JP5348120B2 - Program, method and mechanism for collecting panic dump - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable an operating system to reliably implement a process for acquiring memory data through a panic dump when a malfunction occurs. <P>SOLUTION: In the constituent information of a main memory 5 which is provided to the operating system by boot firmware, at startup of the operating system in a system operation mode and at a dump operation mode, a plurality of different regions (a first main memory region, a second main memory region) are set and a dumping process is executed by restarting the operating system in the dump operation mode while malfunction-time memory information is saved at the time of occurrence of malfunction. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a program, a method, and a mechanism for collecting a panic dump when a failure occurs.

  As information communication technology becomes widespread, high reliability is required for information processing apparatuses, particularly server systems that operate in backbone systems. Therefore, if a failure occurs during system operation, it is essential to collect information promptly and continue operation.

  Generally, when the system operation cannot be continued due to the occurrence of a fatal failure, a so-called panic dump function for dumping memory data at the time of the failure is used as a means for collecting information.

  A panic dump is used to save the memory contents at that time when an abnormality that prevents the system from operating continuously is detected during system operation, and is usually in the OS (Operating System) or in the kernel. A program that runs on is performing dump processing.

  For example, when the CPU is operating a program in the kernel and receives an abnormality detection interrupt signal, the CPU shifts control to a memory dump program incorporated in the kernel and collects a memory dump.

  Since the memory dump program is incorporated in the kernel, it is possible to analyze the system table in the kernel and dump the information necessary for failure analysis with an optimal dump size.

  However, if the cause of the abnormality is, for example, the program itself operating in the kernel, inconsistency in control data, destruction of memory storing the kernel (program), hardware abnormality, etc., dump processing May require resources (eg, corrupted memory) that caused the anomalies. In such a case, the dump process fails because the abnormality occurs again during the dump process.

  In addition, depending on the type of abnormality, the system may hang up, and the process cannot proceed to the panic dump process, so the memory data may not be dumped.

  To solve this problem, reset the system with the memory data saved, reset the hardware resources other than the memory data that you want to dump, restart the dump processing program, and use the memory data in that environment. Stand-alone dump to dump is known.

  For example, if memory dump collection by the memory dump program installed in the kernel fails as described above and the system hangs, the system is reset with the data stored in the memory preserved, and incorporated in the kernel A memory dump is collected by starting a dedicated memory dump program (stand-alone dump program) different from the memory dump program.

  Stand-alone dump allows dumps to be collected regardless of the environment in which the error occurred (kernel control data inconsistency, memory corruption, etc.), and if there is a temporary hardware error, the hardware is reset once This increases the possibility of normal operation. In the case of a permanent hardware abnormality, there is a high possibility that the abnormality can be detected by a POST (Power On Self Test) diagnosis executed when the system is restarted due to a reset or a hardware initialization process.

JP 08-095834 A

However, the conventional stand-alone dump has the following problems.
1) In order to reset the system and download the stand-alone dump, it is necessary to save in advance the data in the memory area overwritten by the stand-alone dump. This must be done with boot firmware for booting the OS before loading (storing) a stand-alone dump.
Therefore, it is necessary that the boot firmware has a hardware resource sufficient to temporarily save the memory data, or a dedicated partition that can be controlled by the boot firmware is secured on a disk or the like and stored as a file there.
It is disadvantageous in terms of cost if it has dedicated hardware resources. If you have a dedicated partition, you must ensure that there is a dedicated partition for temporarily saving memory data on the connected disk. Since it cannot be controlled from firmware, there is a problem that disk partition management becomes complicated.

2) Since the stand-alone dump is booted with the memory data at the time of failure occurrence, the boot firmware and OS loader (program for loading the OS into the memory) data on the memory are completely stored in the system boot process. It will be overwritten.
Therefore, when an abnormality occurs between the kernel and these boot firmware and OS loader, there is a problem that it is difficult to investigate because firmware data cannot be collected as dump data.

3) Since the server system normally has a main memory of several GB or several tens GB, it is not realistic to collect data of all memories by panic dump. For this reason, usually only the areas such as kernel text and kernel data of the operating system necessary for investigation are collected. In order to obtain information in this area, it is necessary to search and analyze the table in the kernel, but since this information differs depending on the version of the kernel, use a stand-alone dump that is a separate program from the operating system. In this case, it is necessary to prepare a stand-alone dump program corresponding to the operating system version. Therefore, the version number of the stand-alone dump needs to match the version number of the corresponding operating system. If it does not match, the table in the kernel cannot be searched, so dumping fails or all memory data to be mounted must be dumped.

  In Patent Document 1, in order to solve these problems 1) and 2), an operating system in which an operating system uses a system dump collection program use area that is not used during normal operation in a main memory area of the system. Set up a system dump collection program from an external storage device after resetting the computer system and before reloading the operating system to the operating system usage area when a system dump cannot be collected due to hang-up etc. A system is disclosed that collects a system dump in an operating system use area by loading and executing the dump collection program use area.

  However, the stand-alone dump system different from the operating system cannot solve the problem pointed out in 3) above. That is, even in the system disclosed in Patent Document 1, a list of areas for which a system dump is to be acquired is created in advance from table information for managing areas on a main storage device that is statically or dynamically allocated. Thus, since a system dump is collected only for the areas shown in this list, a list of target areas must be prepared in advance.

  Since the information for creating the list information usually depends strongly on the version number of the operating system, the problem of complicated version management of the operating system and version management of the list cannot be solved.

