JPH064415A - System for initial diagnosis of memory and device therefor - Google Patents

Abstract

PURPOSE:To always present a useable memory capacity by dividing all mounting memories into areas at the time of initial diagnosis of a hardware, and recording information to each area thereof. CONSTITUTION:A computer is constituted of a processor 101, a nonvolatile memory 102, an input/output means 103, and a memory subsystem 104. The memory subsystem 104 is constituted of a memory subsystem control part 105, and memories 106-(n). Also, the input/output means 103 uses a keyboard and a CRT as an input means and an output device, respectively, and sets an end address of the mounting memory capacity held as data in the nonvolatile memory 102 as MEM-END. In such a state, when a power source is turned on, the mounting memory capacity is recognized by reading out the end address MEM-END of the mounting memory capacity held in the nonvolatile memory 102, and the number of divisions of the mounting memory capacity is recognized by reading out the number of divisions DEV-NO held as data in the contents of the nonvolatile memory 102.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer.

[0002]

2. Description of the Related Art As a known example of using a method of dividing a memory area in a computer system as a countermeasure against a failure, there is JP-A-61-267844. This conventional example
When the memory area is divided according to the program, each program has a flag indicating whether or not it has a fatal effect on system operation, and the address of the error that occurred belongs to a program that has a fatal effect. Aims to improve the operating rate of the computer by stopping the system and continuing the processing when it belongs to only the process that has no fatal influence, depending on the program information running on the system. And

Further, the method at the time of initial diagnosis of the memory before the determination of the operation program is the following two methods (1) The failure diagnosis is performed for each memory board, and when the failure is detected, the board is disconnected for each board. Information is recorded for this purpose, and (2) information is recorded for recording the address at which the failure is detected and separating it.

[0004]

Considering the method disclosed in Japanese Patent Laid-Open No. 61-267844, since the area is divided based on the information of the operating program, the area cannot be divided until the operating program is known. Generic The computer system which programs in advance which memory area can not be determined either loaded is therefore load address of a program in the system until the decision can not be divided memory area.

Therefore, it cannot be used for diagnosing memory resources at the time of initial diagnosis of system hardware.

Further, the conventional system of the memory initial diagnosis before the determination of the operation program has the following problems. That is, in the computer system, the mounting method and the number of mounted memory boards may differ depending on the mounting form. In the conventional method (1) of managing by memory board unit, failure information is recorded depending on the mounting mode. Therefore, when a single memory board is installed and a large capacity memory is mounted, a sufficient amount of information is stored. Even if there is available memory, the available memory becomes empty and the system cannot be started. Further, in the conventional method (2) of recording a failure address, a large amount of failure recording area is required to record failure information when there is a failure in a large-capacity memory area. , The available memory capacity becomes insufficient and the system cannot be started up.

An object of the present invention is to indicate to the software the memory capacity which can always be used for diagnosing the memory resources at the time of initial diagnosis of the hardware of the computer system without depending on the mounting form of the memory and the status of the fault that has occurred. Is to provide a method.

[0008]

In order to achieve the above object, according to the present invention, all mounted memories are divided into areas at the time of initial diagnosis of hardware, and information is recorded in each of the areas. , And a method of indicating to the software the memory capacity that can always be used without depending on the situation of a failure that has occurred.

[0009]

According to the present invention, it is possible to indicate to the software the memory capacity that can be used in the initial diagnosis of hardware and the memory capacity that can always be used without depending on the status of a failure that has occurred. It is possible to construct a computer system that can cope with various failure situations.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. An example will be described in which the mounted memory capacity is 80 Mbytes, the number of divisions is 32, and the failure status is indicated by a 4-byte (32-bit) field.

FIG. 1 shows an embodiment of the configuration of a computer system according to claim 1. The computer includes a processor 101, a non-volatile memory 102, an input / output unit 103, and a memory subsystem 104. Memory subsystem 10
4 is a memory subsystem control unit 105 and a memory 106.
Up to n. The input / output unit 103 uses a keyboard as an input unit and a CRT as an output device. The end address of the mounted memory capacity held as data in the nonvolatile memory 102 is MEM_EN.
Let be D.

