Method and Apparatus for Signaling on-Line Failure Detection

This invention relates generally to test equipment for electrical apparatus and more particularly to built-in test equipment for multi-module digital systems.

On-line built-in test (BIT) equipment is generally known; however, such apparatus is now becoming a vital part of digital systems, such as computers, which are comprised of many interconnected modules, each containing a plurality of large scale integration semiconductor chips which operate, for example, as individual signal processing units. Due to the chip complexity, reliability and pinout capability present in currently utilized large scale integrated (LSI) chips, many different types of constraints are imposed on system design and any built-in test hardware must maintain a very low profile from an overall system viewpoint. By this is meant maximum failure information must be extracted at a minimum cost in hardware and connectivity.

Accordingly, current built-in test equipment for digital systems is normally limited to the detection of occurrences and the conveyance of an error alarm to a system level control. To avoid added hardware and/or signal paths, little or no detailed information is normally conveyed from the site of the observed failure to higher system levels and separate software and/or hardware is relied upon to identify the error source.

Accordingly, it is an object of the invention to provide an improvement in built-in test equipment for digital systems.

It is another object of the invention to provide built-in test equipment for digital systems which is adapted not only to digitally detect the locality of error occurrences, but also the causality thereof.

It is a further object of the invention to perform fault isolation at the module level of a modular type digital signal processing system.

It is still a further object of the invention to provide built-in test equipment having the capability of additionally providing information about the type of failure which has occurred at a particular location.

These and other objects are achieved in accordance with the method and apparatus for providing, on a two wire built-in test digital signaling path, following the binary indication of a failure thereon, binary signals corresponding to a four state indication of failure detail of each of a plurality of digital signal processing units, which may be, for example, large scale integrated semiconductor chips, which are included within a single module of a multi-module system. The four state indicating signals for each processing unit are applied via respective two wire signaling paths to built-in test circuitry associated with each module which is operable to apply a priority type output to higher level built-in test apparatus which can then signal the location of the faulty module and/or take appropriate action to remove the module electrically from the system.

FIG. 1 is an electrical block diagram of a typical module in a multi-module digital system including built-in test apparatus in accordance with the subject invention; and

FIG. 2 is an electrical block diagram illustrative of the preferred embodiment of means for carrying out the subject invention at the module level.

Referring now to the drawings and more particularly to FIG. 1, reference numeral 10 is intended to designate one of a plurality of modules of a multi-module digital system, such as a digital computer, wherein each module is comprised of a plurality of digital signal processing units (SPU) 12.sub.1, 12.sub.2 . . . 12.sub.n which are implemented in respective large scale integration (LSI) semiconductor chips of well known design. Each processing unit 12.sub.1, 12.sub.2 . . . 12.sub.n includes a first set of two wire output signal paths 14.sub.1, 14.sub.2 . . . 14.sub.n which are adapted to report internal failures and output signal faults occurring within the respective SPU and a second set of two wire output signal paths 16.sub.1, 16.sub.2 . . . 16.sub.n for reporting input signal related errors before they are processed by the internal circuit configurations of the particular SPU. Each of the two wire signaling paths 14.sub.i and 16.sub.i, moreover, is comprised of two circuit leads or wires a and b. Each module 10, moreover, includes its own built-in test (BIT) error collection circuitry 18 in addition to the built-in test error detection circuitry located within each signal processing unit 12.sub.1 . . . 12.sub.n. The module BIT circuitry 18 receives signals of built-in test status from the SPU's by way of respective BIT signal combiners 19.sub.1, 19.sub.2 . . . 19.sub.n and four state signaling circuits 20.sub.1, 20.sub.2 . . . 20.sub.n, the latter being coupled to the error collector chip 18 by means of respective two wire signaling paths A.sub.1, B.sub.1 . . . A.sub.n, B.sub.n. For the purpose of minimizing chip pinouts required, the bit signal combiner 19.sub.i and the respective four state signal circuit 20.sub.i are preferably incorporated with the ith SPU 12.sub.i within a common LSI chip.

The built-in test error collection circuitry 18 within the module 10 is coupled to a system level built-in test error collector 22 by means of a two wire signaling path comprised of the leads or wires A' and B'. The system level bit collector 22 is adapted to receive like two wire signaling paths A', B' from other like modules, not shown, in the system and performs the system level function of stopping further signal processing upon detection of a fault at the SPU level by a process such as stopping a system clock 24 which is coupled to the various modules, for example, module 10 shown in FIG. 1. This is shown for purposes of illustration by a circuit lead 26 connected between the system level built-in test collector 22 to the system clock 24 and is intended to carry, for example, a clock inhibit signal thereon. In reality, the clock 24 may comprise a system of synchronized clocks.

