DE102004047363A1 - Processor or method for operating a processor and / or operating system in the event of a fault - Google Patents

Abstract

The invention relates to a method for operating a processor (CPU) and / or operating system (OS), in which the processor (CPU) is operated in an execution mode (N) for a normal operation, wherein in the event of a failure in a repair mode (R) is switched to eliminate the disturbance (f1, f3), and the normal operation in the execution mode (N) is continued after eliminating the disturbance (f7, f8).

Description

The The invention relates to a method of operating a processor and / or operating system with the preamble features of the claim 1 or to a processor with the above-mentioned features of Patent claim 10.

One Processor in a computer or a modern mobile device corresponding operating system will be in an execution mode for one normal operation. A current program malfunction often causes an interruption of execution and a crash of Runtime environment. If the runtime environment is in a monitoring mode, which is also known as kernel mode or privileged level 0 when the machine or processor enters in the event of a fault a fault mode, which is also referred to as panic mode, and is restarted or "frozen", what a turn off required to allow a cold restart. If the runtime environment is an application that is not in the Kernel mode is operating crashes dependent from the storage model and the execution model of the central Processor and the operating system at least the disturbed or faulty Application. In the case of a simple model, the entire machine crashes as described above.

Known is the saving or saving of the execution context in the moment of disturbance, the execution context with the disturbed Condition data in a non-volatile Memory is storable. By external access by an operator is a later one Investigation of the cause of the fault possible.

From Programming development software is generally development programs Known in which a compiler and a debugger as a single Development program are merged, in the case of a compile error or a subsequent runtime error a crash of the compiled program to optimize the program development to prevent.

The The object of the invention is a processor or a method for operating a processor in the event of a failure in normal operation to improve.

These Task is by the method of operating a processor and / or operating system with the features of claim 1 or solved by a processor having the features of claim 10.

Prefers Accordingly, a method for Operating a processor and / or operating system in which the Processor in an execution mode for one normal operation is operated, in case of failure in a repair mode is switched to eliminate the disturbance, wherein normal operation in execution mode after elimination the fault continues becomes.

Device is preferred a processor with an interface for operating system access in an execution mode to run a normal operation. Advantageously, the processor is through this or another interface for operating system access in a repair mode for execution in case of a fault to eliminate the error for the subsequent continuation of the operation in the execution mode.

advantageous Embodiments are the subject of dependent claims.

Prefers in particular, a method in which a runtime environment for the repair mode as independent Operating runtime environment is booted up.

Prefers in particular, a method in which in the event of disruption in switched to a fault mode will and from the fault mode after saving the execution context is switched to the repair mode.

Prefers is particularly a method in which the repair mode in the failure mode and from there into execution mode changed back becomes.

Prefers In particular, a method in which a Troubleshooting by means of the rescued execution context carried out will and the suppressed Data as execution context be restored to the processor and / or a register.

Prefers In particular, a method in which a defective in the repair mode Value or state corrected in the processor and / or in a register will, so a continuation of the execution mode allows becomes.

In particular, a method is preferred in which, in the repair mode, an erroneous value or state in the processor and / or in a register is overwritten with a value or state, which allows a continuation of the execution mode.

Prefers in particular, a method in which in the repair mode for troubleshooting a repair algorithm dependent from a determined cause of the fault is started.

Prefers In particular, a method in which in the execution mode, the operation is optional is continued before or after the operating state of the fault condition.

Prefers In particular, a processor with an interface to at least an operating state memory for storing a current fault operating state of the disturbed Operating sequence and for restoring an undisturbed normal operating state after a fault repair by the repair mode.

Prefers in particular, a processor with a repair database memory for storing various repair algorithms for different types of errors.

Prefers In particular, such a method and / or such a processor, if the disorder a disorder which is a normal continuation of a crash of the Runtime environment, crash the processor, or crash that of the entire system would or would require a cold restart.

Prefers In particular, a mobile device for performing a such method and / or with such a processor. A use of the Process or the processor is also in other applications advantageous. In addition to the use in conventional computers offers itself especially in mobile devices like mobile computers, PDAs etc., which over an external interface connection with other devices communicate. Looking at an online application becomes the user enabled with the repair algorithm or repair mode to react and interact. Ultimately, crucial for use the process or the processor is also how to maintain and maintain the appropriate device is and what advantages in an individual case can be achieved.

According to the method and device thus becomes a solution preferred, which provides a mechanism that after a particularly serious Program malfunction, which was detected becomes active and a Error correction before a crash of the entire system or the Processor with the operating system allows. It is a matter of a repair mechanism, which is a particularly minimal execution environment as a repair environment starts up or activated, if the repair environment already at an earlier time was raised.

