JPS58115558A - Processor monitor system - Google Patents

Abstract

PURPOSE:To monitor operation states of processors to be monitored with one monitor device, by reading the state of a flip flop means by the monitor device in the system provided with one monitor device and plural processors to be monitored. CONSTITUTION:First, a set command for setting a stall indicating flip-flop 206 is transmitted from the monitor device to plural processors to be monitored successively, and setting of respective flip-flop 206 is confirmed by response of respective processors to be monitored. After this circulation, a status information read command including the state of the flip-flop 206 is transmitted successively after TM seconds. When resetting of the flip-flop 206 in every processor to be monitored, the set command is transmitted immediately to the processor, and the state of the flip-flop 206 is confirmed by the response, and the processing for the next processor is started. This processing is repeated in every cycle of TMsec to perform the stall monitor of all processors to be monitored.

Description

[Detailed description of the invention] The present invention relates to a stall monitoring method.

2. Description of the Related Art In information processing systems including a plurality of single processors, a monitoring method is widely used in which a monitoring device is provided within the system, and a dedicated processor for monitoring is provided in the system to monitor the operating states of other processors.

Conventionally, in the monitoring method of this type of information processing system, a monitoring device reads the status of each monitored processor through a monitoring interface, and
lt), and monitors whether an abnormality such as an alarm has occurred.

In this case, the monitoring device can monitor and detect abnormalities in the hardware within each monitored processor and abnormalities directly detected by the programs of each monitored processor, but the programs of the monitored processes and other programs cannot escape from the same loop. It has the disadvantage that abnormalities caused by this, so-called processor stalls, cannot be detected.

SUMMARY OF THE INVENTION An object of the present invention is to provide a monitoring system that eliminates such drawbacks of the prior art.

The method of the present invention provides an information processing system having one monitoring device and a plurality of monitored processors, in which each of the processors receives specific commands included in its own program and instructions from the monitoring device. The monitoring device reads the status of each of the FRI, PFLO, and Pflop means to be controlled in response to the FRI, PFLO, and PFL, and the corresponding FRI, PFLO, and FOP means are not able to escape from the abnormal program loop. Detect failures.

Next, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a diagram showing one embodiment of the present invention. This embodiment shows a monitoring device 1, a plurality of monitored brochures, 2-1
．． . . 2-, .
Contains nine, each monitored bromo.

Each of the frames 2-1 to 2-N includes a side monitoring interface control circuit 20 therein. Further, each of the monitoring interface control circuits has a monitoring output data path line 10.
00 and a monitoring human power data bus line 2000.

Furthermore, as shown in FIG. As shown, multiplexer 201. Parallel-serial conversion circuit 202. Serial parallel conversion circuit 203. Command reception register 204. It includes a decoder 205 and a stall display function 206.

The monitored components 2-1 to 2-N in this embodiment each have different roles, such as an internal data processor (IDP), an input/output control processor (IOP), a mass store, and a digital processor (MSP). Multiple brochures may constitute one system, or the same
Rare multiple bromo, may be. The monitoring device l also includes a service processor (svp) (not shown).

Now, the operation when reading the status information of the service broker of monitoring bag [1, Suga, and a certain monitored Bromo, S (this will be referred to as Bromo, S2-) is as follows.

The service provider outputs command data to the output register 101 via an output line 1010 (FIG. 2) inside the monitoring device 1. As shown in FIG. 4, this command data consists of a processor number field P and a command field C. Field P contains the number of the monitored bromo to which the command is to be sent (currently K ), field C contains a code that specifies the command to be executed (in the current case, a code that specifies input of certain specific status information).

Now, the command data stored in the output register 101 is converted into serial data via the parallel-to-serial conversion circuit 102, and sent to the monitoring output data bus line 1000 in a bit-serial manner.

This command data is transmitted via the Twin 1000 to the monitoring interface control circuit 20 (Fig. 3) of each of the monitoring blocks 2-1 to 2-N, and converted into parallel data by the parallel-serial conversion circuit 203 of each circuit. It is converted and stored in the command reception register 204.

The processor number field P of the data stored in the register 204 is decoded by the decoder 205, and only the stored command data of the number specified by the contents of P becomes valid. This stored command data is ignored in the other control circuits 2.

Now, in the designated bromo, S2-K.

The command field C of the command data stored in register 204 is decoded and an output responsive to the decoded result is applied to the input data selection input terminal of multiplexer 2010.

As a result, data input 2010 of multiplexer 201
Among the status information of this processor added to the processor, the status data specified in the command field C is selected and output in parallel to the parallel-to-serial conversion circuit 202, which converts it into serial data and converts it into bits. The data is serially transferred to the monitoring device side monitoring interface control circuit 10 via the monitoring manual data path line 2000.

On the control circuit IO side, this is connected to the serial/parallel converter circuit 10.
3 (FIG. 2), the data is converted into parallel data and latched into the input register 104. By inputting the contents of this register 104, the service processor of the monitoring device 1 can obtain the designated status information of the designated blog.

