DE102010052726A1 - Method for determining stability of configuration of computer system, involves reading configuration parameter via interface of firmware component, and determining stability of computer system based on parameter - Google Patents

Abstract

The method involves initializing a bus system that comprises a processor bus (8), memory bus (9), system management bus (10), chip set bus (11) and peripheral controller interface-express-bus (12) by a bus control using a preset hardware-specific test routine. A configuration parameter for operating the bus system is detected in an updated configuration based on the test routine by a firmware component. The detected parameter is read via an interface of the firmware component. Stability of a computer system (1) is determined based on the read parameter by an analysis component. Independent claims are also included for the following: (1) a computer system comprising a bus system (2) a computer program product comprising a computer program code for determining stability of configuration of a computer system.

Description

The invention relates to a method for determining a system stability of a system configuration of a computer system. Moreover, the invention relates to a computer system having at least one bus system with a bus controller and at least one bus component and a computer program product comprising program code for determining a system stability.

The quality and stability of a computer system is enhanced by the system components of the computer system and their interconnection. Both so-called on-board components, which are fixedly arranged on a system board of the computer system, and expansion components, which are plugged into, for example, bus slots of the system board, contribute to the overall stability. Due to the large number of different system components, such as processors, memory modules, interface cards and so on, it is usually necessary to check the system stability of the assembled overall system in addition to checking the functionality of the individual components.

Computer system manufacturers that assemble individual components into an overall system usually carry out a so-called production exit test. Known production exit tests are carried out as a so-called go or no-go test, in which it should be determined in as short a time as possible whether a computer system can be delivered in a specific system configuration or not.

For example, an operating system installed on the computer is booted and a test software is started within the operating system. The test software executes functional tests that check, for example, whether files can be written to a hard disk and read back.

With such tests, usually only so-called "hard" errors are found. A statement about the quality of the system stability or individual system components of the system configuration can not be made. In addition, due to time constraints in manufacturing outbound tests, it is usually not possible to vary other constraints on the test environment, such as an operating temperature or operating voltage of the computer system.

It is an object of the present invention to provide methods for determining system stability of a system configuration that provide improved, faster, and / or simpler information about the quality of system stability than known manufacturing benchmarks. In addition, a computer system and a computer program product are described which are suitable for carrying out the method.

• – Initialisieren wenigstens eines Bussystems durch eine Bussteuerung unter Verwendung wenigstens einer vorgegebenen, hardwarespezifischen Testroutine
• – Erfassen wenigstens eines Konfigurationsparameters zum Betrieb des Bussystems in einer aktuellen Konfiguration auf Grundlage der wenigstens einen Testroutine durch eine Firmwarekomponente
• – Auslesen des wenigstens einen erfassten Konfigurationsparameters über eine Schnittstelle der Firmwarekomponente und
• – Bestimmen einer Systemstabilität auf Grundlage des ausgelesenen wenigstens einen Konfigurationsparameter durch eine Analysekomponente:

According to a first embodiment, a method for determining a system stability of a system configuration is proposed with the following steps:

• - Initialize at least one bus system by a bus control using at least one predetermined, hardware-specific test routine
• - Detecting at least one configuration parameter for operating the bus system in a current configuration based on the at least one test routine by a firmware component
• - Reading the at least one detected configuration parameter via an interface of the firmware component and
• Determining a system stability based on the read at least one configuration parameter by an analysis component:

The above method takes advantage of results of a hardware-specific test routine performed during initialization of a bus system. In this case, configuration parameters for operating the bus system in a current configuration are detected by a firmware component and evaluated by an analysis component. For this purpose, the analysis component accesses the stored configuration parameters via an interface of the firmware component.

The above-mentioned method has the advantage, inter alia, that test routines which are executed anyway for the correct operation and for a necessary initialization of a bus system are also used for determining a system stability. By running the test routines at a very low, hardware-specific level of the computer system, it is also possible to detect those system configuration errors or deficiencies that can not be detected in a purely functional go or no-go test of known factory outbound tests.

According to an advantageous embodiment, the system stability is determined based on a deviation of the read-out at least one configuration parameter from a predetermined reference value. The determination of a deviation from a reference value allows the detection of so-called "soft" errors, such as the detection of particularly long bus access times or low data transmission rates, which in Not be recognized under a functional test.