  In addition, since a list of areas for which system dumps should be acquired must be created in advance, the area dynamically allocated by the operating system during operation cannot be predicted in detail, and a large-scale server system with a complicated system configuration cannot be predicted. In some cases, it is difficult to efficiently collect necessary information. Moreover, there is a problem that a necessary area cannot be collected if the version number matching fails due to a patch.

  The present invention has been made in view of the above-described problems, and the problem to be solved is to collect a panic dump that makes it possible to reliably perform a memory data collection process using a panic dump when a failure occurs. It is to provide a program, method and mechanism.

  According to the first aspect of the present invention, in response to the operation mode designation information for controlling the operation of the information processing apparatus, the storage area included in the information processing apparatus is used in an operating system other than the first area. The area definition information acquisition process for acquiring the second area definition information from the area definition information means storing the second area definition information to be drawn in the area 2, and the operating area in the area based on the second area definition information An operating system storage process for storing a system; a dump target area calculation process for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area; A dump process for reading out and outputting information stored in the dump target area, Is a program for causing the line.

  According to the first aspect of the present invention, the first area (used when a failure occurs) corresponding to a predetermined area in the second area acquired by the area definition information acquisition process by the dump target area calculation process. In this case, the predetermined area is calculated and the data of the calculated dump target area is output, so that it is possible to reliably acquire the data of the area used when the failure occurred.

  In addition, the operating system that operates in the first area, that is, the area that was used when the failure occurred, and the operating system that operates in the second area have the same area management (for example, a memory management table). The second area acquired by the area definition information acquisition process by the dump target area calculation process, for example, when managing a memory address of data including a program necessary for operating the operating system or the like. It is possible to easily calculate a predetermined area in the first area corresponding to the predetermined area.

  According to a second aspect of the present invention, there is provided a failure notification process for detecting and notifying that a failure has occurred in a component required for the information processing apparatus to operate, and the information according to the notification by the failure notification process An operation mode designation information storage process for storing operation mode designation information for controlling the operation of the processing apparatus in the operation mode designation information storage means, and the information processing apparatus while maintaining the state of the area used when the failure occurs In response to the restart process to be restarted and the operation mode designation information, the second storage area of the information processing apparatus is defined as a second area other than the first area used for the operating system. Region definition information acquisition processing for acquiring the second region definition information from the region definition information means storing the region definition information, and operating the region based on the second region definition information. An operating system storage process for storing a system; a dump target area calculation process for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area; A program for causing a computer to execute a panic dump collection process including a dump process for reading and outputting information stored in the dump target area.

  According to the invention described in claim 2, in addition to the effect of the invention described in claim 1, when a failure occurs, the information processing apparatus (or the constituent elements constituting the information processing apparatus, and the components are configured by restart processing) From the area definition information means storing area definition information for drawing the area of the storage means included in the information processing apparatus into a plurality of areas by the area definition information acquisition process. The area definition information for the second area other than the first area used for the operating system is acquired, and the operating system is loaded in the area based on the acquired area definition information. It is possible to perform the reset process while retaining the area that has been set.

  According to a third aspect of the present invention, in the dump target area calculation process, the dump target area is set to an offset between a head position of the first area and a head position of the second area, and the second area. The program for collecting a panic dump according to claim 1, wherein the program is calculated based on a difference between the predetermined area and the predetermined area.

  According to a third aspect of the present invention, the second region is determined by the difference between the offset between the leading position of the first region and the leading position of the second region and a predetermined region in the second region. By calculating the first area corresponding to the predetermined area in the area, it is possible to easily calculate and acquire the dump target area.

  According to a fourth aspect of the present invention, in accordance with operation mode designation information for controlling the operation of the information processing apparatus, the area of the storage means included in the information processing apparatus is other than the first area used for the operating system. An area definition information acquisition process for acquiring the second area definition information from the area definition information means storing the second area definition information drawn in the second area, and an area based on the second area definition information. An operating system storing process for storing an operating system; a dump target area calculating process for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area; And a dump process for reading out and outputting information stored in the dump target area.

  According to the fifth aspect of the present invention, in response to the operation mode designation information for controlling the operation of the information processing apparatus, a storage area included in the information processing apparatus is used in an operating system other than the first area. An area definition information acquisition unit that acquires the second area definition information from an area definition information unit that stores the second area definition information to be imaged in the second area, and an area based on the second area definition information. Operating system storage means for storing a system; dump target area calculation means for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area; A panic dump collection mechanism comprising: dumping means for reading out and outputting information stored in the dump target area.

  Also according to the invention of claim 4 and claim 5, as in the case of claim 1, when a failure occurs, the area definition information acquisition process is performed by the dump target area calculation process (or the dump target area calculation unit). The predetermined area in the first area (area used at the time of failure occurrence) corresponding to the predetermined area in the second area acquired by calculating the dump target area is output. Therefore, it is possible to reliably acquire the data of the area that was used when the failure occurred.

  In addition, since the operating system operating in the first area, that is, the area used when the failure occurs, and the operating system operating in the second area perform the same area management, the dump target The predetermined area in the first area corresponding to the predetermined area in the second area acquired by the area definition information acquisition process is easily calculated by the area calculating process (or the dump target area calculating means). It becomes possible.

  As described above, according to the present invention, it is possible to provide a program, a method, and a mechanism for collecting a panic dump that can surely perform a memory data collection process using a panic dump when a failure occurs.