FIG. 2 shows the correspondence between the bit field and the memory space to be fault-managed in this case. Bit 0 of the 32-bit field starts at address 0 (MEM_E
ND + 1) ÷ 32-1 corresponding to the area, and bit 31 of the 32-bit field below is the address (MEM_
It corresponds to the area from END + 1) ÷ 32) × 31 to MEM_END.

An embodiment of the initial diagnosis processing of the present invention will be described with reference to FIG. The processor 101 is firmware 1
The control is started by 02. First, the mounted memory capacity is recognized. First, when the power is turned on, the nonvolatile memory 10
It is possible to recognize the mounted memory capacity by reading the end address MEM_END of the mounted memory capacity held as data in the content of 2.

Next, the number of divisions of the mounted memory capacity can be recognized by reading out the number of divisions DEV_NO held as data in the contents of the nonvolatile memory 102 when the power is turned on.

Since the number of divisions is 32 in this example, a calculation for dividing the mounted memory capacity into 32 equal parts will be shown. Since 32 is 2 to the 5th power, the value obtained by shifting the memory capacity to the right by 5 bits becomes 1/32. Below, the installed memory capacity is 80M
An example when it is byte is shown.

The installed memory capacity of 80 Mbytes is a binary number of 1010000000000000000000.
Since 0 is 0, 1/32 is a binary number, as shown in FIG. 4, which is 101000000000000000000.

Here, a method of creating the diagnostic process management table will be described with reference to FIG. First, TBL

[0] is the start address 0 of the memory space. Next, TBL [1]
Is 10000000000000000000000000 calculated in FIG. TBL [2] is a value obtained by adding 101000000000000000000000000 to TBL [1], and hereinafter, TBL [31] is similarly created. This table is created in the non-volatile memory 102 by the processor 101 by the above algorithm.

Next, the processor 101 performs reading after writing to all the memories 106 to n in order to perform a memory diagnostic process. Then check to see if an error has occurred. Data written to check for an error is held in a register inside the processor, and if the read value is different from the held value, it is recorded as an error.

When recording next, the diagnostic processing management table created in FIG. 5 is used. The processor 101 sets the address of the faulty memory to the TBL of the table created in the non-volatile memory 102.

The contents of [0] to TBL [31] are sequentially compared, and the addresses are TBL [N-1] and TBL [31].
Find N such that it lies between [N]. At this time, the bit that records the failure information is the N-th field of the 32-bit field.
Become a bit. However, N is 0 to 31. The processor 101 records the failure information and writes it in the non-volatile memory 102 until the diagnosis is completed up to the end of the memory. Further, the failure information is output via the input / output unit 103.

By the above processing, it is possible to construct a computer system which can always show the available memory capacity to software without depending on the mounting form of the memory and the situation of the failure that has occurred.

Next, an embodiment of the configuration of the computer system of claim 12 will be shown. The basic configuration is as shown in FIG. FIG. 6 shows an example in which hardware is added to the memory subsystem control unit. Each part will be described below.

The memory subsystem control unit includes a memory fault information recording register 601, a fault recording non-volatile memory 6
02, division number setting register 603, division number saving nonvolatile memory 604, mounting memory capacity register 605, mounting memory capacity saving nonvolatile memory 606, memory failure address setting register 607, memory address failure information data conversion table 608, memory It is composed of the subsystem controller 609.

The relationship of each hardware is shown below. The memory subsystem controller 609 can read the contents of the mounted memory capacity saving non-volatile memory 606. For setting the mounted memory capacity, input / output unit 1
The information input by 03 is taken into the system, and the processor 101 sets the mounted memory capacity register 60.
5 can be set in the mounted memory capacity storing non-volatile memory 606. Further, the memory subsystem controller 609 can read the contents of the division number storage non-volatile memory 604.