In prior art built-in test equipment associated with digital systems, while the BIT signal combiners 19.sub.1, 19.sub.2 . . . 19.sub.n were known to be used, the four state signal circuits 20.sub.1, 20.sub.2 and 20.sub.n were lacking and signaling of built-in test status of the SPU's was conveyed from the signal processing units 12.sub.1, 12.sub.2 . . . 12.sub.n to the signal processing unit error collector 18 either directly by means of the two sets of two wire signaling paths 14.sub.1 . . . 14.sub.n and 16.sub.1 . . . 16.sub.n or by way of the BIT signal combiners 19.sub.1, 19.sub.2 . . . 19.sub.n via, for example, a respective pair of two wire signaling paths a.sub.1 ', b.sub.1 ' and a.sub.2 ', b.sub.2 ' which conveyed combined internal failures and combined input signal failures, respectively. In operation a failure (error) or no-failure (no error) of BIT status was reported in terms of the binary digital state of the signal on each signal path pair a and b out of the particular SPU. Typically, if mutually different binary values appeared on wires a and b, (wire a=0, wire b=1 or wire a=1, wire b=0) a "no error" status was meant to be indicated, while the mutually same binary value on both wires (wire a=0, wire b=0 or wire a=1, wire b=1) was meant to be indicative of an "error" status. This type of operation, however, fails to provide any significant information as to the cause of the "error" except possibly for a system's built-in test indication of which module or modules, as the case may be, reported the error. Because error sources often cause apparent errors in destinations rather than or in addition to the true source, simple knowledge of which module(s) reported an error is usually of little use in fault location. It is to this aspect that the present invention is directed.

Referring now to FIG. 2, shown in detail is one of the BIT signal combiners 19.sub.1 coupled to signal processing unit 12.sub.1, and which feeds into a four state signaling circuit 20.sub.1 included in the module 10.

As noted above, for the purpose of minimizing LSI pinout requirements, bit signal combiner 19.sub.1 and signaling circuit 20.sub.1 are preferably co-resident within the LSI chip containing the SPU 12.sub.1. It is understood that the other BIT signal combiners 19.sub.2 . . . 19.sub.n and signaling circuits 20.sub.2 . . . 20.sub.n in module 10 are identical in configuration and are respectively coupled to their respective signal processing units 12.sub.2 . . . 12.sub.n.

The BIT signal combiner 19.sub.1 is comprised of a first and second fault combiner 28 and 30 which are respectively coupled to the multiple pairs of two wire signaling paths, a.sub.1, b.sub.1 . . . a.sub.n, b.sub.n of the two sets of signaling paths 14.sub.1 and 16.sub.1 out of the signal processing unit 12.sub.1. The fault combiners 28 and 30 comprise combinational logic which produce an error output indication whenever one or more of the respective input signals indicate an error.

The respective two wire output leads a' and b' of combiners 28 and 30 are respectively fed into a fault register 32 and 34. These registers function as holding registers and are required for retiming fault indications and freezing the fault status existing at report time (via the stoppage of clocks). The start (INIT) and clock (CLK) input terminals of register 32 and 34 receive a "start" signal and a "clock" signal from a system clock 24, for example, shown in FIG. 1. The fault register 32 includes two wire output signal leads a'.sub.1 and b'.sub.1 which provide fault status indication to another fault combiner 36 as well as a first EXCLUSIVE OR circuit 38 which forms part of the four state signal circuit 20. The fault register 34 in a like manner has a two wire output comprised of circuit leads a'.sub.2 and b'.sub.2 which are commonly coupled to the fault combiner 36 and a second EXCLUSIVE OR circuit 40 in the four state signal circuit 20.sub.1.