It is a figure which shows the example of the principal part of the structure which concerns on the Example of this invention. It is a flowchart which shows the process of ROM firmware which concerns on the Example of this invention. It is a figure which shows the example of the memory information table at the time of the system operation mode which ROM firmware produces. It is a figure which shows the example of the memory information table at the time of the system operation mode which ROM firmware produces. It is a figure which shows the example of the memory information table at the time of dump operation mode which ROM firmware produces. It is a flowchart which shows the process of the boot firmware which concerns on the Example of this invention. It is a flowchart which shows the process of OS loader and OS concerning the Example of this invention. It is a figure which shows the specific example of the process of step S407 to step S409 shown in FIG. It is a flowchart which shows the process after OS detects abnormality until a restart process is performed. 10 is a flowchart illustrating a modification of processing from when an OS detects an abnormality until restart processing is performed. . An example of processing from when the OS detects an abnormality until restart processing is performed when the power supply control means is provided with dump operation mode / system operation mode switching means is shown.

Hereinafter, embodiments of the present invention will be described with reference to FIGS. 1 to 8.
FIG. 1 is a diagram illustrating an example of a main part of a configuration according to an embodiment of the present invention.
The information processing apparatus 1 shown in FIG. 1 stores a CPU 2 (not shown) that controls each component of the information processing apparatus and performs processing according to the program according to the present embodiment, ROM firmware, boot firmware, and the like according to the present embodiment. Read only memory (ROM) 3 and environment variables used by ROM firmware and environment variables used by boot firmware (for example, ROM firmware operation mode designation flag, boot firmware operation mode designation flag, etc.) It is an information processing apparatus including at least a nonvolatile memory 4 configured by (volatile RAM) or Flash Memory, and a main memory 5 for storing boot firmware, an OS loader, an OS kernel, and the like.

  The information processing apparatus 1 according to the present embodiment is connected to an external recording device 6 (for example, a magnetic disk device) that stores an OS loader and an OS kernel. The external recording device 6 shown in the figure is connected to the outside of the information processing device 1, but may be inside.

  When the main unit is powered on, the ROM firmware stored in the ROM 3 is activated, performs hardware initialization and diagnosis, and performs hardware configuration recognition processing. In the configuration recognition processing, the number of mounted CPUs 2 and the configuration of the mounted memory (mounted main memory 5) are recognized, and the configuration information is stored in the ROM firmware environment variable storage area.

  The ROM firmware reads boot firmware data from a predetermined address in the ROM 3, stores it in the main memory 5, develops it as executable boot firmware on the main memory 5, and passes control to the boot firmware.

  Here, the ROM firmware provides a method for passing the memory configuration information of the hardware recognized in the configuration recognition process when passing control to the boot firmware or notifying the memory configuration information.

The boot firmware recognizes the memory mounting state of the entire system and operates the OS loader and the OS in this memory mounting environment.
Further, some programs remain on the main memory 5 after the OS is started, and provide various services to the OS loader and the OS.

Therefore, the OS loader and the OS kernel do not directly view the hardware information, but abstract configuration information (for example, memory configuration information notified by the boot firmware, not actual memory configuration information). ) To improve versatility without depending on individual hardware specifications.
The first main memory area and the second main memory area are determined based on the memory configuration information notified from the ROM firmware to the boot firmware.

  The main memory 5 according to the present embodiment described above is composed of the first main memory area 7 and the second main memory area 8, but it is only necessary to be logically divided into two memory areas. Physically, it may be composed of one memory. Further, it may be composed of a plurality of (physically divided) memories. Alternatively, it may be logically divided into two or more memory areas (for example, a first memory area, a second main memory area, and a third main memory area).

  FIG. 2A to FIG. 4 show processing flowcharts when the information processing apparatus 1 is turned on and when it is restarted (restart processing when the OS panics), and the processing is performed based on the drawings. explain.

FIG. 2A is a flowchart showing processing of the ROM firmware according to the embodiment of the present invention.
When the information processing apparatus 1 is turned on or restarted (step S201), the CPU 2 provided in the information processing apparatus 1 refers to the ROM firmware stored at a predetermined address of the ROM 3 and performs hardware initializing according to the instruction. Processing is performed (step S202). For example, hardware such as the main memory 5 and the external recording device 6 is initialized.

  When the hardware initialization process is completed in step S202, the CPU 2 determines, for example, the reset factor notified from the firmware performing power control (in this case, not the operation mode information related to the ROM firmware operation mode designation flag, for example, the power supply It is determined whether the power is turned on by the switch or information indicating whether the power is turned on by restarting, etc.) is due to the power-on or the restarting ( Step S203).

  If the reset factor is due to power-on, the CPU 2 moves the process to step S204, and initializes the ROM firmware operation mode designation flag stored at a predetermined address in the ROM 3 to the system operation mode. When the initialization of the ROM firmware operation mode designation flag is completed, the process proceeds to step S205.

  In step S203, if the reset factor is not power-on (restart), the CPU 2 proceeds to step S205 without initializing the ROM firmware operation mode designation flag.

In step S <b> 205, the CPU 2 performs a process of recognizing mounting information and the like (hereinafter simply referred to as configuration information) related to the components constituting the information processing apparatus 1.
For example, the configuration information of the CPU 2 and the memory, etc., which is provided in the information processing apparatus 1 and has been normally initialized in step S202 (for example, the number of CPUs 2 mounted, the memory mounting capacity, etc.) is read out from the register provided in each component. To obtain the configuration information.

  When the recognition processing of the configuration information such as the CPU 2 and the memory is completed, the CPU 2 shifts the processing to step S206, and stores the mounted CPU 2 and memory information in the ROM firmware environment variable storage area in the ROM 3.

  Further, in step S208, the CPU 2 reads a ROM firmware operation mode designation flag stored at a predetermined address in the ROM 3 and performs a process of determining whether the system operation mode is set.