According to a further advantageous embodiment, the steps of initializing, detecting and reading out are repeatedly carried out and the system stability is determined on the basis of a statistical evaluation of the at least one read back configuration parameter. The repeated execution of the method steps together with a statistical evaluation of the read configuration parameter allows detection of changes and in particular detection of a gradual deterioration of the system stability. In this way, for example, age-related error sources of the computer system can be detected in good time before a total failure.

According to a further advantageous embodiment, the predetermined, hardware-specific test routine comprises a training sequence for training a data transmission between the bus control and at least one bus component connected to the at least one bus system. Such training sequences, which are used in particular to set a bus timing, provide information about the correct function, performance and arrangement of individual bus components.

According to a second embodiment of the invention, a computer system is described which comprises at least one bus system with a bus controller and at least one bus component connected to the bus system, at least one firmware component with program code and an analysis component with program code. The firmware component includes program codes for initializing the bus system by the bus controller using at least one predetermined hardware-specific test routine comprising a training sequence for training data transfer between the bus controller and the bus component and capturing at least one configuration parameter for operating the bus system in a current configuration of the computer system the at least one test routine. The at least one analysis component has program code for reading out the at least one detected configuration parameter via an interface of the firmware component and for determining a system stability on the basis of the read-out at least one configuration parameter.

Such a computer system is suitable for carrying out the above-mentioned method. Furthermore, such a computer system is particularly suitable for determining system stability of a current system configuration independently of a special test environment. Thus, the system stability can be determined at any time. Such a computer system is particularly suitable for the continuous monitoring of a current system configuration, even after delivery of the computer system by its manufacturer when using the computer system by a user.

According to an advantageous embodiment, the firmware component and / or the analysis component are integrated in a BIOS component of the computer system. Thereby, the method for determining a system stability can be performed independently of an operating system and further test software of the computer system.

According to a further advantageous embodiment, the analysis component is integrated in a system management component of the computer system, which is set up for coupling to a maintenance device. Such a computer system can either be analyzed continuously or, if necessary, with a maintenance device. It is possible to draw conclusions about hard and soft errors as well as misconfiguration at the hardware level.

• – Auslesen wenigstens eines während einer Initialisierung wenigstens eines Bussystems durch eine Bussteuerung unter Verwendung wenigstens einer vorgegebenen, hardwarespezifischen Testroutine erfassten Konfigurationsparameters über eine Schnittstelle einer Firmwarekomponente und
• – Bestimmen einer Systemstabilität auf Grundlage des ausgelesenen wenigstens einen Konfigurationsparameters.

According to a third aspect of the present invention, a computer program product comprising program code for determining a system stability of a system configuration is described. The computer program product is characterized in that when the program code is executed by a processor of a computer system, the following steps are carried out:

• - Read at least one during an initialization of at least one bus system by a bus control using at least one predetermined, hardware-specific test routine detected configuration parameter via an interface of a firmware component and
• Determining system stability based on the read at least one configuration parameter.

Further advantageous embodiments of the invention are disclosed in the dependent claims and the following detailed description of exemplary embodiments.

The invention will be explained in more detail below using different embodiments with reference to the figures.

In the figures show:

1 an exemplary system configuration of a computer system,

2 a software architecture for implementing a method for determining system stability and

3 a flowchart of a method for determining a system stability.

1 schematically shows an exemplary system configuration of a computer system 1 , The computer system 1 includes in the embodiment a processor 2 , a hub module 3 as well as two memory modules 4a and 4b , Furthermore, the computer system includes 1 a BIOS module 5 and a so-called PCI bridge 6 , Finally, the computer system includes 1 expansion components 7a to 7c located in corresponding bus slots of the computer system 1 are plugged in.

In the embodiment, the processor 2 with the hub module 3 via a processor bus 8th coupled. Furthermore, the hub component 3 over a memory bus 9 with the memory modules 4a and 4b coupled. The hub module 3 is with the BIOS module 5 via a system management bus 10 and with the PCI bridge 6 via a chipset bus 11 coupled. The expansion components 7a to 7c are with the PCI bridge 6 via a PCI Express bus 12 coupled.

Of course, the configuration described above is exemplary only. In practice, the type and number of system components used varies depending on the purpose of the computer system 1 , Also the architecture and topography of the bus systems of the computer system 1 can be varied in many ways. For example, other system architectures are known in which a plurality of different system components are connected to a common system bus. The specific arrangement and design of the individual bus systems, some of which are described above, depends on their respective intended use.

