FR2657182A1 - Device for aiding diagnosis - Google Patents

Abstract

The device for aiding diagnosis of a computerised machine comprises a maintenance processor (PM) and diagnosis circuits (CD) installed in each module (C) of the machine, the modules being printed-circuit boards and the internal buses of the machine, the said boards being of any type and including or not including a microprocessor. The diagnosis circuits and the maintenance processor are linked by a serial bus (BA) and the maintenance processor is linked by the switched public telephone network (RTC) to a telediagnosis centre (CTD). Each diagnosis circuit comprises means for storing module signals, linked to the module by two buses (BS, BE) and means for storing software context which are linked to the microprocessor when the module includes one. In the diagnosis circuit, an error-detection handler delivers an error signal in order to freeze the software context and signals storage means, whether the error originates from the module or from another diagnosis circuit via the serial bus (BA). On detection of an error originating from the module, the error-detection handler also delivers an attention signal on the serial bus (BA).

Description

Diagnostic support device
The invention is in the field of diagnostic means for computer machines.

 For several years, the various computer system manufacturers have endeavored to increase the availability of their systems vis-à-vis their customers. In addition to increasing the reliability of the equipment, increasing the availability rate involves remote diagnosis, remote maintenance and remote prevention.

These methods are implemented by connecting the various client sites to an expert center which, using repatriated information, issues a diagnosis and decides on the actions to be taken.

 In general, the information retrieved is software information: log of task sequences, software error codes, copying of memory areas, etc. This information is transmitted by data transmission links.

 This solution has the advantage of not requiring the installation of additional material means, but of using the software means in place by adapting them to create trace files.

 On the other hand, this solution has several drawbacks - the information is generally logged in application software and is therefore distant both in time and in functional space from the origin of the error. The volume of information processed between the origin of the fault and its manifestation at the software level means that it is difficult to locate in the log the location of the error and to draw up an accurate diagnosis.

- the analysis is made difficult by the large volume of information to be processed in order to trace the cause of the error.

- the analysis of this information requires the participation of 3 levels of experts
- at the application software level.

 - at the system software level.

 - at the material level.

- the analysis of this information may require access to protected data which has already been processed and may also be tainted with error, and therefore complicate the search.

- the retrieval of information to assist in the diagnosis goes through the internal communication channels (data paths) of the machine to be possibly transmitted remotely via a channel reserved for this purpose but materially and logically dependent on the machine .

- the diagnosis by the software could be improved in certain cases by imposing at the time of the hardware design of the system constraints of distribution of functions, this to the detriment of the simplicity of the diagrams.

 The object of the invention is to overcome these drawbacks and to obtain an accurate picture of the state of a computer machine at the time of a hardware or software fault, in order to give a diagnostic center, local or remote, the maximum of information so that it can establish the most precise diagnosis possible concerning faults or disturbances of the system on site for assistance in maintenance.

 The subject of the invention is a circuit for assisting in the diagnosis of a computer machine of a station connected to a remote diagnostic center common to all computer machines of the same type, a computer machine consisting of modules such as printed circuits with or without microprocessor, buses internal to the machine, characterized in that a station comprises a maintenance processor and at least one computer machine, that each module of a computer machine is equipped with a diagnostic circuit which constitutes a device for collecting basic information characteristic of the operation, good or bad, of the module, that all the diagnostic circuits of a machine are connected, by a serial bus specific to the diagnosis, to the maintenance processor, and that the maintenance processor is connected to the remote diagnostic center by the public switched telephone network.

 The diagnostic aid device of the invention is located in a computer machine or in a station having two computer machines. It consists of diagnostic circuits and a maintenance processor connected on the one hand to the diagnostic circuits and on the other hand to a diagnostic center. Each diagnostic circuit is a means of removing hardware and / or software information from the sub-assemblies of a computer machine; the subsets will be called modules in the following description. The diagnostic circuits are located at the very heart of the functions to be monitored and are used to take software information from the microprocessor-based cards, hardware information constituted by the set of important signals representative of the good functioning of a card (with or without microprocessor), internal buses of the machine, supervision modules, power supplies, ... etc; in this way - the information taken is very close to error in time and in functional space; they are therefore not yet altered by higher hardware and software processing, - the material information taken contains information before, during and after the error, which is important for the accuracy of the diagnosis, - on cards comprising microprocessors, the circuit takes contextual software information at the time of the error, - the taking at different functional levels of the machine offers the possibility of making correlations between the different data, which is an important aid in the development of a reliable diagnosis, - the processing of this type of information to extract a diagnosis requires only the intervention of experts in the design of the machine, - the transmission of the collected information is done by a communication channel distinct from the standard channels of the machine and reserved only for this purpose, materially and logically independent and conceptio n as simple as possible, - the introduction of the diagnostic circuit in the diagrams does not imply any constraint on the study.