  In this embodiment, the ROM firmware operation mode designation flag uses 1-byte data, and the dump operation mode is determined when the flag is 0xFF. When the flag is other than 0xFF, it is determined that the system operation mode is set. However, the flag is not limited to the above case, and a value determined in advance may be used so that the dump operation mode and the system operation mode can be distinguished.

  In step S208, if the CPU 2 determines that the system operation mode is not selected, the process proceeds to step S209, and the memory information table used in the dump operation mode from the mounted memory information acquired in the process of step S205, that is, the first A memory information table defining two main memory areas is created and passed to the boot firmware as a system memory table.

  Here, the system memory table is a table that defines a memory area (or memory configuration) that can be accessed by the boot firmware, the OS loader, and the OS kernel (or the OS including the OS kernel).

Therefore, in the processing described below (the processing shown in FIGS. 2A to 4), when operating in the dump operation mode, the processing is performed on the second main memory area (for example, the boot firmware in step S213 is replaced with An area to be expanded, an area to expand the OS loader in step S305 in FIG. 3, an area to expand the OS kernel in step S401 in FIG. 4, and the like).
By step S209, the ROM firmware creates, for example, the table shown in FIG. 2B.

  The memory information table shown in FIG. 2B is an example in which only the 512 MB continuous area starting from 0x000000C000000000000 is defined as an area usable by the boot firmware, the OS kernel, and the OS kernel when the ROM firmware is in the dump operation mode.

  In the memory information table shown in the figure, for example, the first 128 MB area is used by the boot firmware, and the next 358 MB area is used by the OS loader and the OS kernel.

  If the CPU 2 determines in the system operation mode in step S208, the process proceeds to step S210, and the memory information table used in the system operation from the mounted memory information acquired in the process in step S205, that is, Then, a memory information table defining the first main memory area is created and passed to the boot firmware as a system memory table.

  Therefore, in the processing described below (the processing shown in FIGS. 2A to 4), when operating in the system operation mode, the processing is performed on the first main memory area (for example, the boot firmware in step S213 is changed). An area to be expanded, an area to expand the OS loader in step S305 in FIG. 3, an area to expand the OS kernel in step S401 in FIG. 4, and the like).

  By step S210, the ROM firmware creates a table shown in FIG. 2C or 2D, for example.

  The memory information table shown in FIG. 2C has a 4 GB continuous area at the beginning of each space of four 256 GB, the boot firmware uses the 128 MB area from the address 0x00000000F8000000, and the other areas use the boot firmware, OS loader, and A table when the OS kernel is used is shown.

  That is, the ROM firmware is an example in which all the main memory 5 is defined as the first main memory area when the system operation mode is set (therefore, in this case, the first main memory area is only set in the dump operation mode). And a second main memory area).

  In addition, the memory information table shown in FIG. 2D has a 4 GB continuous area at the beginning of each space of four 256 GB, the boot firmware uses 128 MB from 0x00000000F8000000, and the last 512 MB of the continuous area starting from 0x000000C000000000000 is hidden. The table shows a case where only 4 GB-512 MB can be accessed from the OS kernel or the like as a continuous area.

  That is, in the ROM firmware, the continuous area starting from 0x000000C000000000000 is an example in which the last 512 MB is defined as the second main memory area and the other areas are defined as the first main memory area. Therefore, in the system operation mode, only the first main memory area can be accessed from the boot firmware, the OS loader, and the OS kernel.

  When the creation of the memory information table is completed, the CPU 2 shifts the processing to step S211, and executes memory initialization processing and diagnostic processing for the memory area set in the system memory table passed to the boot firmware.

  For example, the memory initialization process is performed by writing a predetermined pattern (for example, all 0) in the memory data. Further, the memory diagnosis process can be performed by writing and reading a predetermined pattern, comparing the written pattern with the read pattern, and determining whether there is a match or not.

  In the present embodiment, all available memory areas set in the system memory table are targeted for memory diagnosis. That is, in the system operation mode, the entire memory area used during system operation is subject to memory diagnosis. Similarly, in the dump operation mode, the entire memory area used during dump operation is stored in memory. It is targeted for diagnosis.

  When the memory initialization process and the diagnostic process are completed in step S211, the CPU 2 shifts the process to step S212, and determines the address of the memory storing the boot firmware from the memory area managed by the system memory table.

  For example, the start address (offset address from the memory area) for storing the boot firmware in the memory area defined in the system memory table is read from the predetermined address in the ROM firmware environment variable storage area of the ROM 3.

  When the address of the memory storing the boot firmware is determined, the CPU 2 shifts the processing to step S213, reads the boot firmware stored at a predetermined address in the ROM 3, and expands the boot firmware read at the address. .

  When the process of expanding the boot firmware into the main memory 5 is completed in step S213, the CPU 2 moves the process to step S214, and sets the pointer to the system memory table and the value (setting value) of the ROM firmware operation mode designation flag as parameters. Control to the boot firmware.

  Here, the process of passing control to the boot firmware using the pointer to the system memory table and the value of the ROM firmware operation mode designation flag as parameters, for example, to the system memory table at a predetermined address in a memory area accessible by the boot firmware. And a ROM firmware operation mode designation flag are stored, and an instruction start address of the main memory 5 where the boot firmware is expanded is read out and set in a PC (Program Counter).

  Through the processing described above, the boot firmware is expanded on the main memory 5, and thereafter, the CPU 2 operates according to the boot firmware command.

FIG. 3 is a flowchart showing the boot firmware process according to the embodiment of the present invention.
In step S214 shown in FIG. 2A, when the control of the CPU 2 shifts from the ROM firmware to the boot firmware, the CPU 2 shifts the process to step S301, and a pointer from the ROM firmware to the system memory table and the ROM firmware operation mode designation flag. Get the value of as a parameter.