Particularly in modern, particularly powerful computer systems, the bus systems used often have to have very high data transmission rates. In order to simultaneously achieve flexibility with respect to usable system components, data transmission between two bus components, usually a bus control as a bus master and a further bus component as a so-called slave, is trained in the first initialization of the respective bus system. In this way, optimal configuration parameters for operating the bus system can be found.

Individual or all such configuration parameters can also be specified by the individual system components. In the exemplary embodiment, for example, the memory modules include 4a and 4b each a non-volatile memory with so-called Serial Presence Detect (SPD) data 13a and 13b ,

In the SPD data are configuration parameters for automatically configuring the memory modules 4a and 4b , in particular their storage capacity, supply voltage and timing parameters, such as required refresh and latency times, stored. These data are from a firmware component of the computer system 1 during the initialization of the memory bus 9 retrieved as part of the so-called power-on self-test (POST) and to configure one in the hub module 3 integrated memory bus controller used. The values determined for operating the memory bus controller are stored in suitable registers, for example the hub module 3 or the BIOS device 5 stored. In the embodiment according to 1 are exemplary nonvolatile CMOS registers 17 of the BIOS device 5 shown.

Similar parameters for configuring and operating individual components of different bus systems or associated bus control components of a bus system, such as the PCI bridge 6 , are also in the expansion cards 7a to 7c saved. In the 1 they are as configuration data 14a to 14c shown.

In addition, the computer system 1 Include sensors for detecting environmental conditions. In the exemplary embodiment, the computer system comprises 1 a temperature sensor 15 and a voltage sensor 16 which detects an operating temperature or an operating voltage of a system board used and via the system management bus 10 provides.

In the 2 is an example software architecture 20 for determining system stability of the computer system according to 1 shown. In the exemplary embodiment, the software architecture comprises 20 a firmware component 21 as well as an operating system 22 ,

In the firmware component 21 For example, it is a conventional BIOS component with program code that is backward compatible with the BIOS of the original IBM PC. Alternatively, the firmware component may be 21 also be a modular extendable firmware according to the Extensible Firmware Interface (EFI) act.

The firmware component 21 includes in the embodiment a boot component 23 , a bus control component 24 and an analysis interface 25 , The boot component 23 is called first after applying an operating voltage to the processor 2 and the BIOS device 5 executed. The boot component 23 loads additional components of the firmware component 21 and optionally other firmware components of other system components of the computer system 1 after to the computer system 1 and to initialize the system components contained therein.

In the exemplary embodiment, the boot component loads 23 the bus control component 24 for example, the memory bus 9 initialized. For this purpose, in the exemplary embodiment, first the SPD data 13a and 13b from the memory modules 4a and 4b accessed. Subsequently, in the course of a so-called training phase, different data transfer rates and / or memory access modes between the memory bus controller and the memory modules 4a and 4b tested. This recorded configuration parameters, such as available data transmission modes or their latencies are in registers, such as the CMOS register 17 of the BIOS device 5 for the further operation of the computer system 1 deposited. These data can be subsequently transmitted via the analysis interface 25 be retrieved.

In the embodiment, within the operating system 22 an analysis component 26 as well as a remote maintenance component 27 executed. Of course, the components can 26 and 27 alternatively also part of the firmware component 21 or an independent monitoring component, such as a system management module, also known as a baseboard management controller (BMC).

The analysis component 26 calls via the analysis interface 25 that of the bus control component 24 ascertained configuration parameters for a bus system. On the basis of predetermined reference values, which may be fixed for example for different system configurations or have been determined in an earlier determination of configuration parameters in an original system configuration, it is thus possible to determine an optionally existing deviation from desired setpoint values. For example, an impending error or misconfiguration of the computer system 1 detected when the latency for addressing certain bus components increases or when certain data transmission modes are no longer available. Of course, it may also be checked if detected bus components for use with other components of the computer system 1 are basically suitable.

The captured data can either be evaluated locally and used to display alerts in the computer system 1 be used or via the remote maintenance component 27 be forwarded to a central maintenance computer. Unless the computer system 1 integrated into a network, for example in a company network, this transfer can take place fully automatically in the background. Otherwise, the data may be transmitted through the remote maintenance component 27 stored locally on a non-volatile storage medium and delivered as needed to a corresponding maintenance component.

In the 3 is a flowchart of a method 30 to determine system stability.

In a first step 31 becomes at least a bus system of a computer system 1 initialized. For example, the processor bus 8th , the memory bus 9 , the system management bus 10 , the chipset bus 11 and / or the PCI Express bus 12 of the computer system 1 according to 1 be initialized.