 The invention will be better understood from the following description of embodiments illustrated by the appended figures in which - FIG. 1 represents a remote diagnostic system using the diagnostic aid device of the invention, - FIGS. 2A and 2B schematically represent a diagnostic circuit of the device of the invention.

The remote diagnosis system of FIG. 1 is constituted by stations S1 to Sn, by the public switched telephone network PSTN, and by a remote diagnosis center CTD connected by the switched network to the stations. Each station Sl to Sn comprises a computer machine, or two computer machines when a machine is duplicated, as shown in FIG. 1 in which the station Sl comprises two machines
SlA and SlB; each station also includes a PM maintenance processor. In each machine of the stations, a CD diagnostic circuit is located at the level of each of the functional modules C of the machine, such as for example cards equipped or not with a microprocessor, the internal buses of the machine, etc. diagnostic circuits of a machine are connected to each other and to the maintenance processor PM by a serial bus; in station Sl, the diagnostic circuits CD of the machine SlA are connected to a serial bus BA and the diagnostic circuits of the machine SlB are connected to a serial bus BB, the serial buses BA and BB being connected to the maintenance processor PM , and being distinct from the internal buses of the SIA S1B machines and functionally independent of the said machines.

 The maintenance processor PM is connected to the switched PSTN network by an asynchronous serial type LPM link, and to each machine, by an asynchronous serial type LS link; in station S1 there are therefore two links, LSA and LSB; the PM maintenance processor is also connected to an operator console by an asynchronous serial type LOP link.

 In a station, the diagnostic circuits CD and the maintenance processor PM constitute a diagnostic aid device of the invention.

 A CD diagnostic circuit is a device for collecting basic information; on a particular event, error, stop, etc., it takes information and transmits it to the PM maintenance processor. The maintenance processor is used to collect all the information recorded by each diagnostic circuit of the station, and it has the possibility of preprocessing said information. He communicates his information to the CTD remote diagnostic center. The remote diagnostic center, connected to all maintenance processors by the PSTN switched telephone network, collects all the information from the maintenance processors and prepares diagnostics, statistics and forecasts. Obviously the remote diagnosis center is provided for 'the same type or known types, of a manufacturer, and the remote diagnostic center is generally located at said manufacturer. Such a remote diagnostic center therefore supervises a set of machines and makes it possible to establish possible correlations between the faults of said machines, to identify cyclic faults and their context on one machine, to identify identical faults on several machines, etc. This centralization makes it possible to have a global view of the cases of faults and their contexts, and to create databases gathering all the histories of faults. The communications between the remote diagnostic center and a maintenance processor are established either on the initiative of said center or on the initiative of the maintenance processor.

 On the serial bus, such as BA for example, which connects the maintenance processor PM to the diagnostic circuits CD of a computer machine, only the maintenance processor is master and can address a diagnostic circuit; the diagnostic circuits are slaves and cannot address the maintenance processor or any other diagnostic circuit. The BA, or BB, serial bus consists of an incoming data line, an outgoing data line and a bidirectional attention line. The transmission of information between master and slave is done in both directions, in asynchronous mode and NRZI coding. The dialogue between the PM maintenance processor and a CD diagnostic circuit consists of two sequences and a permanent observation of the bus rests by the maintenance processor. The two sequences are as follows: -addressing a diagnostic circuit by the maintenance processor, - response from the diagnostic circuit.

 The addressing / command sequence always consists of a fixed number of bytes. On the other hand, the response, depending on the type of command, comprises a variable number of bytes from one type to another.

 The addressing sequence is divided into three phases - checking the validity of the sequence data, - address recognition, - command recognition.