  Here, for example, in step S214 shown in FIG. 2A, the ROM firmware stores a pointer to the system memory table and a ROM firmware operation mode designation flag at a predetermined address in a memory area accessible by the boot firmware. By referring to the predetermined address, the CPU 2 can acquire the pointer to the system memory table set by the ROM firmware and the ROM firmware operation mode designation flag as parameters.

  When the CPU 2 acquires the pointer of the system memory table and the ROM firmware operation mode designation flag, the CPU 2 shifts the processing to step S302, and uses the acquired pointer of the system memory table and the ROM firmware operation mode designation flag as the boot firmware of the nonvolatile memory 4. Stored in an area for storing a pointer of the system memory table in an area for storing environment variables and an area for storing a boot firmware operation mode designation flag.

  In step S303, a memory management table for managing the boot firmware memory is created from the system memory table received as a parameter.

  When the memory management table of the boot firmware is created, the CPU 2 shifts the control to step S304, and the boot firmware driver (for example, a driver for accessing the external recording device 6 in which the OS loader and the OS kernel are stored) (Such as a SCSI driver) or a driver for operating a network device to enable network communication) and loading and initialization processing.

  Then, the OS loader is read from the external recording device 6 and expanded in the main memory 5 (step S306), and the control of the CPU 2 is transferred from the boot firmware to the OS loader (step S306).

In this case as well, the instruction start address of the main memory 5 in which the OS loader is expanded may be read and set in the PC as in the process of step SS214 in FIG. 2A.
Through the processing described above, the OS loader is expanded on the main memory 5, and thereafter, the CPU 2 operates according to the instruction of the OS loader.

FIG. 4 is a flowchart showing the processing of the OS loader and the OS according to the embodiment of the present invention.
In step S306 shown in FIG. 3, when the control of the CPU 2 shifts from the boot firmware to the OS loader, the CPU 2 shifts the process to step S401, reads the OS kernel data from the external recording device 6, and stores it in the main memory 5. Expand (store).

  When the process of expanding the OS kernel data to the main memory 5 is completed, the CPU 2 shifts the process to step S402, arranges the OS kernel data expanded in the main memory 5 in an executable format, and transfers control to the OS kernel. (For example, the CPU 2 is executed from the head address of the OS kernel instruction).

When the control of the CPU 2 shifts to the OS kernel, the process shifts to step S403, and the OS kernel initialization process is performed.
When the initialization processing of the OS kernel is completed, the CPU 2 calls the boot firmware service routine, and acquires the boot firmware operation mode designation flag from a predetermined address in the area where the boot firmware environment variable of the nonvolatile memory 4 is stored. (Step S404).

  Here, the boot firmware service routine is an interface between the boot firmware and the OS kernel, and is a program that makes the boot firmware function available from the OS kernel.

  In step S405, if the boot firmware operation mode designation flag is not in the dump operation mode, the process proceeds to step S406, and normal operation processing is performed as the system operation mode.

If the boot firmware operation mode designation flag is the dump operation mode in step S405, the process proceeds to step S407, and the boot firmware service routine is called to obtain the physical address information of the main memory area to be dumped. get.
Then, the process proceeds to step S408, and the memory dump process of the main memory area 1 is executed.

In step S409, the boot firmware service routine is called to set the firmware operation mode designation flag stored in the nonvolatile memory 4 to the system operation mode.
When the above memory dump process is completed, the CPU 2 shifts the process to step S410, calls a restart process, and restarts the system in the system operation mode.

  FIG. 5 is a diagram showing a specific example of a process in which the dump program of the OS kernel operating in the second main memory area shown in step S407 of FIG. 4 determines the dump target area in the first main memory area. is there.

  In the figure, a first main memory area and a second main memory area arranged in the main memory 5 shown in FIG. 1 are shown. Further, since the OS is activated in the dump operation mode, the access to the first main memory area is impossible in principle, and only the second main memory area is accessible. Therefore, the OS can access the physical address of the first main memory area only by the boot firmware service routine or the ROM firmware service routine activated by the dump program.

  In the second main memory area shown in the figure, the OS kernel is expanded into a kernel core area 7a, a kernel physical memory discontinuous area 7b, and a kernel dynamic allocation area 7c. 2 operates in accordance with instructions of the OS kernel developed in the main memory area 2.

  Also in the first main memory area shown in the figure, the OS kernel is expanded into a kernel area 9a, a kernel physical memory discontinuous area 9b, and a kernel dynamic allocation area 9C. The OS kernel is expanded before the OS kernel expanded in the second main memory area is started (restarted).

  Hereinafter, a process of determining a desired dump target area expanded in the first main memory area by the dump program of the OS kernel expanded in the second main memory area will be described.

  (1) When a text area or a data area (hereinafter simply referred to as a static text area or a static data area) assigned to an area where physical addresses are continuous, such as the kernel core area 9a, is to be dumped first, Since the logical address of the static data 8 and the logical address of the static data 10 in the kernel are the same if they are the same kernel, the boot firmware service routine is called from the OS kernel, and the kernel in the first main memory area is called. The base address of the core area 9a (the head address of the kernel core area 9a) is acquired.

  Then, a value (offset 15) is calculated by subtracting the head address of the kernel core in the first main memory area from the head address of the kernel core area 7a in the second main memory area.

