Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
2000-04-14
2004-04-27
Baderman, Scott (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C714S048000, C714S002000, C714S025000
Reexamination Certificate
active
06728907
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to computer system crash analysis. More specifically, the invention relates to the identification of a component responsible for a computer system crash.
BACKGROUND OF THE INVENTION
Today, diagnosing a computer system crash (due to operating system or device driver software bugs, hardware errors, configuration problems, or the like) is a very time consuming and expensive process. Typically, a system administrator or developer is left to access books, websites, or colleagues, and often resorts to trial and error to determine what exactly caused the system crash. The diagnosis is generally manual and involves setting particular diagnostic configurations, rebooting the system (likely many times), manually evaluating the diagnostic results, and attempting to reproduce the crash.
In some operating systems, when a crash occurs, a dump file may capture the operating state of the computer at the time of the crash. The traditional dump file helps solve the mystery of what caused the crash, but is typically a very large file. For instance, large systems may have several gigabytes of memory. Writing out the traditional dump file may take upwards of thirty minutes on such a system. Users typically disdain that much down time, and administrators prefer to avoid such time-consuming steps toward diagnosing the system crash.
Moreover, as suggested above, using the information stored in the dump file has traditionally been a time-intensive, manual process. A system administrator or developer is left to read many lines of information in an attempt to determine what caused the crash. Hours of human intervention may be spent simply identifying the diagnostic steps to be taken in search of the offending component that caused the crash.
Further complicating the diagnosis of system crashes is that they are often difficult to reproduce. For example, a device driver may have a bug that does not arise unless memory is low, and then possibly only intermittently. In that case, a test system may not be able to reproduce the error because it does not reproduce the conditions.
In sum, diagnosing system crashes has long vexed system administrators and users of computing systems. A system that overcomes the problems identified above has eluded those skilled in the art.
SUMMARY OF THE INVENTION
Briefly described, the present invention provides a system and method for self-diagnosing system crashes by identifying a type of system crash that occurred, and automatically taking diagnostic steps based on that type of crash. The invention may make use of a stop code contained in a memory dump file stored in response to the system crash. Preferably, the invention makes use of a “minidump” that contains an intelligently selected subset of the available pre-crash computer information, including the stop code that identifies the particular type of crash that occurred.
In one implementation, a mechanism of an operating system is configured to write an abbreviated dump file of a selected portion of the system memory at the time of a system crash. For example, a “crash driver” may be implemented that, when instructed by the operating system, reads from system memory certain information considered to be likely the most relevant to the diagnosis of a system crash, and writes that information to the dump file. Typically, a component of the operating system (e.g., a memory manager component) identifies the occurrence of a system fault, such as corrupt or exhausted memory, and informs the operating system that the system crash has occurred. In response, the crash driver may be instructed to write the dump file so that the crash may be diagnosed.
In accordance with an aspect of the invention, another mechanism within an operating system, such as a memory management component of a system kernel, checks for the existence of the dump file at each startup of the machine. The existence of the dump file may indicate that the system crashed during the previous session. The existence of the dump file is but one technique that may be used to determine that a system crash occurred, and is only given as an example. In any case, once the occurrence of the system crash has been discovered, the mechanism of the invention analyzes the dump file to determine what type of crash occurred (e.g., out of memory or corrupt memory), and implements a self-diagnostic routine or procedure corresponding to the type of crash. More particularly, the mechanism may read the stop code from the dump file and implement a self-diagnostic procedure that corresponds to that stop code.
Through the described construct, the mechanism self-diagnoses the likely cause of the crash by automating many of the tasks normally performed manually. If the crash occurs again, the mechanism identifies, through the self-diagnostic procedures automatically implemented, the likely cause of the crash, e.g. the particular faulty driver or configuration error, and may report that information to a system administrator. This significantly simplifies the corrective measures that typically need to be taken by system administrator or the like to correct the fault. Moreover, the self-diagnostic procedure may enable special code to provoke the problem into reoccurring sooner, and, more importantly, to also catch it before it causes too much damage so the culprit can be easily identified. And still further, the invention enables non-experts to quickly diagnose and resolve computer problems, thereby ameliorating both the cost and delay of finding an “expert.”
In accordance with another aspect of the invention, during startup, the mechanism may change the stop code stored in the dump file to avoid a situation where the system suffers another, different type of crash before the mechanism is able to address the first crash (such as later in the startup process).
REFERENCES:
patent: 5043871 (1991-08-01), Nishigaki et al.
patent: 5355469 (1994-10-01), Sparks et al.
patent: 5948112 (1999-09-01), Shimada et al.
patent: 5999933 (1999-12-01), Mehta
patent: 6047124 (2000-04-01), Marsland
patent: 6070254 (2000-05-01), Pratt et al.
patent: 6101617 (2000-08-01), Burckhartt et al.
patent: 6163858 (2000-12-01), Bodamer
patent: 6170067 (2001-01-01), Liu et al.
patent: 6178528 (2001-01-01), Poisner
patent: 6226761 (2001-05-01), Berstis
patent: 6393560 (2002-05-01), Merrill et al.
patent: 6457112 (2002-09-01), Hostetter
Hendel Matthew D.
Wang Landy
Baderman Scott
Damiano Anne L.
Law Offices of Albert S. Michalik PLLC
Microsoft Corporation
LandOfFree
System and method for self-diagnosing system crashes does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for self-diagnosing system crashes, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for self-diagnosing system crashes will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3245430