To In activation the repair environment forms a malfunction type or an error code to a corresponding repair algorithm in a repair algorithm database, performs the repair algorithm and causes a continuation the normal operation in the execution mode after the trouble shooting. Unless the troubleshooting using a stored or saved execution context processor contents and processor registers, etc., become the execution context or the corresponding data after the fault elimination again in the processor or the corresponding registers, etc. reloaded. An otherwise inevitable after activation of the fault or panic mode A crash or a cold restart can be avoided become. After leaving the repair mode, the execution of the Operation in the previously defective program again included in the program section where the malfunction or disruption otherwise crashes would have caused.

The Such repair algorithms can be very diverse be. In a simple case, a text message can be used to inform user's output before the crash occurs workarounds to propose to avoid the crash. More elaborate repair algorithms can advantageously, correct the program at the error location and the execution the normal operation of this program without a crash Resume.

described thus becomes a mechanism which bypasses software defects allows and availability the machine or the device improved with the appropriate method or processor. Especially Mobile facilities should not crash and have the best possible availability Offer. As usual, however There is no perfect software and errors leading to a crash to lead, occur, the use of such mechanisms is advantageous.

One embodiment will be explained in more detail with reference to the drawing. Show it:

1 schematically individual components in connection with runtime environments of a processor; and

2 schematically a process flow to illustrate an exemplary repair process to eliminate a fault.

1 exemplifies a processor CPU with three symbolic processor sections for symbolizing various operations. The operations are an execution mode N for executing a normal operation and a panic mode P for saving processor states and register states in the event of a failure which, when the operation progresses, crashes the execution mode, the processor or even of the overall system. Execution mode N and disturbance mode P work in a manner known per se in such a processor CPU or under a corresponding operating system OS.

Also provided is additional a repair mode R. This is for repairing or removing a fault or an error that would otherwise cause the system to crash. the Repair mode R is expediently assigned a separate runtime environment, which as independent Operating runtime environment is booted up. The boot up can in this case, for example, with the starting of the device in which the processor CPU is used, done. Booting up the repair mode R may, however, advantageously also only when starting a critical program, an update, or after activating the failure mode P performed become.

the Repair mode R is preferably in a memory M, a repair database RDB, in which various repair algorithms Alg1, Alg2 are stored, on which the repair mode R depending on the determined Cause of error or cause of fault to initiate appropriate measures can help prevent a crash.

In usual In one way, the processor comprises CPU registers, in particular processor registers MC, in which for needed the normal operation Data will be deposited. about a bus B is also a connection to an external memory M given. In the external memory M or possibly also a Memory integrated in the processor CPU are memory areas for Saving the operating system OS, the repair database RDB as well an execution context memory S for storing an execution context ps at the moment of a fault or previously provided.

To Start of operation, the operating system OS is loaded into the processor CPU, around this in execution mode N to operate. In case of a fault, especially a fault, which lead to a crash would, becomes the execution mode N interrupted and in the fault mode P switched f1. In fault mode P becomes the execution context ps in a subsequent step f2, as far as possible, saved or in the execution context memory S stored. Subsequently, in the repair mode R is switched f3 to the fault, as far as possible, to eliminate. In repair mode R, the stand-alone operating runtime environment is booted, unless they have already started up earlier has been. In addition, will a repair algorithm started. Preferably depends on a detected or determined error or the cause of the fault from repair database RDB a suitable repair algorithm Alg1, Alg2 loaded and in the operating runtime environment of the repair mode R executed. Thereby can the disorder directly in the processor or processor registers MC etc. or in the Image of the disturbed execution context ps in the execution context memory S are processed f5. The following is preferably with interposition of the fault mode P in the execution mode N backwards f7, f8. In execution mode N can then continue the normal operation, without that it crashes.

2 FIG. 12 illustrates an example flow of a method in the area of the execution mode N, the failure mode P, and the repair mode R in the case of a failure of the normal operation. Shown in particular is the relationship between a repair manager running in the repair mode R in the corresponding mini-runtime environment of the repair mode R, the repair mode being started from the panic context P in which N is entered in the event of a malfunction from the execution mode. Also shown are exemplary major tasks in each of the modes and the execution of a repair algorithm in repair and execution mode R, N. In an exemplary application for commercial purposes, the exemplary code below is to be implemented:
"If" card account "= 0;
then start "empty"
otherwise start "buy"
fi ".

Thus, a routine is provided which, in the case of an empty card account, starts either an "empty" routine or a "buy" routine. For example, if the routine "empty" does not have a correct pointer to a suitable memory area, it will crash, but instead of abruptly terminating the application, the repair manager R will be started in its minicontext, retrieving the error code and thus retrieving the error code Repair algorithm table or repair database indexed to load and execute a corresponding repair algorithm Alg1.

