JP2806321B2 - Hardware connection status monitoring method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a computer system having a hot-swappable device, and more particularly to a hardware connection state monitoring method.

[0002]

2. Description of the Related Art As disclosed in Japanese Patent Application Laid-Open No. 4-195455, in order to monitor hardware failure or configuration change in a system that can be inserted and removed during operation, all ports that can be connected on the software side are conventionally used. Needs to issue a confirmation command at regular intervals. In this case, even when monitoring is performed for a fixed time, the implementation is delayed due to the load state of the system or the device. If the latest connection status is not reflected, the system may issue an incorrect request. For example, an I / O request may be issued even though a device has been deleted. Another method using a watchdog timer is described in Japanese Patent Application Laid-Open No. 4-195455. In this technology, a CPU device is constantly monitored by hardware using a watchdog timer, and when a timeout occurs, another device is notified through an interface circuit. In the case of this method, a fault notification is realized to another CPU by providing a signal line. In the case of this technology, the notification is performed at a high speed, but the signal line is used, so that it is not suitable for a system in which the device is inserted and removed during operation.

[0003]

In the case of the prior art, there is a problem that the timing of monitoring the connection state of the hardware at regular intervals is delayed. The reason is that when another interrupt occurs, it is processed with higher priority than the management processing. For this reason, there is a problem that the frequency of issuing an I / O request is high even though the device is faulty.

An object of the present invention is not to issue a command on a software basis at regular time intervals to diagnose a connection state of a device, but when hardware is connected, the hardware itself shows itself at regular time intervals. An object of the present invention is to use hardware that includes a device that issues an interrupt, and to use the hardware to receive the interrupt so that the timing of monitoring the connection status of the hardware is not delayed.

[0005]

According to the hardware connection state monitoring method of the present invention, a hardware having a built-in device for issuing a special interrupt indicating its own presence at regular time intervals and receiving the special interrupt And a connection information management unit comprising an update / table management unit for updating a table for recording hardware connection status, and receiving a special interrupt from the hardware. The content decoding unit adds the interrupted slot number to the device information notified by the interrupt and notifies the update / table management unit, and the update / table management unit notifies the update / table management unit of information on the connection device of each slot. A connection information management table to be recorded, and a mutual information to classify each slot into a predefined group and record information on devices connected to slots belonging to each group. It has a logical table, and compares the device information notified based on the slot number notified from the content decoding unit with the device information recorded in the connection information management table to determine the validity of the connected device. If it is valid, the connection information management table and the mutual management table are updated with the notified device information, and if determined to be invalid, the device is disconnected, and the device information of the slot number of the same group as the notified slot number is updated. Is extracted from the mutual management table, the latest update time is examined, and the presence or absence of the special interrupt from hardware is determined to determine the presence / absence of a device abnormality.

According to the present invention, there is no need for the software to send a command for confirming the connection state to the hardware, and it is possible to confirm whether the device is connected only by receiving an interrupt from the hardware. It is. As a result, even when the device is checked in a state where the system load is high, the time lag is eliminated, so that the latest connection status information can be obtained as compared with the conventional method.

FIG. 6 (a) shows the flow of execution of the prior art. FIG. 6B shows a flow of execution of an embodiment of the present invention. S-1, S- to check the normal condition
2. A command is issued to the hardware in S-3, and A-
1, A-2 and A-3 receive the results from the apparatus. Both techniques reflect results at regular intervals. However, in the case where another interrupt process occurs, the delay time T occurs when the command issuance timing is a process having a high priority such as the interrupt process in the related art, and the command is issued, and is shifted to the rear by T. . If such situations overlap, a situation may occur that is significantly different from the current apparatus situation. According to the present invention, since the issuing operation is stopped, the difference in the timing of updating the connection state can be reduced accordingly. In this way, the hardware status can be reflected at a more accurate time than before, and the system is less likely to issue an erroneous request to the device.

