to the discrimination result of the discriminating means.

[0030]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. FIG. 1 is a configuration diagram showing a computer system according to Embodiment 1 of the present invention. In the figure, reference numeral 1 denotes a computer system, 6a, 6b, and 6c denote input / output devices connected to the computer system 1, and 6a denotes a hard disk ( Hereinafter, I / O device A), 6b are communication terminals (hereinafter I / O device B), and 6c is a display (hereinafter I / O device C). Hereinafter, the I / O devices A6a, B6b, C6c are collectively referred to as the input / output device 6.

Reference numeral 2 denotes an RAS processing unit which receives an error interrupt signal from the input / output device 6 and performs processing necessary for maintaining reliability, availability and maintainability, and 3 stores the state of the input / output device 6. In the device management table, information 3a for specifying the input / output device 6 and its status 3b are held.

Further, 4a, 4b, and 4c are I /
I / O for operating O devices A6a, B6b, C6c
A device driver for device A, a device driver for I / O device B, and a device driver for I / O device C. The device driver 4a for the I / O device A, the I / O device
The device driver 4b for the device B and the device driver 4c for the I / O device C are generically referred to as a device driver 4. Reference numeral 5 denotes a PCI bus that connects the RAS processing unit 2 and the input / output device 6.

The device management table 3 is initialized when the computer system 1 is started up in the process of configuring the PCI bus 5 in accordance with the PCI regulations. Since the information included in the device management table 3 at the time of this initialization can be obtained by a well-known ordinary technique, the description of the configuration procedure of the device management table 3 is omitted, and the device management table 3 is prepared in advance. .

Next, the operation will be described with reference to the flowchart of FIG. 2 showing the processing of the RAS processing section 2 and the flowchart of FIG. 3 showing the processing of the device driver 4.

When an error interrupt of a failure occurs from any of the I / O devices A6a, B6b, C6c, the interrupt is notified to the RAS processing unit 2 via the PCI bus 5. In response to the notification, the RAS processing unit 2 first specifies a faulty device for specifying which input / output device 6 has failed according to the flowchart shown in FIG. 2 (step ST1), and can identify the faulty device. It is determined whether or not it has been performed (step ST2).

If the faulty device cannot be identified, the computer system administrator is notified that the faulty device cannot be identified (step ST4), and the processing operation is terminated. This notification to the administrator is sent to the I / O device C6c.
This is done by storing the fact that the faulty device could not be identified by the display on the top, the recording in the I / O device A6a, and the like. If the faulty device can be identified, the status 3b of the corresponding I / O device in the device management table 3 is set to N
G (step ST3), and the processing operation ends. In the example of the device management table 3 in FIG. 1, a failure has occurred in the I / O device B6b, and NG is recorded.

On the other hand, the device driver 4 operates according to the flowchart of FIG. 3 asynchronously with the operation of the RAS processing unit 2. The device driver 4 that has received the input / output request to the input / output device 6 (step ST5) refers to the state 3b of the device management table 3 corresponding to the input / output device 6 to be accessed (step ST6), and
b is confirmed (step ST7). At this time, if the state 3b of the input / output device 6 is OK (that is, a state in which access to the input / output device is possible), the input / output processing to the input / output device 6 is directly performed (step ST8), and
In other cases (that is, in a state where access to the input / output device 6 is not possible), the input / output processing is interrupted (step ST9) to avoid actual access to the input / output device 6 in which a failure has occurred.

As described above, according to the first embodiment, only the failed input / output device 6 can be separated from the computer system 1, and the effect that the computer system 1 can be degraded is obtained. Can be In addition, the RAS processing unit 2 collectively performs processing when a failure occurs in the input / output device 6 and holds the state in the device management table 3 so that the overall operation of the computer system 1 can be managed in a comprehensive manner. Is obtained. Further, since it is not necessary to add a failure handling function to the device driver 4 alone, an effect that the configuration of the device driver 4 is simplified can be obtained.

Embodiment 2 FIG. 4 is a configuration diagram showing a computer system according to Embodiment 2 of the present invention. In the figure, the same components as those of Embodiment 1 shown in FIG. Is omitted.