  Further, the logical address of the static data 8 in the kernel of the second main memory area is converted into a physical address based on the address conversion table (system memory table), and the offset 15 is subtracted, whereby the second main memory area The physical address of the static data 10 in the first main memory area corresponding to the static data 8 is obtained. The same processing is performed for the static text area as well, based on the static text area of the second main memory area, the physical address of the static text area in the first main memory area to be dumped (for example, , The physical address of the dump target area 12).

  Further, normally, since the offset 15 is equal to the start address of the second main memory area minus the start address of the second main memory area, the static data area and the static text in the kernel core area 7a. If it is an area, the base address information of the first main memory area and the second main memory area and the address conversion table (system memory table) in the second main memory area operating in the dump operation mode are used. The address of the static data area or text area of the kernel core 9a which is the first main memory area which is the dump target area can be easily obtained.

  (2) Since the physical address is assigned to a discontinuous area, the calculation based on the offset 15 described in the above (1), the calculation in the first main memory area based on the second main memory area information When the physical address of the dump target area (for example, the dump target area 13 and the dump target area 14) cannot be calculated or the logical address itself is dynamically allocated instead of the static data area, the second main memory area When dump processing is performed on a data area in the kernel that is located at a logical address different from that during operation, a system memory table (address conversion) in the kernel of the first main memory area is used. The corresponding physical address may be calculated by referring to the data in the table (11).

Here, the physical address of a place in the system memory table 11 in the kernel core area 9a in the first main memory area can be obtained by the process (1).
Here, the kernel core areas 7a and 9a are basic parts of the kernel text (static text area) and kernel data (static data area), and the data to be linked to various control data in the kernel is stored in the kernel core areas 7a and 9a. It is defined that physical addresses are assigned to consecutive areas.

  The above-described processing based on FIGS. 2A to 5 is processing until the OS is started due to power-on of the information processing apparatus 1 or reboot due to OS panic, but in FIG. 6, the OS is abnormal. The process from when the process is detected until the restart process is performed will be described.

FIG. 6 is a flowchart illustrating an example of processing from when the OS detects an abnormality until restart processing is performed.
In the state where the information processing apparatus 1 (OS) is operating in the system operation mode, when a fatal hardware failure occurs in order to continue system operation such as memory destruction, an interrupt signal from the hardware or the OS kernel In response to the error notification from the OS, the OS detects an abnormality (step S601).

When the OS detects an abnormality, the CPU 2 shifts the control to step S602 and performs a dump process entry. For example, when an interrupt handler is activated in response to an interrupt signal from hardware, the instruction start address of the dump program is set in the PC provided in the CPU 2, and the CPU 2 operates according to the instruction of the dump process entry program.
Note that in the case of an OS in which a plurality of programs (processes, tasks, and applications) operate at the same time, programs other than the dump processing entry program also operate naturally.

  Then, the process proceeds to step S603, the boot firmware service routine is called from the dump processing entry program, and the ROM firmware stored at a predetermined address in the ROM firmware environment variable area secured in the nonvolatile memory 4 The operation mode specification flag or the ROM operation mode specification flag and the firmware operation mode specification flag stored at a predetermined address in the firmware environment variable area are set to the dump operation mode.

  When the process for setting the dump operation mode is completed, the CPU 2 shifts the control to step S604, calls the restart process, and starts the restart. That is, the process proceeds to step S201 illustrated in FIG. 2A, and the restart process (reset process) is started.

  Further, in the figure, in the process of step S603, the process of setting the dump operation mode is performed by the firmware service routine. However, the ROM firmware is prepared by preparing an interface (means) for directly accessing the ROM firmware from the OS. The dump operation mode may be set in the operation mode designation flag, or the ROM firmware operation mode designation flag and the firmware operation mode designation flag.

FIG. 7 shows a modification of the process from when the OS detects an abnormality until the restart process is performed.
As in FIG. 6, when a fatal hardware failure occurs in order to continue system operation such as memory corruption in the state where the information processing apparatus 1 (OS) is operating in the system operation mode, The OS detects an abnormality by an interrupt signal or an error notification from the OS kernel (step S701).

When the OS detects an abnormality, the CPU 2 shifts the control to step S702 and makes a dump process entry (starts the dump process entry program).
Then, the process proceeds to step S603, and the ROM firmware service routine is called from the dump process entry program, and one or all of the ROM firmware operation mode designation flag and the firmware operation mode designation flag reserved in the nonvolatile memory 4 are set. Set the dump operation mode.

  When the dump operation mode is set, the CPU 2 shifts the control to step S704, calls a restart process, and starts restart. That is, the process proceeds to step S201 illustrated in FIG. 2A, and the restart process is started.

  Further, power control means (not shown in FIG. 1) (for example, a system monitoring mechanism that performs power control or monitoring) sets the ROM firmware operation mode designation flag and the firmware operation mode designation flag to the dump operation mode when the system is started. May be.

  FIG. 8 shows an example of processing from when the OS detects an abnormality until restart processing is performed when the power supply control means is provided with dump operation mode / system operation mode switching means.

  As in FIG. 6, when a fatal hardware failure occurs in order to continue system operation such as memory corruption in the state where the information processing apparatus 1 (OS) is operating in the system operation mode, The OS detects an abnormality by an interrupt signal or an error notification from the OS kernel (step S801).

When the OS detects an abnormality, the CPU 2 shifts the control to step S802 and performs a dump process entry (starts the dump process entry program).
Then, the process proceeds to step S803, and the ROM firmware operation mode designation flag is set from the dump process entry program to the system monitoring mechanism that performs power supply control at the next system startup (the process shown in FIGS. 2A to 4), or The ROM firmware operation mode designation flag and the firmware operation mode designation flag are notified to be the dump operation mode.