As stated above. The feature of this embodiment of monitoring by the service processor is that this method includes the acquisition of stall information of the monitored broker. This will be explained below.

First, the service information field P contains the processor number (K) of the monitored processor to be checked in the processor number field P. The command data/1 containing a predetermined code for instructing the setting of the stall display flip 70 knob 206 in the command field C is created, and is output to the output register 101 of the control circuit 10.

This command data is exactly the same as above,
The command field C is decoded in the decoder 205 of the control circuit 20 of the designated processor 2-, thereby setting the stall indicating flip-flop 206 included in this circuit via line 2050. At the same time, the output information 206 of this Furi, Pufuro, Pu 206
Status information containing 0 (currently set to II'') is selected by multiplexer 201 and transferred to input register 104 of monitoring device I in the manner described above.

By reading this, the service processor sends a stall indication flip-flop to the specified processor 2-.
Confirm that 06 is set according to the command.

Now, the operating program running in the monitored program has a program path that must be passed within a certain time interval T seconds if it is operating normally. A reset command is inserted into this path to reset the stall indicating block flop 206 via line 2061.

As a result, if the Furi, Guraro, Pu206ke, Bromo, and Su2-K set as described above are operating normally, they will always be reset within T seconds after being set. There will be. Therefore, Service Bromo Suha set the above-mentioned stall display flip-flop 206.
After sending the command to load and confirming the set, after T seconds have elapsed, the program selects and reads out the status information including the status of the stall display flip-flop 206 to S2-. In response, the monitored processor 206, which creates the specified command data and sends it as described above, returns status information including the state of the flip-flop processor 206 at that time. do.

By reading this returned information, the service processor learns the current state of the 1 display flip-flop 206, and if it has been reset, the service processor
is working fine and this is set to it
If so, it can be determined that Bromo and Su2-K are in a stalled state. If Bromo.

If it is determined that the processor 2-K is in a stalled state, the service processor 7 will restart the monitored processor 2 in the same manner as described above after a predetermined timeout period has elapsed. -Ni no Furi, Puflo, Pu20
Reconfirm the set state of 6, and this Bromo,
I am glad that I finally decided that I was in a stalled state.

In addition, the above-mentioned time T generally differs depending on the type of service, but the maximum value of these T is set as 1, and this 1 is used as a cycle for each target. The above-described stall monitoring for the monitoring processor may be repeated.

That is, first, the stall display slip 70, the slip 2
06 is sequentially sent to each of the monitored processors 2-1 to 2-N, and at the same time, it is confirmed that each flip-flop 206 has been set by a response from each processor 2-1 to 2-N. After this cycle, after Tk1 seconds, a read command to read the status 4 including the state of the flip-flop 206 is sequentially sent to each processor, and each processor confirms that the flip-flop 206 has been reset. Immediately, the above-mentioned set command is sent to the brochure, and at the same time, this flip 70. After confirming the steps 206 and 206, the process proceeds to the next step. By repeating such processing at a cycle of 1 second thereafter, it is possible to perform stall monitoring on all monitored processors, 2-1 to 2-N, using the same cycle.

Of course, if necessary, choose the most appropriate T for each Brosse.
It is also possible to set a value K and perform stall monitoring at a different cycle from this value K.

Additionally, the supervisory interface control circuits IO and 2 are generally
The data bus lines 1000 and 2000 between 0 and 0 transmit a relatively small amount of information, so in this embodiment,
Although a bit-serial transmission system is used here, a bit-parallel transmission system may of course also be used.

Furthermore, the monitoring device and the monitored unit described in the above embodiments do not need to be physically isolated from each other; they may exist in the same housing, the same panel, or even the same board. There is no problem even if you do.

As described above, by using the present invention, it is possible to monitor the operating states of a plurality of monitored processors, including monitoring the stalls of each monitored processor, using a single monitoring device.

This K makes it possible to provide a more reliable monitoring system.

[BRIEF DESCRIPTION OF THE DRAWINGS] FIG. 1 is a professional diagram showing an embodiment of the present invention, and FIGS. 2 and 3 are professional diagrams and diagrams for explaining some details of this embodiment. FIG. 4 is a diagram for explaining 7 ohms and 7's of command data used in this embodiment. In the figure, l...Monitoring device, 2-1 to 2-N
...Monitored block, 10...Monitoring device side monitoring interface, -S control circuit, 20...
Brose, side monitoring interface control circuit, 101...
Mark/output register, 102, 202...Parallel-serial conversion circuit % 103゜203...Serial-parallel conversion circuit, 1o4...Input register, 201...
...Multiplexer, 2o4, river...! Command reception register S 205...Decoder, 206...
... Flip flop for stall display.

Claims (1)

[Claims] In an information processing system having one monitoring device and a plurality of monitored processors, each processor is controlled in response to specific instructions included in its own program and instructions from the monitoring device. A processor monitoring method comprising flip-flop means, wherein the monitoring device detects a failure caused by a corresponding process being unable to escape from an abnormal program loop by checking the state of each of the flip-flop means. .