In the second embodiment, the device management table 3A includes, in addition to the information 3a for specifying the input / output device 6 and its status 3b, the application that is currently issuing an access request to the input / output device 6. A number 3c for identifying a process such as software is recorded. The configuration other than this is the same as that of the first embodiment.

Next, regarding the operation, the flowchart of FIG.
This will be described with reference to the flowchart of FIG. In the flowchart of FIG. 5 and the flowchart of FIG. 6, steps that perform the same processing as the steps in the flowchart of FIG. 2 and the flowchart of FIG. 3 are assigned the same step numbers, and detailed description thereof will be omitted. .

The device driver 4 accepts an input / output request as shown in the flowchart of FIG.
T5) With reference to the device management table 3A (step ST6), it is determined whether or not each of the input / output devices 6 can be accessed (step ST7).

If the access is possible, the identification number of the process that issued the access request is recorded as the number 3c for identifying the process that issued the access request in the device management table 3 (step ST11).
An input / output process is performed (step ST8). Thereafter, when the processing is completed normally, the recorded process identification number 3c is deleted from the device management table 3 (step ST12).

On the other hand, if the status 3b of the device in the device management table 3 is other than OK (that is, an inaccessible status), the input / output process is interrupted (step ST9), and an error interrupt from the faulty device is processed by RAS. Notify part 2.

Further, the RAS processing unit 2 which has received the error interrupt notification of the occurrence of the fault from the input / output device 6 specifies the faulty device in accordance with the flowchart shown in FIG. 5 (step ST1). Step S
T2) After changing the status 3b of the corresponding device in the device management table 3A to NG (step ST3), access from the part of the device management table 3A where the number 3c identifying the process issuing the access request is recorded. The process that issued the request is detected, and the process is notified that the input / output processing has abnormally ended (step S
T10).

As described above, according to the second embodiment, when a failure occurs in the input / output device 6 of the computer system 1, the process requesting access to the failed input / output device is performed. Thus, it is possible to notify that the request does not end normally.

Embodiment 3 FIG. 7 is a configuration diagram showing a computer system according to Embodiment 3 of the present invention. In the figure, the same components as those of Embodiment 1 of FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. I do.

In FIG. 7, reference numeral 7 denotes a specific input / output device 7a.
Alternatively, it is a device use table that holds a relationship between 7b and a process name 7c that keeps using it regularly.
The information held in the device usage table 7 will be described again with a specific example. For example, assuming a computer using UNIX as the OS, as shown in the first row of the device usage table 7 in FIG. 7, the syncer command is used as a daemon process for repeatedly accessing the hard disk as the I / O device A6a. And sys
This indicates that there is a logd command. Both processes are started while the computer system is operating normally, and the I / O is performed at a certain period or by event driving.
O device A6a is continuously accessed, and I
An abnormality in the / O device A6a has a significant effect on the operation of these processes.

It is assumed that the device usage table 7 is defined and prepared in advance by the designer or administrator of the computer system 1, and it is assumed that the device usage table 7 already exists. The other components in FIG. 7 are the same as the components of the first embodiment. Also, device driver 4
Is abbreviated.

Next, regarding the operation, the RAS processing unit 2 shown in FIG.
A description will be given with reference to a flowchart showing the processing.
The RAS processing unit 2 that has received the error interrupt signal indicating that a failure has occurred from the input / output device 6 identifies the failure generating device according to the flowchart shown in FIG. 8 (step ST1).
If it can be identified (step ST2), after changing the status 3b of the corresponding device in the device management table 3 to NG (step ST3), the faulty input / output device 7a or 7b exists in the device usage table 7. It is confirmed whether or not this is the case (step ST13). As a result of the confirmation, if it does not exist, the process is terminated as it is, but if it does exist, the process name 7c that regularly uses the input / output device 6 in the device use table 7 is obtained, and the process is started. If so, an instruction is issued to stop the operation (step ST14), and the operation of the RAS processing unit 2 ends.

As described above, according to the third embodiment, the process of continuously issuing an input / output request to a specific input / output device 6 of the computer system 1B is a prerequisite for the normal operation of the process. When a failure occurs in the input / output device 6, it is possible to avoid in advance making an I / O request that is apparently abnormally terminated, and to prevent the failure from spreading to other parts of the computer system 1B. The effect that it can be obtained is obtained.