[0008]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention. In this embodiment, the device information 30 is periodically updated.
Hardware 10 that generates an interrupt notifying 4
1, 102, and 103, and a connection information management unit 100 that receives and analyzes the state. Hardware 10
1, 102 and 103 have built-in interrupt generators 104, 105 and 106 which periodically generate hardware interrupts. Connected hardware 101, 10
2 and 103 generate interrupts W1 and W2, and notify the device information 304 to the connection information management unit 100. When the connection information management unit 100 receives the notification, the content decoding unit 200 sends the device information 304 and the notified slot position to the update / table management unit 300. The update / table management unit 300 updates the connection status. The case where the hardware 103 is not connected is shown. Since the interrupt from the hardware 103 is not notified to the connection information management unit 100, the update in the update / table management unit 300 is not performed.

FIG. 2 shows the content decoding unit 200 in FIG.
The decoding unit includes a receiving device 201, an interrupt processing table 202, and a hardware register 203.

FIG. 4 shows an interrupt process in the content decoding unit 200.

Referring to FIG. 2 and FIG.
01 and 102, the receiving device 201 receives the interrupt signals.
2, the type of the interrupt is determined, and in the case of a hardware connection state notification interrupt, the slot position of the hardware register 203 is read, and the device information 304 and the slot position are passed to the processing of the update / table management unit 300 in FIG. If the interrupt is not a hardware connection state notification interrupt, the same interrupt processing as in the related art is executed in step 406.

FIG. 3A shows the update / table management unit 30 in FIG.
FIG. 7 is a diagram showing details of a zero. The connection information management table 301 holds management information for all slots, and stores the contents of the corresponding data 303. The contents of the corresponding data 303 include the device position (slot) “locate”, the unique number “d_id” for each hardware, and the type identification code “type”.
e ”,“ time_s ”of the time stamp value at the time of interruption
The value is stored in the connection information management table 301. Each of the values is stored in the connected device, and includes “old_time” and “old_time” of the latest update time.
Is further stored in the mutual management table 302. A slot recognized as being connected forms a management group with another slot. The mutual management table 302 holds management group information. The contents of the device information 304 include “LOCATE” of the position of the device, “D_ID” of a unique number for each hardware, and “T” of the type recognition code.
YPE ”,“ TIME ”of the time stamp value at the time of interruption
_STAMP. ”“ TI ”of the device information 304
The value of "ME_STAMP" is incremented by one for each interrupt by the hardware interrupt generator, and is sent to the content decoding unit M2.

FIG. 5 shows the processing of the update / table management unit 300. First, the connection information management table 301 is searched based on the slot position notified from the content decoding unit 200, and the corresponding data is extracted (step 501). If the corresponding data 303 cannot be searched, error processing is performed (step 51).
2) End (step 517). Next, a determination process of the device connection state is performed. First, the identification condition of the device is determined. If "type" and "TYPE" are compared and the values do not match (condition 1), the device is determined to be an invalid device (step 503). If “d_id” and “D_ID” do not match (condition 2), it is determined that the device has been changed (step 504). If the condition 1 is not satisfied, a cutting process (step 513) is performed. In this disconnection processing, the corresponding data 303 (FIG. 4) is initialized. Condition 2
Is not satisfied, a disconnection process (step 513) and a definition process (step 514) are performed. In this definition process,
The corresponding data 303 is set and the mutual management table 3 is set.
02 corresponding data is set. Next, the connection status of the device is determined. There are two conditions for determining an abnormality. Condition 3 is that the current time and the “old_ti
In this case, the threshold value exceeds a certain threshold value (step 505).
e_stamp ”and“ TIME_ST ”of device information 304
AMP "is compared and" TIME_STAMP "is 0 or less or 2 or more (step 506) If conditions 3 and 4 are not satisfied, a confirmation request is issued to the device because an abnormality is expected (step 515). ), If there is a response to the request (step 508), start again (step 50).
The process starts from 0). If there is no response to the request (step 508), disconnection processing (step 513) is performed. After all the comparisons are completed, the update time of the corresponding data 303 (FIG. 3A) is set to old_time,
time_stamp is the TI of the device information (FIG. 3A).
A management group update process (1) to be replaced with ME_STAMP is performed (step 507). When performing this update process, the management group information Group (see FIG. 3) of the corresponding management data M (FIG. 3B) of the mutual management table 302
(B)), and obtains slot positions included in the management groups 30a and 30b (FIG. 3 (b)) including the management data M (FIG. 3 (b)). The comparison is performed on the corresponding data (step 511). In this case, condition 3
Only when the data match, the corresponding data 303 (FIG. 3)
An update process (2) for setting the update time of (a)) to old_time is executed (step 516), and if the conditions are not met, a connection request is issued to the device (step 51).
5) If there is no response (step 508), delete processing is performed (step 513). The process ends when the comparison of all members is completed (step 517).