At least some bus systems, such as a so-called frontsidebus (FSB) of the processor 2 , an interprocessor bus such as Intel's Quickpath (QPI) or a PCI Express bus are optimized via a so-called training. This means that such bus systems are re-measured and initialized at each restart or at each firmware initialization. This training is part of the program code of the firmware component 21 For example, the bus control component 24 , To carry out the training, test mechanisms that have already been implemented at the chip level are often used by the respective manufacturers of the participating microchips. In these test algorithms, so-called margin tests are often included test algorithms in which tests are performed under variation of certain boundary conditions, such as the operating voltage, or variation of timing values.

As a result of this test training and optimization phase are in step 32 Configuration parameters for operating the respective bus system in a current configuration of the computer system 1 through a firmware component 21 detected.

For example, such parameter values may include the timing of particular bus signals or the presence and location of expansion components located in insertion locations 7 in the form of appropriate register values or variables in the memory of the computer system 1 be filed. These configuration parameters are then available for the operation of the respective bus component or the respective bus system and can then for example, be read by a bus component or a bus control.

In one step 33 the captured configuration parameters are via an interface of a firmware component 21 read. In this way, the hardware-specific results of the bus initialization in step 31 and the one based on it in the step 22 captured configuration parameters also provided for other software components. For example, further so-called EFI expansion modules of the firmware component 21 access the training results of a hardware driver. Alternatively, the results may also be as in the 2 represented at a higher software level, for example at an operating system level.

The advantage of using the results of hardware-specific test routines is, inter alia, that they can be carried out particularly quickly due to the hardware-oriented implementation, which leads to a reduction in the required test time. In addition, test results are available that can usually not be retrieved or reproduced by other means, or only with considerable effort, because chip manufacturers often do not completely disclose the details of the individual test routines.

In one step 34 On the basis of at least one read-out configuration parameter, a system stability is then determined by an analysis component. The results of the initialization or training phase can be evaluated in various ways.

For example, it is possible to store the acquired configuration parameters for a given system configuration under a unique identifier. The stored configuration parameters can then be used in particular for comparisons with other system configurations, statistical monitoring of system stability and error tracking. For example, over a statistical view, the life cycle of the computer system 1 be tracked. It is also possible, via a correlation with configuration parameters of known error situations, to a misconfiguration of the computer system 1 close. In addition, it can be concluded directly from certain parameters on the quality of the respective system configuration, for example via the evaluation of data transfer rates available for data transmission via a bus system. In this way, system configurations of a computer system can possibly be optimized before delivery. This avoids configurations that, while still a functioning computer system 1 but do not use the optimal performance of the components involved.

The method steps described can be either once, for example, prior to delivery of a completely assembled computer system 1 , or repeatedly, for example as part of the power-on-self-test (POST) to be executed.

In an initial test, conspicuous systems can be sorted out with such examinations, which, although they have passed a functional initial test, clearly deviate from known nominal values with regard to the acquired configuration parameters. In this way, a later failure of the computer system 1 be prevented with subsequent maintenance and warranty claims. Of course, such an initial test also completely or completely replace known, sometimes time-consuming initial test.

The method sections described are also suitable for monitoring the system quality of already delivered computer systems 1 , The monitoring can be triggered both preventatively, for example, at a given time, for example at every boat or once a month, or specifically in case of a problem. Should be in the step 34 relatively complex analyzes are carried out, the process can be carried out at longer intervals.

It is also possible to perform or initiate the procedure when predetermined system management interrupts (SMI) occur, such as are regularly caused by a timer or when severe faults occur. Of course, only individual parts of the method, such as the step 33 to retrieve the captured configuration parameters each time a particular interrupt occurs, during the subsequent step 34 only in previously given. Time intervals, for example, after 100 hours of operation or once a year, carried out.

In order to keep the maintenance costs low, it is recommended to be able to trigger the method with means for remote maintenance. This is in addition to the already with respect to the 2 described remote maintenance component 27 on the software level, a baseboard management controller (BMC), which provides remote access independent of the processor 2 on components of a system board. By means of repeated or continuous monitoring of configuration parameters, it is thus possible, for example, to detect the aging of components, for example dried electrolytic capacitors or corroded plug-in connections, and to deduce therefrom a possible deterioration of the system stability.