Embodiment 4 FIG. FIG. 9 is a configuration diagram showing a computer system according to Embodiment 4 of the present invention. In the figure, the same components as those of Embodiment 1 of FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. I do. The notation of the device driver 4 is omitted.

In FIG. 9, the device management table 3B according to the fourth embodiment includes an I / O device C6c and an I / O device A.
8a, information 3a for specifying the I / O device B 8b and its state 3b, an I / O device C6c, an I / O device A
8a, the bus number and the device number 3d of the input / output device of the I / O device B8b are recorded. 8 is an input / output device and 1
One input / output device has the functions of two I / O devices A8a and I / O device B8b. According to the PCI regulations, it is recognized that a plurality of (two for convenience) input / output devices having different functions are mounted in one input / output device, and these have the same bus number and device number.

The bus number / device number 3d of the input / output device 8 is determined at the time of the bus configuration at the time of starting the computer system 1C and can be obtained in this process. The description of the configuration procedure is omitted, and the device management table 3B is prepared. The other components in FIG. 9 are the same as the components of the first embodiment.

Next, the operation will be described with reference to a flowchart showing the processing of the RAS processing section 2 in FIG. The RAS processing unit 2, which has received the error interrupt signal of the occurrence of the failure from the I / O device C6c or the I / O device A8a or the I / O device B8b, specifies the failure occurrence device according to the flowchart shown in FIG.
1) If it can be specified (step ST2), the state 3b of the corresponding input / output device in the device management table 3B is changed to NG (step ST3), and then the bus number / device number 3d of the device management table 3B is changed. It is checked whether there is an entry having the same bus number and device number as the failed device by referring to (step ST).
15). As a result of the confirmation, if the device does not exist, the process is terminated. If the device is present, the status 3b of the device is also changed to NG (step ST16), and the process is terminated.

As described above, according to the fourth embodiment, even when one input / output device connected to the computer system 1C has a plurality of different functions, a failure is found in any of the functions. In this case, other functions on the same I / O device can also be prevented from accepting I / O requests, so that it appears that the system is operating normally without a failure being detected, but a failure occurs. An input / output request for a function having a high possibility of being obtained can be avoided in advance, and the effect of increasing the stability of the operation of the computer system 1 can be obtained.

Embodiment 5 FIG. 11 is a flowchart showing processing of the RAS processing unit 2 of the computer system according to Embodiment 5 of the present invention. The configuration of this embodiment is as follows.
The configuration is the same as that of the fourth embodiment shown in FIG.

RAS processing unit 2 which has received an error interrupt signal indicating that a failure has occurred from I / O device A8a or B8b
In accordance with the flowchart shown in FIG. 11, the faulty device is specified (step ST1). If the faulty device can be specified (step ST2), the status 3b of the corresponding device in the device management table 3B is changed to NG (step ST2). ST
3) Later, it is checked whether there is an entry having the same bus number and device number as the failed device by referring to the bus number / device number 3d of the device management table 3B (step ST15). As a result of the check, if the device does not exist, the process is terminated as it is. If the device exists, it is checked whether the device is operating normally (step ST17). As a result of the check, if it is confirmed that the device is normal (step ST18), the process is terminated as it is.
Is changed to NG (step ST19), and the process ends.

As described above, according to the fifth embodiment, by confirming whether or not normal operation is possible, it is possible to reconfirm the normal operation of a device having a high risk of occurrence of a failure, and to reduce the function more than necessary. The advantage is that the operation of the computer system can be stably continued without any problem.

Embodiment 6 FIG. FIG. 12 is a configuration diagram of a computer system according to Embodiment 6 of the present invention. In the figure, the same components as those of Embodiment 1 of FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. . The notation of the device driver 4 is omitted.

The device management table 3C according to the sixth embodiment records an address 3e for designating post-processing to be executed when a failure occurs in each input / output device 6. The address 3e for designating post-processing to be executed when this failure occurs can be freely changed by the device driver 4 or another program. The other components in the figure are the same as the components of the first embodiment.