FIG. 3B will be described. This is a conceptual diagram of a management group, which shows that two management groups are associated with one slot, and one management group is composed of two devices. When the management data A is to be updated, the condition 3 (step 505) is performed on the slot M included in the management group 30a. Since the slot M also belongs to the management group 30b, the same processing is performed when the management data B is updated. In other words, the slot M is monitored by two devices. If the device in the slot M fails, the management data B cannot be updated by itself, but when the management data A or the management group B is updated. The condition 3 (step 505) is implemented for the cutting process (step 51).
3) is always executed. As described above, the connection information management table 301 can be updated even when the notification cannot be made due to the failure of the apparatus main body, and the slot M is double monitored by the management group. This configuration means is an example, and multiple monitoring can be performed by changing the method of configuring the management group.

[0016]

As described above, according to the present invention,
When monitoring the hardware connection status on the software side, there is no need to send a command to the hardware to check the connection status, and it is possible to check whether the device is connected only by receiving an interrupt. is there. Further, it is possible to eliminate a delay in transmission timing that occurs when an interruption occurs during transmission processing. Thus, when the connection state of the device is updated in a state where the system load is high, the latest connection information can be obtained as compared with the conventional system.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a block diagram showing a content decoding unit in FIG. 1;

FIG. 3A is a diagram showing details of an update / table management unit in FIG. 1, and FIG. 3B is a diagram showing details of a mutual management table in the diagram (a).

FIG. 4 is a flowchart showing an interrupt process of the content decoding unit of FIG. 2;

FIG. 5 is a flowchart showing processing of a connection information management table in FIG.

FIGS. 6A and 6B are diagrams showing a flow of execution of a conventional example and an embodiment of the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 connection information management unit 101, 102, 103 hardware 104, 105, 106 interrupt generation device 200 content decoding unit 201 reception device 202 interrupt processing table 203 hardware register 300 update / table management unit 301 connection information management table 302 mutual management Table 303 Applicable data 304 Device information

Claims (1)

(57) [Claims] 1. A hardware having a built-in device for issuing a special interrupt indicating its existence at regular intervals, a content decoding unit for receiving the special interrupt and classifying received information, and a hardware connection. A connection information management unit comprising an update / table management unit for updating a table for recording a situation, wherein the content decoding unit receiving the special interrupt from the hardware interrupts the device information notified by the interrupt. The update / table management unit notifies the update / table management unit by adding the existing slot number, and in the update / table management unit, a connection information management table for recording information of the connection device of each slot, and each slot is defined in advance. A mutual management table for recording information on devices connected to the slots belonging to each group by classifying them into groups, wherein the slot number notified from the content decoding unit is also included. The device information reported to the connection information management table is compared with the device information recorded in the connection information management table to determine the validity of the connected device. If the device information is valid, the connection information management table and the mutual Update the management table, if it is determined to be invalid, perform device disconnection, take out the device information of the slot number of the same group as the notified slot number from the mutual management table, check the latest update time, and check the latest update time from the hardware. A hardware connection state monitoring method for determining the presence or absence of a device abnormality by determining the presence or absence of the special interrupt.