 These three phases are executed in order by an automaton of the diagnostic circuit; if one of the first two phases is not carried out, the controller does not continue on the next phase or phases and returns to the bus observation condition. If the first two phases have passed successfully and the PLC does not recognize the command, the diagnostic circuit returns by the response sequence a received command code.

Response sequence of the diagnostic circuit. The diagnostic circuit which is recognized without error goes into transmission on the serial bus to respond to the command - for any command, the response by the diagnostic circuit includes at least and in order the following fields
answering circuit address code,
- code of the order received (used to acknowledge or not the
command issued),
- control code by CRC 16.

- For read commands, the response also includes a data field which is inserted between the command field received and the control code field.

 In all cases, the control code field relates to all the fields which precede it in the same response.

 Silence sequence on the bus. The maintenance processor having sent the data fields of the addressing sequence places its amplifiers connected to the bus in the high impedance state, giving the various diagnostic circuits time to carry out the phases of the addressing sequence. The circuit which has recognized itself must begin its response within a minimum / maximum interval of silence on the bus. In order to check the correct functioning of the interrogated circuit, the maintenance processor must have two supervisions - supervision of the start of response of the diagnostic circuit; this must begin to respond to an addressing sequence before a maximum time. After this time, this means that the diagnostic circuit queried cannot respond, for different reasons, and that the maintenance processor can launch another sequence of addressing, - supervision of the byte interval; the circuit questioned recognized itself, started its response in time, then at any time no longer emits bytes. In this case also the maintenance processor cannot wait indefinitely, and at the end of the byte interval supervision, considers the diagnostic circuit at fault and can send another addressing sequence.

 The diagnostic circuit of the invention provides the following functions - it detects errors, - it keeps track of the states of the module with which it is associated, before during and after the error, - it transmits this information to the maintenance processor.

 The diagnostic circuit receives the error signals specific to the module (card, bus, etc.) with which it is associated. When an error condition is detected, it freezes its own signal sampling and generates an SFG attention signal, on the attention line; this attention signal is received by all the other diagnostic circuits of the computer machine and signifies that they also freeze their sampling. This attention signal SFG is also received by the maintenance processor to inform the latter of the presence of information in the diagnostic circuits. The diagnostic circuit stores operating information of the module, information which is constituted by a set of signals. important representative of the proper functioning of the module. This type of information is taken from all modules.

 When the module includes a microprocessor, the diagnostic circuit stores software context information which are error and module status registers, microprocessor registers and batteries, etc .; this type of information is memorized under control of the microprocessor, when it detects an error condition.

 On receiving an attention signal SFG, the maintenance processor PM initiates a reading sequence of all the diagnostic circuits, and each diagnostic circuit transmits its information via the serial bus, BA, connecting the processor maintenance to the diagnostic circuits of the computer machine, SlA, and by means of the protocol indicated above.

 The CTD remote diagnostic center ensures the supervision and management of all the computer machines of the stations Sl to Sn, and communications with the machines are established either on the initiative of the remote diagnostic center or on the initiative of one of the processors of maintenance.

 The CTD remote diagnostic center makes available to an operator, or the application software it contains, all the means of connection to any PM maintenance processor, therefore to a station to which it is attached. This makes it possible to bring back the diagnostic data stored in the maintenance processor, to modify its parameters, to dialogue with a local operator connected to the maintenance processor, to carry out a debugging of a system board of the maintenance processor in order to refine a diagnosis, etc ...

 A PM maintenance processor can access the remote diagnostic center in two ways - either by software; the maintenance processor can, on certain criteria, such as exceeding storage capacity, event thresholds, detection of serious errors, connect to the remote diagnostic center to transmit information to it, - either by a local operator connected by the LOP link to the maintenance processor; the remote diagnostic center must respond to requests from a station operator: transmission of configuration files, machine statuses, bearings in progress, etc.

In addition, the remote diagnostic center performs operations on diagnostic information.
- the center, upon receipt of data, creates files
directly usable by processing software: format,
organization of data, marking of additional parameters
mental. . .etc. These files are stored on the memories of
mass of the remote diagnostic center,
- processing software allows analysis and
classification of the different error information of a
machine to extract a diagnosis, trigger actions
to refine the analysis of the information received,
look for possible correlations with events
occurred on other stations in order to detect possible
regressions due to increments (hardware, firmware, or
software) etc ...