Next, the operation will be described with reference to a flowchart showing the processing of the RAS processing unit 2 in FIG. The RAS processing unit 2, which has received the error interrupt signal indicating that a failure has occurred from the input / output device 6, specifies the failure generating device according to the flowchart shown in FIG.
1) If it can be specified (step ST2), after the status 3b of the corresponding device in the device management table 3C is changed to NG (step ST3), it is executed when a failure occurs in the device in the device management table 3C. It is confirmed whether the address 3e for specifying the post-processing to be performed is specified (step ST20). As a result of confirmation, address 3
If e does not exist, the process is terminated as it is, but if it does exist, the corresponding process specified in the post-processing address 3e to be executed is executed (step ST21), and the process ends.

As described above, according to the sixth embodiment, when a failure occurs in the input / output device 6 of the computer system 1D, post-processing required for each failed device is performed for each device. Since the definition can be made easily and freely, an effect is obtained that an optimal method can be executed to keep the computer system operating after a failure occurs in the input / output device 6.

Embodiment 7 FIG. FIG. 14 is a configuration diagram of a computer system according to Embodiment 7 of the present invention. In the figure, the same components as those of Embodiment 2 of FIG. 9 are denoted by the same reference numerals, and description thereof is omitted. . The notation of the device driver 4 is omitted.

In FIG. 14, reference numeral 9 denotes a computer system 1E.
In the upper host bus, 10a is a host bridge A for connecting the host bus 9 and the PCI bus 5a, and 10b is a host bridge B for connecting the host bus 9 and the PCI bus 5b. Note that host bridges A10a, B
10b is generally referred to as a host bridge 10. By the way, in the above-mentioned embodiment, since it is not necessary to consider the host bus 9 and the host bridge 10, illustration and description are omitted.

11a is a PCI bus 5b and a PCI bus 5c.
And the PCI-PCI bridges a and 11b connecting the
PCI-PC connecting I bus 5b and PCI bus 5d
The I bridges b and 11c are connected to the PCI bus 5d and the PCI bus 5
e to a PCI-PCI bridge c. The PCI-PCI bridges a11a, b11b, c1
1c are collectively referred to as a PCI-PCI bridge 11,
The CI buses 5a, 5b, 5c, and 5d are collectively referred to as a PCI bus 5. Reference numeral 12 denotes a bus configuration table in which connection relationships between the PCI buses 5 are recorded.
The device number 12a for identifying the CI bridge 11, the number 12b of the primary PCI bus (closer to the computer main body) of the PCI-PCI bridge 11, the number 12c of the secondary PCI bus (farther from the computer main body), and the PCI -An item indicating the state 12d of the PCI bridge 11 itself.

The above bus configuration conforms to the PCI rules. The numbering of the PCI bus is set at the time of execution of the configuration, and the numbers of the primary bus and the secondary bus are also available at this time. Therefore, description of the configuration procedure of the bus configuration table 12 is omitted, and
It is assumed that the bus configuration table 12 is prepared. The other components in the figure are the same as those of the second embodiment.

Next, the operation will be described with reference to the flowchart showing the processing of the RAS processing section 2 in FIG. I / O device 6 or PCI-PCI bridge 11
RAS that has received an error interrupt signal indicating that a fault has occurred
In the processing unit 2, according to the flowchart shown in FIG.
The faulty device is specified (step ST1), and if the faulty device is specified (step ST2), the device is set to the input / output device
Or the PCI-PCI bridge 11 is checked (step ST22). As a result of the check, if the input / output device 6 is found, the status 3b of the corresponding device in the device management table 3B is changed to NG (step ST3), and the process ends. In the case of the PCI-PCI bridge 11, the status 12d of the device corresponding to the failed PCI-PCI bridge 11 in the bus configuration table 12 is changed to NG,
Further, the bus number 12c existing on the secondary side is checked.

Based on the bus number 12c, it is determined whether each input / output device 6 having an entry in the device management table 3B is connected to the secondary side of the PCI-PCI bridge 11 in which the fault has occurred. Number 12a
Is checked with reference to the bus number / device number 3d in which the status 3b of the input / output device 6 is N.
G is recorded (step ST23), and the processing ends.