  For example, a memory area referred to by a program having the function of the system monitoring mechanism is reserved in advance in the nonvolatile memory 4, and a system monitoring mechanism operation mode designation flag is stored at a predetermined address in the memory area (for example, If 1 byte data is 0xFF, the dump operation mode is set, and other than 0xFF is the system operation mode).

  Then, as shown in FIG. 6, the OS calls the boot firmware service routine to set the dump operation mode in the system monitoring mechanism operation mode designation flag.

  On the other hand, the system monitoring mechanism refers to the system monitoring mechanism operation mode designation flag when the system is started (for example, step S201 shown in FIG. 2A). The designation flag, or the ROM firmware operation mode designation flag and the firmware operation mode designation flag may be set to the dump operation mode.

  When the dump monitoring mode is notified to the system monitoring mechanism, the CPU 2 shifts the control to step S804, calls the restart process, and starts the restart. That is, the process proceeds to step S201 illustrated in FIG. 2A, and the restart process is started.

  In addition, the system monitoring mechanism may, for example, set the ROM firmware operation mode designation flag, or the ROM firmware operation mode designation flag and the boot firmware operation mode designation flag to the dump operation mode and execute the reboot. Also good.

  As a result, even if the OS hangs up and control cannot be transferred to the dump process entry, the system can be restarted in the dump operation mode, so that dump collection can be performed.

  In the above description, the first main memory area and the second main memory area are divided in both the system operation mode and the dump operation mode. However, the present invention is not limited to this. . That is, all the areas of the main memory 5 are used as the first main memory area shown in FIG. 1 in the system operation mode, and the first main memory area and the second main memory area are only in the dump operation mode. It may be divided into In this case, an area that is less frequently used in the first main memory area or an area that is not necessary for failure analysis may be allocated as the second main memory area.

  When analyzing the failure of the data collected by the dump processing described above, it is necessary to refer to a kernel-specific table (for example, a table in which data allocation is defined) in order to determine an area necessary for failure analysis. However, since the kernel operating in the dump operation mode is the same as the kernel operating in the system operation mode, the same table is used to obtain information such as the address where the predetermined data is located. It is possible to perform failure analysis easily.

  In addition, since the system dump collection program operates in the same kernel (operating system) as in the system operation mode, table information for managing the area on the main memory to determine the dump collection area (eg, memory management table) It is not necessary to create a list of areas from which system dumps are to be obtained in advance, and it is necessary to manage the version number of the kernel and the version number of the area list (or the version number of the table information for creating the area list) Disappears.

  Therefore, due to a mismatch between the kernel version number and the area list version number (or the version number of the table information used to create the area list), an incorrect area is dumped or an unimplemented area in memory is accessed. Therefore, it is possible to avoid the system dump collection from being terminated abnormally. In addition, there is no need to perform unnecessary version management.

  Further, since the dump collection program incorporated in the kernel operates at the time of system dump, it is possible to collect the optimum area necessary for failure analysis from the memory data at the time of failure occurrence.

  Since the reset is executed once to collect the dump, it is possible to collect the dump on a stable operating system. It is possible to analyze the data and identify the cause of the failure and the suspected part to some extent when the failure occurs. In addition, it is possible to cope with such as further expanding the dump collection area according to the analysis result at that time.

  Also, in the dump operation mode, the operating system is started in a physical memory space different from that in the system operation mode, including the boot firmware that boots the operating system, so an abnormality occurs at the interface between the operating system and the boot firmware. Even if it hangs up, it is possible to dump without destroying the data in the boot firmware area when an error occurs during dump collection.

There is no need to temporarily save memory data on the boot firmware side because the data in the memory area used in the system operation mode to be dumped is not overwritten, and there is a dedicated hardware resource for temporarily saving memory data. Since it is not necessary, it is possible to reduce costs while increasing reliability. Therefore, it is not necessary to secure a dedicated partition for temporarily saving memory data, so that disk partition management becomes easy.
Furthermore, dump collection can be performed regardless of the environment in which the failure occurred (kernel control data inconsistency, memory corruption, etc.).

  In addition, if a temporary hardware abnormality has occurred, it is possible to increase the possibility of normal operation by resetting it temporarily. Even if a permanent hardware abnormality has occurred, it is possible to It is possible to increase the possibility that an abnormality can be detected by POST diagnosis or hardware initialization processing executed when the system is restarted.