 The center must monitor maintenance operations and for this purpose record any operation, whether or not reported on each machine in order to have a history of the evolution of each one.

 The center must manage a station documentary database. Each station corresponds to a documentary level, and the center, using monitoring files for maintenance operations by station, keeps the database up to date.

 The set of all the files and their analysis allows the remote diagnostic center to draw up a certain number of reports leading to statistics. The balance sheets and statistics are compared with thresholds and their exceeding generates alarms, triggering preventive interventions (predictions of breakdowns).

 FIGS. 2A and 2B schematically represent a diagnostic circuit CD of the diagnostic aid device of the invention.

 The diagnostic circuit takes eighty specific signals from the module on which it is installed, and representative of the proper functioning of the module. These signals are divided into three categories - the sampled signals, SI to S68, sixty eight of which are signals specific to the operation of the module, and whose state may be interesting in the vicinity of the error, - the sampler signals , EO5 to El2, eight in number, which allow the state of the preceding signals to be recorded with a view to their analysis: it is on a transition of these sampler signals that the signals are sampled, - the error signals, EOl to E04, four in number, the appearance of which is significant of an error on the module where they are generated; the transition of each of them causes the sampling of all the signals, in a single register per error signal.

 The sampler signals and the error signals, which are sampler signals, are also sampled.

 In FIG. 2A, a first bus BS routes the sampled signals S1 to S68, and a second bus BE routes the sampler signals E05 to E012 and the error signals EOl to E04, the two buses connecting the module associated with the diagnostic circuit to a storage block constituted by eight memories M1 to M8 of the FIFO type, and to a set of memory for errors constituted by five registers R1 to R5.

 The memories M1 to M4 each have a capacity of 16 words of 80 bits, each memory receiving the 80 signals from the buses BS and BE; memories M5 to M8 each have a capacity of 16 words of 24 bits, and receive 24 signals including the 12 sampler signals, the other 12 signals taken from the 68 sampled signals being different for each memory. Each of the memories M1 to M8 has for clock signal a different sampler signal E05 to E12, respectively. Each memory is permanently filled at each active transition of its clock signal; error detection by the diagnostic circuit, or reception of the attention signal SFG emitted by another diagnostic circuit, inhibits their writing with a certain delay, which makes it possible to keep track of the samples before, during and after the mistake.

 The registers R1 to R5 are 80-bit registers; the clock signal of the registers R1 to R4 is an error signal EOl to E04, respectively; the clock signal of register R5 is the attention signal SFG, transmitted or received by the diagnostic circuit.

 This register R5 takes all of the 80 signals from the buses BS and BE and samples them on transition from the attention signal SFG. This makes it possible to have the image of the 80 bits at the time of the transition of the signal SFG by the diagnostic circuit.

 Once a register has memorized, it is locked until the sampling is restarted or the diagnostic circuit is reset.

 The memories M1 to M8, and the registers RI to R5 are connected, at the output to a common bus BC, itself connected to the input of a multiplexing assembly RT1 to RT10. The output of this assembly is connected to an internal bus Bl of the diagnostic circuit. A redundant cyclic control CRC circuit is connected to the internal bus BI.

 The records in the memories M1 to M8 and the registers R1 to R5 are completely asynchronous with respect to each other; there is therefore no relative anteriority or posteriority of one memory compared to another. For the diagnosis it is however necessary to restore this notion of relative time of a sample in a memory compared to a sample in another memory.

It is the role of the spatio-temporal time table TP which is associated with memories M1 to M8 and registers R1 to R5. This table is connected at input to the bus BE and therefore receives the 12 sampler signals EOl to E12 therefrom; it is connected, at the output to the internal bus
BI of the diagnostic circuit.

 In a certain number of modules, important signals of the function provided by the modules have meanings independent of the sampler signals EOl to 12. In order to know the state of these signals, during an error, the diagnostic circuit, figure 2A, comprises a block of BRE state registers having eight inputs connected by wires 1 to 8 to the module, for picking up eight signals.