As described above, according to the seventh embodiment, in a computer system having a plurality of PCI buses, when a failure occurs in the PCI-PCI bridge connecting the PCI buses, the PCI-PCI bridge This makes it possible to avoid in advance making an input / output request to an input / output device connected before the bridge, and to prevent the failure from spreading to other parts of the computer system.

Embodiment 8 FIG. FIG. 16 is a block diagram showing a computer system according to the eighth embodiment of the present invention. In the figure, the same components as those of the seventh embodiment shown in FIG. . The notation of the device driver 4 is omitted.

In FIG. 16, reference numeral 13 denotes a device management table 3B and a bus configuration table 12 according to the seventh embodiment.
3 is a device management table tree (device management table) created based on the information recorded in (1) and (2). The device management table tree 13 includes a host bus 9, host bridges A10a and B10b, and a PCI-PCI bridge a11.
a, b11b, c11c, I / O devices A8a, B6b, etc. are respectively stored in the memory as a set of data (hereinafter, each set of data is referred to as a node), and data indicating a connection relationship between the nodes is stored in the memory. This is added to a block of data. This device management table tree 13
Since the information necessary to configure the information is information that can be obtained in the process of PCI configuration, the configuration procedure is not described here. The other components in the figure are the same as the components of the seventh embodiment.

Next, the operation will be described with reference to FIG.
This will be described with reference to the flowchart of FIG. In the RAS processing unit 2 which has received an error interrupt signal indicating that a failure has occurred from the input / output device 6 or the PCI-PCI bridge 11,
According to the flowchart shown in FIG. 15, the faulty device is specified (step ST1). If the faulty device can be specified (step ST2), the device is determined by the device management table tree 13 according to the bus number and device number of the input / output device 6.
To the node of the corresponding device. It is checked whether or not the device in which the failure has occurred in this node is the PCI-PCI bridge 11 (step ST22). As a result of the check, if the input / output device 6 is found, the status of the corresponding device in the device management table tree 13 is changed to NG (step S
T3) and ends. In the case of a PCI-PCI bridge, the status of the corresponding device in the device management table tree 13 is set to NG, and further, all the child nodes ahead of the node are searched, and the input / output recorded in each node is The state of the device 6 is recorded as NG (step ST2).
3) End. As a result, input / output requests to all devices on the secondary side of the failed PCI bridge can be stopped.

As described above, according to the eighth embodiment, in a computer system having a plurality of PCI buses, information for managing each input / output device 6 is in a format corresponding to the configuration state of the bus. As a result, even in a computer system having a complicated bus configuration, information access becomes easy due to management processing, the possibility of errors being included in a program is reduced, and a specific bus and an input / output device below it are used. An effect is obtained that processing such as designating a part on the bus configuration can be performed quickly.

Embodiment 9 FIG. FIG. 17 is a flowchart showing a reset processing operation of the computer system according to the ninth embodiment of the present invention. The configuration of the ninth embodiment is shown in FIG.
The configuration is the same as that of the eighth embodiment shown in FIG.

A certain device (the input / output device 6 is also a PCI-PC
A failure occurs in the I-bridge 11), and an instruction from the administrator or an instruction from the RAS processing unit 2 (in this case, a reset process when a failure occurs is specified in the device management table tree 13 in advance). When the reset processing is requested to the faulty device (step ST24), the device management table tree 13 is searched for the node of the corresponding device based on the bus number and the device number of the device whose reset is requested.

Next, it is checked whether the device which has received the reset request in this node is the input / output device 6 or the PCI-PCI bridge 11 (step ST25). As a result of examination, P
In the case of the CI-PCI bridge 11, the state of all devices located below the node, that is, on the secondary side, is set to a reset waiting state so as not to accept an input / output request (step ST26).

After this processing is completed, the state of the apparatus which has received the reset request is changed to the state of executing reset (step ST27), that is, the new input / output request is not accepted and the reset processing is executed. (Step ST28). After the reset process is completed, the result is recorded in the device management table tree 13 (step ST29).

Next, the type of the device to be reset is checked (step ST 30). If the device is the input / output device 6, the process is terminated, the PCI-PCI bridge 11, and the reset process is performed normally. If the processing has been completed, the state of the device management table 13 relating to all the devices under that node is restored (step ST31), and the processing ends.