Regarding the setting of the number of divisions, the input / output means 10
The information input by 3 is taken into the system, and the processor 101 writes it in the division number setting register 603 to set it in the division number storage nonvolatile memory 604.

Further, the memory subsystem controller 6
Reference numeral 09 denotes the table of FIG. 5 based on the contents of the mounted memory capacity storage non-volatile memory 606 and the division number storage non-volatile memory 604 according to the algorithm described with reference to FIGS. 4 and 5. Created in the data conversion table 608.

The processor 101 is connected to the memory 106 through n.
, Write data and then read
The failure address is set to the memory failure address setting register 60
The fault address of the memory is set in 7. For that processing, the memory subsystem controller 609 refers to the contents of the memory address failure information data conversion table 608 and

The contents of [0] to TBL [31] are sequentially compared to obtain N such that the address is between TBL [N-1] and TBL [N]. At this time, the bit that records the failure information is the Nth bit of the 32-bit field. However, N is 0 to 31.

Memory subsystem controller 609
Stores the Nth bit in the memory failure information recording register 60.
When set to 1, failure information is set in the failure recording non-volatile memory 602. When the processor 101 sets 0 in the memory failure information recording register 601, the failure information in the failure recording nonvolatile memory 602 is cleared.
After diagnosing all the memories, the processor 101 outputs the failure information via the input / output unit 104.

With such hardware, it is possible to construct a computer system that can always indicate to the software the memory capacity that can be used without depending on the memory mounting mode and the status of a failure that has occurred.

[0029]

According to the present invention, it is possible to always indicate to the software the memory capacity that can be used without depending on the memory mounting mode and the status of a failure that has occurred. Therefore, various mounting modes and various failures can be shown. It is possible to provide a computer system that can handle situations.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram showing a correspondence between a memory area of the initial diagnosis system of the present invention and fault record information.

FIG. 3 is a flowchart showing a flow of an initial diagnosis method of the present invention.

FIG. 4 is an explanatory diagram of a method of dividing the mounted memory into 32 equal parts when the mounted memory is 80 Mbytes.

FIG. 5 is an explanatory diagram of an example of a diagnostic processing management table.

FIG. 6 is a block diagram showing an embodiment of a memory subsystem control unit.

[Explanation of symbols]

101 ... Processor, 102 ... Nonvolatile memory, 103
... I / O means, 104 ... Memory subsystem, 105 ...
Memory subsystem control unit 106-n ... Memory.

Claims (13)

[Claims] 1. A computer system comprising a processor, a memory subsystem having means for indicating mounted memory capacity, firmware for holding programs and data of the processor, and input / output means, and having a plurality of total mounted memory areas. The memory initial diagnosis method is characterized in that the failure information is recorded for each of the above division units. 2. The memory initial diagnosis system according to claim 1, wherein the size of each of the divided areas is the same when the area is divided. 3. A memory initial diagnosis apparatus having means for recording fault information for each division unit of a memory area in order to use the memory initial diagnosis method according to claim 1. 4. The memory initial diagnosis device according to claim 3, further comprising means for setting the number of divisions of the area. 5. The memory initial diagnosis device according to claim 3, further comprising a non-volatile information recording means for storing the set division number of the area. 6. The memory initial diagnosis device according to claim 3, further comprising input / output means for setting the number of divisions of the area. 7. The memory initial diagnosis device according to claim 3, further comprising means for holding fault information of the area-divided memory. 8. The memory initial diagnosis apparatus according to claim 3, further comprising means for recording fault information of the area-divided memory. 9. The memory initial diagnosis device according to claim 3, wherein the division is performed by a power of two. 10. The memory initial diagnosis method according to claim 1, wherein the number of divisions is a power of two. 11. The memory initial diagnosis according to claim 3, further comprising means for identifying to which of the divided memory areas the memory address belongs from the means for indicating the number of divisions and the mounted memory capacity. apparatus. 12. The memory initial diagnosis device according to claim 3, further comprising means for converting an error occurrence address into information for recording failure information. 13. The memory initial diagnosis device according to claim 3, further comprising means for outputting fault information record information.