These signals are taken into account - in a register sampled by activation of the attention signal
SFG, - in eight individual flip-flops each assigned to a signal and memorizing on the edge of the signal, - each incoming signal comes to a receiver, without memorization, which makes it possible to know the state of the eight signals at the time of reading by the processor PM maintenance. This block of registers of the BRE states is connected at output to the internal bus BI; it is read by the maintenance processor PM. The reception by the diagnostic circuit of the read command of the register block or the command to restart the sampling, or even the initialization of the diagnostic circuit, resets the flip-flops of the BRE block to zero and unlocks the register having the error signal for clock signal.

The diagnostic circuit is connected to the maintenance processor
PM by an outgoing data line LE, an incoming data line LR and an attention line LSFG, FIG. 2B, these three lines constituting the bus BA, or BB, of FIG. 1. The internal bus BI is connected to a serialization register RES, the serial output of which is connected by an emission amplifier AE to the outgoing data line LE. The incoming data line LR is connected to a reception amplifier AR which is itself connected to a deserialization register RED, the parallel output of which is connected to a control circuit CC. The bidirectional attention line LSFG is connected to a transmission amplifier A1 and to a reception amplifier A2. The transmission amplifier A1 is connected at the output of an error detection and generation automaton. attention signal
AUTSFG; the reception amplifier A2 is connected to the input of the AUTSFG PLC. Line F1, result of the "OR" function in the AUTSFG between the incoming SFG and the outgoing SFG, is connected to the register R5. The error signals EOl to E04 are applied to the automaton AUTSFG in which they are put into logic OR function to generate an error signal which is delivered on the one hand to a wire F2 and on the other hand to the emission amplifier A1 which delivers the attention signal SFG on the attention line LSFG. The error signal on line F2 is used to freeze the samples in the diagnostic circuit, for this the line F2 is connected to a AUTMR memory and register management automaton which delivers an error signal to freeze the sampling.

The diagnostic circuit also includes a block of registers
BRM connected to the microprocessor of the module by a data bus BD, a memory of the MLOG software contexts connected to the block of registers BRM, and an automaton of dialogue AUTD connected to the microprocessor of the module by a parallel link LSC carrying control signals. The BRM register block includes a command register, a status register and an internal register; these registers are accessible for reading by the maintenance processor PM. A bit in the internal register is set by the diagnostic circuit when it receives an SFG attention signal, which allows the maintenance processor to know, when it interrogates the diagnostic circuit, whether it has sent or received the attention signal The memory of software contexts
MLOG is a memory of 256 words of 8 bits, of FIFO type, accessible in writing to a single address by the microprocessor; Memory
MLOG is connected at output to the internal bus BI and it is read by the maintenance processor PM. The dialogue automaton is connected by a line F3 to an OR gate, 10, connected at output to a control input of the register block BRM, and to an OR gate, 11, connected at output to a control input of the MLOG memory.

 The AUTD dialog controller is also connected at output by a line F4 to the AUSTSFG controller, to the CC control circuit, and to the AUTMR controller, to which it delivers a reset signal.

 The diagnostic circuit also includes a register of REF functions connected at input by a function bus BF to the module, and connected at output to the internal bus BI. This function register receives bits of function codes from the module; this register is read by command from the maintenance processor PM.

 The control circuit CC has an input connected by an address link LAD to the module which delivers its address by said link. The control circuit is connected by a line F5 to the automaton AUTMR for managing memories and registers to which it delivers the orders received from the maintenance processor PM by the incoming data link LR.

 The AUTMR management automaton is connected at the output, by a line F6 to an input for controlling memories M1 to M8, registers R1 to R5, multiplexing sets RT1 to RT10, the scoreboard TP, the block of registers of the BRE states, of the CRC circuit, to an input of the OR gates, 10 and 11, and to a command input of the register of the REF functions. The AUTMR automaton delivers by this line F6 read control commands from the various memories, registers and circuits, and a sampling stop command (error signal).

 The diagnostic circuit also includes an AUTRSD automaton for managing incoming LR and outgoing LE data links; this automaton is connected at input to line F4 by which it receives a reset signal; it is connected at the output to the AUTMR management machine; it is connected at output, by a line F7 to a control input of the registers RES AND RED connected to the outgoing LE and incoming data links LR, respectively.