When the PCI-PCI bridge 11 is to be reset, resetting of all devices connected to the secondary side of the device can be performed, or the stability of the entire computer system can be improved. It is effective to secure Further, the operation of the reset process of the ninth embodiment can be applied to an input / output management system having a device management table 3 and a bus configuration table 12 as shown in FIG.

As described above, according to the ninth embodiment, in a computer system having a plurality of PCI buses, the input / output device for which reset processing is to be executed is the input / output device at the end of the bus. However, even in the case of a PCI-PCI bridge which is an intermediate component of the bus, an effect that reset processing can be performed without giving a problem to the operation of the entire computer system can be obtained.

[0082]

As described above, according to the first aspect of the present invention, the computer system is provided with the device management table which can be referred to from the device driver and can be changed by the error signal interrupt processing. When a failure occurs in an input / output device while the computer system is operating, a degraded operation in which only the failed device can be separated can be performed, and processing at the time of occurrence of a failure in the input / output device is collectively performed by a RAS processing unit. By maintaining the state in the device management table, the operation of the entire computer system can be managed in an integrated manner. Further, since it is not necessary to add a failure handling function to the device driver alone, the configuration of the device driver is reduced. This has the effect of being easier.

According to the second aspect of the present invention, since the device management table is provided with a column for recording information for identifying the process which has issued the access request, the input / output device of the computer system is provided. When a failure occurs, it is possible to notify the process that has requested access to the failed device that the request is not completed normally, and the process that is waiting for the result of the I / O request can be notified. There is an effect that it is possible to avoid waiting for the response indefinitely, and to continue the subsequent processing.

According to the third aspect of the present invention, a device use table is provided to hold a relationship between a specific input / output device and a process name that keeps using the input / output device repeatedly. A process that continues to issue I / O requests to a specific I / O device issues an I / O request that is apparently abnormally terminated if a failure occurs in the I / O device that is the prerequisite for the process to operate normally. This prevents the failure from spreading to other parts of the computer system.

According to the fourth aspect of the present invention, since the device management table is provided with a column for recording information on the physical connection position of the input / output device, one input / output device connected to the computer system is provided. Even when a device has a plurality of different functions, if a failure is found in any one of the functions, it is possible to prevent input / output requests for other functions on the same input / output device as well, It is possible to avoid input / output requests for functions that seem to be operating normally without detection of a failure but are likely to cause a failure in advance, thereby improving the stability of operation of the computer system. There is.

According to the fifth aspect of the present invention, the RAS processing unit is provided with a function of confirming whether or not another function at the same physical connection position as the failed device operates normally. Since it is configured, 1 connected to the computer system
Even if one I / O device has a plurality of different functions, if a failure is found in any of the functions, it is possible to confirm whether the other functions on the same I / O device can operate normally. Thus, the normal operation of the device having a high risk of occurrence of a failure can be confirmed again, and the operation of the computer system can be stably continued without deteriorating the function more than necessary.

According to the sixth aspect of the present invention, since the bus configuration table for holding the configuration information indicating the relationship between each of the plurality of buses is provided, the computer system having the plurality of PCI buses has When a failure occurs in the PCI-PCI bridge connecting the buses,
It is possible to avoid in advance making an I / O request to an I / O device connected earlier than the PCI-PCI bridge, and to prevent a failure from spreading to other parts of the computer system. is there.

According to the seventh aspect of the present invention, the device management table is provided with a column for recording information for designating processing to be performed when a failure occurs in the input / output device. When an I / O device fails, post-processing required for each failed I / O device can be defined easily and freely for each I / O device. There is an effect that an optimum method for keeping the computer system operating can be executed later.

According to the invention of claim 8, the device management table is configured to store the information of the input / output device in a data format having a tree structure according to the configuration of the bus. In a computer system equipped with, the information for managing devices is in a format that is aware of the bus configuration state, so that even computer systems with complicated bus configurations can easily access information for management processing. As a result, there is an effect that the possibility that an error is included in the program is reduced, and processing such as designating a part of a bus configuration such as a specific bus and an input / output device thereunder is quickly performed.