In the case, that the repair algorithm Alg1 determines that the faulty Pointer the cause for the malfunction or malfunction could be different Actions performed become.

According to one first variant, the user of the corresponding device with be informed to the processor, for example, to be informed that his card account is empty, the user charge the card should and then try the last operation again. After issue The information then becomes a jump to the repair algorithm End fi made the actual algorithm, making a sequel in execution mode N without a crash allows becomes f7b. Advantageously, an error report can also be recorded and optionally to e.g. the manufacturer of the device or sent to the issuer of the account card become. To this embodiment too support, should the explanation the function will be created in conjunction with a crash text hint, which by the repair manager or by the repair mode R is used.

According to one second variant could a text issued to the user also merely informs the user about that inform incorrectly working program and then to the end the application or at the end of the program without jumping respectively. This may be due to incorrect registry entries or processor entries the repair algorithm Alg1 before continuing the execution mode N corrected.

According to one third exemplary variant can by the repair algorithm Alg2 the faulty code of the faulty program will be repaired, for example, by replacing the faulty pointer by the correct address of the function or routine "empty." This correct address can be, for example due to automatic learning from previously correct executions of the program. Below is a sequel possible at the point at which the error or the fault occurred. This means, the operation in execution mode N will be at the incident continued with a correct value f7a.

According to 2 Thus, for example, a program flow in execution mode N and a malfunction of the program sequence are assumed. It is changed to the fault mode P f1, by which the execution context stored ns and possibly the crash site is completely saved or stored. It follows the start of the repair routine f3, wherein preferably an error type with the subsequent repair mode R is passed. In the repair mode R, a suitable repair algorithm Alg1 is called, executed and, after repair, a suitable execution context ns is stored back. Thereafter, the repair mode R is returned to the trouble mode P f6. From fault mode P or possibly also directly from repair mode R, the execution context is restarted and the execution mode N is advanced f7. The continuation of the erroneous program in execution mode N is possible f7a or f7b at the end fi of the faulty program, depending on the nature of the repair at the faulty and corrected location of the program algorithm.

Claims (14)

Method for operating a processor (CPU) and / or operating system (OS), in which the processor (CPU) is operated in an execution mode (N) for a normal operating sequence, characterized in that in the event of a failure in a repair mode (R) is switched to eliminate the fault (f1, f3), wherein the normal operation in the execution mode (N) after the elimination of the fault is continued (f7, f8). The method of claim 1, wherein for the repair mode (R) a runtime environment as a standalone operating runtime environment is raised. Method according to Claim 1 or 2, in which case the disorder in a fault mode (P) is switched (f1) and the fault mode (P) after a rescue the execution context (ps, f2) is switched to the repair mode (R) (f3). Method according to any preceding claim, in which from repair mode (R) to fault mode (P) and from there into the execution mode (N) changed back becomes (f6, f7). Method according to claim 3 or 4, wherein in the repair mode (R) a troubleshooting by means of the rescued execution context (ps) performed becomes (f4) and the suppressed Data as execution context (ns) in the processor (CPU) and / or a register (MC) restored become (f6 *). Method according to any preceding claim, in which in repair mode (R), an erroneous value or state in the processor (CPU) and / or in a register (MC) is corrected (f5), so that a continuation of the execution mode allows becomes (f7a, f7b). Method according to any preceding claim, wherein in the repair mode (R) an error Any value or state in the processor (CPU) and / or in a register (MC) is overwritten with a value or state (f5), which allows a continuation of the execution mode (N) (f7a, f7b). Method according to any preceding claim, in which in repair mode (R) for troubleshooting a repair algorithm (Alg1, Alg2) depending on a determined malfunction cause is started (f4). Method according to any preceding claim, in which in execution mode (N) the operation optionally before (f7a) or after (f7b) the operating state of the fault condition will continue. Processor (CPU) with an interface (B) for a Operating system access in an execution mode (N) to execute a normal operation, characterized by this or another Interface (B) for an operating system access in a repair mode (R) to execute in case a fault to eliminate the error for the subsequent continuation (f7) of the operation in the execution mode (N). Processor according to claim 10 with an interface (B) to at least one operating state memory (M, F) for storing a current fault operating state (ps) of the disturbed Operating sequence and for restoring (f6 *) of an undisturbed normal operating state (ns) after a fault correction (f4) by the Repair mode (R). Processor according to claim 10 or 11 with a repair database memory (RDB) for storing various repair algorithms (Alg1, Alg2) for various Error types. Method according to one of claims 1 - 9 and / or processor according to one of the claims 10 - 12, being the disorder a disorder which is a normal runtime crash that crashes the runtime environment, a crash of the processor or a crash of the lead the entire system would or would require a cold restart. mobile device to perform A method according to any one of claims 1-9 and / or with a processor (CPU) according to one of claims 10 - 12.