 The dialogs between a diagnostic circuit and the maintenance processor are done according to the protocol indicated above. Following a fault on a module, the diagnostic circuit of this module can be caused to cyclically send an SFG attention signal. To avoid permanently disturbing the other diagnostic circuits and the maintenance processor PM, the latter commands the diagnostic circuit in question to invalidate the attention signal SFG.

 The diagnostic circuits and the maintenance processor of the same station, or of a computer machine, are interconnected according to the principle of daisy chain connection, without regeneration, to minimize the risk of breakage of the default bus circuits. This type of connection also allows the extraction of a module (with its diagnostic circuit) without bus break.

Claims (5)

1 / Device to aid in the diagnosis of a computer machine of a station (S1) connected to a remote diagnostic center (CTD) common to all computer machines of the same type, a computer machine (S1A) consisting of modules ( C) such as: printed circuit boards with or without microprocessor, buses internal to the machine, characterized in that a station comprises a maintenance processor (PM) and at least one computer machine (S1A), that each module d a computer machine is equipped with a diagnostic circuit (CD) which constitutes a device for collecting basic information characteristic of the operation, good or bad, of the module, that all the diagnostic circuits of a machine are connected by a serial bus (BA) specific to the diagnosis, to the maintenance processor (PM), and that the maintenance processor (PM) is connected to the remote diagnostic center (CTD) by the public switched telephone network (PSTN). 2 / diagnostic aid device according to claim 1, characterized in that a station comprises two computer machines (S1A, SlB) each connected by a specific serial bus (BA; BB) to the maintenance processor (PM) of the station (S1). 3 / diagnostic aid device according to claim 1, characterized in that the serial bus (BA) comprises an incoming data line (LR) for reception by the diagnostic circuits of signals transmitted by the maintenance processor, a outgoing data line (LE) for transmission of signals by the diagnostic circuits (CD) intended for the maintenance processor, and an attention line (LSFG), bidirectional, carrying an attention signal (SFG) transmitted by a diagnostic circuit (CD) intended for the maintenance processor (PM) and the other diagnostic circuits (CD), said attention signal (SFG) being emitted by a diagnostic circuit (CD) having detected a fault in the module with which it is associated. 4 / diagnostic aid device according to claim 3, characterized in that each diagnostic circuit (CD) comprises storage means (M1 to M8, R1 to R4) of signals from the module to which they are connected by a first bus (BS) carrying first signals to be sampled (S1 to S68) and by a second bus (BE) carrying second signals to be sampled (EOl to E12) which are also used as clock signals for said storage means and a spatio-temporal time table (TP) connected to the second bus (BE), that a register (R5) is connected as an input to the first (BS) and to the second (BE) bus and has a clock input connected to the line attention (LSFG) by which it receives the attention signal (SFG) to store said first and second signals upon receipt of said attention signal, that said register (R5) and said storage means are connected at output to a bus internal (BI) of the dia circuit gnostic (CD), that said internal bus is connected to a serialization register (RES) itself connected at the output to the outgoing data line (LE), that a control circuit (CC) is connected at input to a deserialization register (RED) itself connected at the input to the incoming data line (LR), that a management automaton (AUTMR) is connected at the input to the control circuit (CC) from which it receives command orders, and at the output to the storage means and to the register to which it delivers the control orders, that an error detection and attention signal generation automaton (AUTSFG) receives among the second signals (EOl to E12) signals (EOl to E04) each characterizes an error, and is connected to the management automaton (AUTMR) and to the attention line (LSFG), said error detection automaton delivering, in the event of an error and as well as upon reception of an attention signal, an error signal to the management controller (AUTMR) in. ur freeze the storage means, and delivering in the event of an error an attention signal (SFG) on the attention line (LSFG).  5 / diagnostic aid device according to claim 4, characterized in that the diagnostic circuit (CD) also includes a dialog automaton (AUTD) and means for memorizing software context (BRM, MLOG), said automaton and said software context storage means being connected to a microprocessor of the module when the latter includes one, and the software context storage means are connected at output to the internal bus (BI) of the diagnostic circuit and receive control commands from the management automaton (AUTMR) and the dialogue automaton.