Furthermore, such monitoring also allows the active detection of system configuration changes, for example, by the subsequent installation of unknown or inappropriate components by a customer. If further operating parameters, such as the temperature, are included in the monitoring, it can also be recognized whether or, if appropriate, which components are particularly susceptible to an increased or reduced operating temperature and thus lead to a reduction in system stability. Finally, it can also be recognized if any in the computer system 1 used firmware component, in particular a BIOS version, which is in the computer system 1 to correctly control the components contained in it or, if necessary, to carry out an update.

LIST OF REFERENCE NUMBERS

1

computer system

2

processor

3

Hub module

4

memory module

5

BIOS chip

6

PCI Bridge

7

extension component

8th

processor

9

memory

10

system management

11

Chipset

12

PCI Express bus

13

SPD data

14

configuration data

15

temperature sensor

16

voltage sensor

17

CMOS register

20

software architecture

21

firmware component

22

operating system

23

boat component

24

Bussteuerungskomponente

25

Analysis interface

26

analysis component

27

Remote maintenance component

30

Method for determining system stability

Claims (10)

Procedure ( 30 ) for determining system stability of a system configuration of a computer system ( 1 ) comprising the steps of: - initializing at least one bus system ( 8th . 9 . 10 . 11 . 12 by a bus controller using at least one predetermined, hardware-specific test routine; Detecting at least one configuration parameter for operating the bus system ( 8th . 9 . 10 . 11 . 12 ) in a current configuration based on the at least one test routine by a firmware component ( 21 ); - Reading the at least one detected configuration parameter via an interface ( 25 ) of the firmware component ( 21 ); and determining system stability based on the at least one configuration parameter read out by an analysis component ( 26 ). Procedure ( 30 ) according to claim 1, characterized in that the system stability is determined based on a deviation of the read-out at least one configuration parameter from a predetermined reference value. Procedure ( 30 ) according to claim 1 or 2, characterized in that the steps of initialization, detection and read-out are performed repeatedly and the system stability is determined on the basis of a statistical evaluation of the at least one repeatedly read configuration parameter. Procedure ( 30 ) according to claim 3, characterized in that the steps of initializing, detecting and reading upon activation of the at least one bus system ( 8th . 9 . 10 . 11 . 12 ), upon a change in the system configuration, upon the initiation of an interrupt request, or upon expiration of a predetermined time interval. Procedure ( 30 ) according to one of claims 1 to 4, characterized in that the predetermined, hardware-specific test routine a training sequence for training a data transfer between the bus control and at least one with the at least one bus system ( 8th . 9 . 10 . 11 . 12 ) connected bus component. Procedure ( 30 ) according to one of claims 1 to 5, characterized in that the at least one configuration parameter comprises at least one of the following parameters: number of bus slots, type of bus slots, number of bus components, type of bus components, mode of bus components, position of bus components, access times Bus components, temperature of bus components, height of a supply voltage of the bus system ( 8th . 9 . 10 . 11 . 12 ). Computer system ( 1 ), comprising: - at least one bus system ( 8th . 9 . 10 . 11 . 12 ) with a bus controller and at least one bus component connected to the bus system; At least one firmware component ( 21 ) with program code for initializing the bus system ( 8th . 9 . 10 . 11 . 12 ) by the bus controller using at least one predetermined hardware-specific test routine comprising a training sequence for training a data transfer between the bus controller and the bus component and for detecting at least one configuration parameter for operating the bus system in a current configuration of the computer system ( 1 ) based on the at least one test routine; and at least one analysis component ( 26 ) with program code for reading the at least one detected configuration parameter via an interface ( 25 ) of the firmware component ( 21 ) and determining system stability based on the read at least one configuration parameter. Computer system ( 1 ) according to claim 7, characterized in that the firmware component ( 21 ) and / or the analysis component ( 26 ) in a BIOS component of the computer system ( 1 ) are integrated. Computer system ( 1 ) according to claim 7 or 8, characterized in that the analysis component ( 26 ) in a system management component of the computer system ( 1 ), which is adapted for coupling with a maintenance device. Computer program product comprising program code for determining a system stability of a system configuration, characterized in that when the program code is executed by a processor ( 2 ) of a computer system ( 1 ) the following steps are carried out: - reading at least one during an initialization of at least one bus system ( 8th . 9 . 10 . 11 . 12 ) via a bus control using at least one predetermined, hardware-specific test routine detected configuration parameter via an interface ( 25 ) a firmware component ( 21 ); and determining system stability based on the read at least one configuration parameter.

Images