According to the ninth aspect of the present invention, since the reset target is determined and the reset processing is performed in a format according to the result, the reset processing is performed in a computer system having a plurality of PCI buses. Even if the input / output device to be executed is the input / output device at the end of the PCI bus, the PCI-PCI which is an intermediate component of the PCI bus is used.
Even with a bridge, there is an effect that reset processing can be performed without giving a problem to the operation of the entire computer system.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a computer system according to Embodiment 1 of the present invention.

FIG. 2 is a flowchart showing an operation of the RAS processing unit according to the first embodiment of the present invention.

FIG. 3 is a flowchart showing an operation of the device driver according to the first embodiment of the present invention.

FIG. 4 is a configuration diagram showing a computer system according to a second embodiment of the present invention.

FIG. 5 is a flowchart showing an operation of the RAS processing unit according to the second embodiment of the present invention.

FIG. 6 is a flowchart showing an operation of the device driver according to the second embodiment of the present invention.

FIG. 7 is a configuration diagram showing a computer system according to Embodiment 3 of the present invention.

FIG. 8 is a flowchart illustrating an operation of a RAS processing unit according to Embodiment 3 of the present invention.

FIG. 9 is a configuration diagram showing a computer system according to Embodiment 4 of the present invention.

FIG. 10 is a flowchart showing an operation of a RAS processing unit according to Embodiment 4 of the present invention.

FIG. 11 is a flowchart showing an operation of the RAS processing unit according to the fifth embodiment of the present invention.

FIG. 12 is a configuration diagram showing a computer system according to Embodiment 6 of the present invention.

FIG. 13 is a flowchart showing an operation of the RAS processing unit according to the sixth embodiment of the present invention.

FIG. 14 is a configuration diagram showing a computer system according to Embodiment 7 of the present invention.

FIG. 15 is a flowchart showing an operation of the RAS processing unit according to the seventh embodiment of the present invention.

FIG. 16 is a configuration diagram showing a computer system according to Embodiment 8 of the present invention.

FIG. 17 is a flowchart showing the operation of the RAS processing unit according to Embodiment 9 of the present invention.

FIG. 18 is a configuration diagram showing a conventional computer system.

FIG. 19 is a flowchart showing the operation of another conventional computer system.

[Explanation of symbols]

1, 1A, 1B, 1C, 1D, 1E, 1F Computer system, 2 RAS processing unit, 3, 3A, 3B, 3C device management table, 4a I / O device A device driver (device driver), 4b I / O Device driver for device B (device driver), 4c I / O Device driver for device C (device driver), 6a I
I / O device A (input / output device), 6b I / O device B (input / output device), 6c I / O device C (input / output device), 7 device utilization table, 8 input / output device, 8a I / O device A (Input / output device), 8b I / O device B (input / output device), 12 bus configuration table, 13 device management table tree (device management table).

Claims (9)

[Claims] An RAS processing unit that receives an error detection signal from an input / output device and performs necessary processing, and a device that holds a state of each input / output device that can be changed according to a signal from the RAS processing unit A computer system comprising: a management table; and a device driver that determines whether to output an input / output request in accordance with a state of an input / output device with reference to the device management table. 2. The computer system according to claim 1, wherein a column for recording information for identifying a process issuing a request to the input / output device is provided in the device management table. 3. The computer system according to claim 1, further comprising a device use table for holding a relation between a specific input / output device and a process that operates while continuing to use the specific input / output device. 4. The computer system according to claim 1, wherein a column for recording information on a physical connection position of each input / output device is provided in the device management table. 5. The RAS processing unit has a function of confirming whether or not another function at the same physical connection position as the failed device operates normally. Computer system. 6. The computer system according to claim 4, further comprising a bus configuration table for holding configuration information indicating a relationship between each of a plurality of buses of the computer system. 7. The computer system according to claim 1, wherein a column for recording information for specifying a process to be executed when a failure occurs in the input / output device is provided in the device management table. 8. The device management table according to claim 1, wherein the information of the connected input / output device is stored in a data format having a tree structure according to a configuration of a PCI bus of the computer system. Computer system as described. 9. The target of reset processing is an input / output device or a PC
2. The computer system according to claim 1, further comprising: a determination unit that determines whether the bridge is an I-PCI bridge; and a reset unit that performs a reset process in a format according to a determination result of the determination unit.