(Appendix 1)
In accordance with operation mode designation information that controls the operation of the information processing apparatus, a second area that maps the storage means area of the information processing apparatus to a second area other than the first area used for the operating system An area definition information acquisition process for acquiring the second area definition information from the area definition information means storing the area definition information;
An operating system storage process for storing an operating system in an area based on the second area definition information;
A dump target area calculation process for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area;
Dump processing for reading and outputting information stored in the dump target area;
A program for causing a computer to execute a panic dump collection process.
(Appendix 2)
A failure notification process for detecting and notifying that a failure has occurred in a component required for the information processing apparatus to operate; and
An operation mode designation information storage process for storing the operation mode designation information in the operation mode designation information storage means in response to the notification by the failure notification process;
A restart process for restarting the information processing apparatus while maintaining the state of the area used when the failure occurred;
The program for making a computer perform the panic dump collection process of Additional remark 1 provided before the said area | region definition information acquisition process.
(Appendix 3)
In the dump target area calculation process, the dump target area is determined based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The program for collecting a panic dump according to appendix 1 or 2, characterized by being a calculation process.
(Appendix 4)
In the dump target area calculation process, the dump target area is determined based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The program for collecting panic dump according to any one of appendices 1 to 3, wherein the program is acquired based on information to be calculated and information for managing the first area .
(Appendix 5)
A second area that demarcates the area of the storage means included in the information processing apparatus to a second area other than the first area used in the operating system in accordance with operation mode designation information that controls the operation of the information processing apparatus. A region definition information acquisition process for acquiring the second region definition information from the region definition information means storing the region definition information;
An operating system storage process for storing an operating system in an area based on the second area definition information;
A dump target area calculation process for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area;
Dump processing for reading and outputting information stored in the dump target area;
How to collect panic dump.
(Appendix 6)
A failure notification process for detecting and notifying that a failure has occurred in a component required for the information processing apparatus to operate; and
An operation mode designation information storage process for storing the operation mode designation information in the operation mode designation information storage means in response to the notification by the failure notification process;
A restart process for restarting the information processing apparatus while maintaining the state of the area used when the failure occurred;
The panic dump collection method according to appendix 5, further comprising: before the area definition information acquisition process.
(Appendix 7)
In the dump target area calculation process, the dump target area is determined based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The panic dump collection method according to appendix 5 or 6, wherein the panic dump collection method is a calculation process.
(Appendix 8)
In the dump target area calculation process, the dump target area is determined based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The panic dump collection method according to any one of appendices 5 to 7, which is processing acquired based on information to be calculated and information for managing the first area.
(Appendix 9)
In accordance with operation mode designation information that controls the operation of the information processing apparatus, a second area that maps the storage means area of the information processing apparatus to a second area other than the first area used for the operating system Area definition information acquiring means for acquiring the second area definition information from the area definition information means storing the area definition information;
Operating system storage means for storing an operating system in an area based on the second area definition information;
Dump target area calculating means for calculating a predetermined area in the first area corresponding to a predetermined area in the area based on the second area definition information as a dump target area;
Dumping means for reading out and outputting information stored in the dump target area;
A panic dump collection mechanism.
(Appendix 10)
A failure notification means for detecting and notifying that a failure has occurred in a component required for the information processing apparatus to operate;
An operation mode designation information storage means for storing the operation mode designation information in response to the notification by the failure notification means;
Restarting means for restarting the information processing apparatus while maintaining the state of the area used when the failure occurred;
The panic dump collection mechanism according to appendix 9, further comprising:
(Appendix 11)
The dump target area calculation means determines the dump target area based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The panic dump collection mechanism according to appendix 9 or 10, wherein the panic dump collection mechanism is a calculation process.
(Appendix 12)
The dump target area calculation means determines the dump target area based on a difference between an offset between the start position of the first area and the start position of the second area and a predetermined area in the second area. The panic dump collection mechanism according to any one of appendices 9 to 11, which is a process to be acquired based on information to be calculated and information for managing the first area.

1 Information processing device 2 CPU
3 ROM
4 Nonvolatile memory 5 Main memory 6 External recording device 7a Kernel core area 7b Kernel physical memory discontinuous area 7c Kernel dynamic allocation area 8 Static data 9a Kernel core area 9b Kernel physical memory discontinuous area 9c Kernel dynamic allocation area 10 Static data 11 Address information 12 Dump target area A
13 Dump target area B
14 Dump target area C
15 offset

Claims (4)

A control method for an information processing apparatus capable of operating the same operating system in a system operation mode for executing a normal system operation operation and a dump operation mode for collecting dump information,
The arithmetic processing unit of the information processing apparatus executes a program,
When starting the information processing apparatus, refer to information indicating an operation mode of the information processing apparatus,
When the information indicates a system operation mode, the first area among the areas included in the memory of the information processing apparatus is set as an area used for the startup process of the information processing apparatus, while the information is If the dump operation mode is indicated, a second area different from the first area among the areas included in the memory is set as an area used for the startup process of the information processing apparatus,
Perform initialization and diagnosis of the set area,
Start the information processing apparatus using the set area,
When the information indicates the dump operation mode, the offset calculated from the difference between the base addresses of the first area and the second area, and the second operating in the dump operation mode. A dump process for reading out information stored in the first area calculated using the address of the area corresponding to the dump target area in the first area, and the dump process A control method, wherein the information processing apparatus is restarted after completion of the operation.   The operation processing unit restarts the information processing apparatus after updating the information to information indicating a system operation mode when the dump process is completed by executing a program. The control method described in 1. The arithmetic processing unit develops an operating system in the set area after executing initialization and diagnosis of the set area by executing a program,
Characterized in that activating the information processing apparatus by operating the expanded the operating system, control method according to claim 1 or 2. A control method for an information processing apparatus capable of operating the same operating system in a system operation mode for executing a normal system operation operation and a dump operation mode for collecting dump information,
The arithmetic processing unit of the information processing apparatus
By executing the first program that executes the startup process of the information processing apparatus,
When starting the information processing apparatus, refer to information indicating an operation mode of the information processing apparatus,
When the information indicates a system operation mode, a first area of the areas included in the memory of the information processing apparatus is set as an area used for the startup process of the information processing apparatus, and the information is When the dump operation mode is indicated, the management information in which the second area different from the first area among the areas included in the memory is set as an area to be used for the startup process of the information processing apparatus Create
Perform initialization and diagnosis of the set area,
Expand the operating system on the set area to start the information processing apparatus,
Passing control from the first program to the operating system;
By executing the operating system,
When the information indicates the dump operation mode, the offset calculated from the difference between the base addresses of the first area and the second area, and the second operating in the dump operation mode. A dump process for reading information stored in the first area calculated using the address of the area corresponding to the dump target area in the first area,
A control method, comprising: restarting the information processing apparatus after completion of the dump process.

Images