The fault combiner 36 in circuit 20.sub.1 provides two output signals on circuit leads a".sub.1 and b".sub.1 while the EXCLUSIVE OR circuits 38 and 40 provide single outputs on respective output circuit lead a".sub.2 and b".sub.2. These four output lines a".sub.1, b".sub.1, a".sub.2 and b".sub.2 connect to four AND gates 42, 44, 46 and 48 along with a control input shown comprising an LSD signal applied to a terminal 50 and its complement provided by an inverter 52. As shown, output lead a".sub.1 is coupled to one input of AND gate 42 while its other input is coupled to terminal 50 through the signal inverter 52. Output lead b".sub.1 is coupled to one input of AND gate 46 along with the LSD signal out of the inverter 52. Signal lead a".sub.2 is connected to one input of AND gate 44 whose other input is directly connected to the LSD signal terminal 50. Terminal 50 is also connected to one input of AND gate 48 whose other input is connected to circuit lead b".sub.2 from the EXCLUSIVE OR gate 40. The outputs of the AND gates 42 and 44 couple to a first NOR gate 52 which provides an output signal on circuit lead A.sub.1 which comprises one of the leads of the two wire signaling path to the SPU error collector 18 shown in FIG. 1. The other output lead comprises the lead B.sub.1 which is connected to the output of a second NOR gate 54 coupled to the outputs of the AND gates 46 and 48.

In operation, the four state signal circuit 20.sub.1 couples the combined status of the two two wire signal paths a'.sub.1, b'.sub.1 and a'.sub.2, b'.sub.2 from the BIT signal combiner 19.sub.1 through its logic gates to the two wire output pair A.sub.1, B.sub.1 while a "no error" condition exists. In such instance, mutually opposite binary values will appear on circuit leads a".sub.1, b".sub.1 from the fault combiner 36. Concurrently, the LSD signal applied to terminal 50 will typically have a binary 0 applied thereto which causes the inverter 52 to output a binary 1. It can readily be seen that under such conditions, the binary output signals on circuit leads A.sub.1 and B.sub.1 will be of mutually opposite values, 0 and 1 or 1 and 0, indicating a "no error" condition. However, where an "error" or failure is outputted from either register 32 or 34, the logic circuitry of the signaling circuit 20.sub.1 will provide the same binary values on signal leads A.sub.1 and B.sub.1 , i.e. 1 and 1 or 0 and 0 which are fed to the system bit error collector circuitry 22 as shown in FIG. 1 which operates to clock this report termed a "primary failure report" into one or more registers, not shown. Concurrently with this report being coupled to the system bit error collector 22 over the two wire signal lines A', B', the system clock 24 is stopped and an LSD signal having a binary 1 value is now applied to terminal 50 in FIG. 2 which acts to convert the binary outputs on lines A.sub.1, B.sub.1 from a two state indicator into a four state indicator which is now adapted to provide in a "secondary failure report" mode details of the failure previously provided in the "primary failure report".

Due to the fact that the fault register 32 which outputs SPU internal and output signal errors only to the EXCLUSIVE OR circuit 38 while the register 34 outputs only SPU input signal errors which are coupled to the EXCLUSIVE OR circuit 40, the following truth table is thereby implemented in the four state signalcircuit 20.sub.1.

When these four secondary reports are coupled to the module's BIT error collector 18 along with any four state built-in test status from the other signal processing units 12.sub.2 . . . 12.sub.n, the module BIT error collector circuitry 18 applies a predetermined priority selection to the collected data to provide specific module built-in test status to the system built-in test error collector 22. The highest error priority is assigned to BIT indication inconsistencies, meaning that SPUs which have not declared the presence of an error or failure but do not signal a "no error" when the four state mode of signal circuit 20.sub.1, for example, has declared the presence of an error. The next highest priority is assigned to error indications of the type where signal processing unit errors are not of the input signal type but are either internal failures or output signal failures. The third priority comprises a status when all error reports provided to the module built-in test circuitry 18 are all input signal faults or special errors. For all error reporting conditions which do not satisfy a higher priority, the fourth priority defines an input fault to the system built-in test error collector 22 at the wires A' and B' of FIG. 1. Such a priority selection is set forth in Table II.

This prioritization can be implemented by conventional techniques or, when desirable, can be implemented by circuitry similar to the four state signal circuitry 20.sub.1 shown in FIG. 2. In either case the initial conveyance of an error occurrence in a module 10 as a result of more faults or failures occurring in the signal processing units 12.sub.1, 12.sub.2 . . . 12.sub.n is logically provided to the system level built-in test circuitry 22 (FIG. 1) with the same two wire signaling path techniques as used within the module 10.

Subsequent to the arrival at the system built-in test error collector 22 of the prioritized module built-in test error reports from the error collector circuitry 18, further built-in test circuitry, not shown, can implement further decision making rules such as those shown in the following Table III.

Further fault analysis, either using more capable built-in test system circuitry or via added diagnostic equipment is aided by providing system level access to all processing unit fault report detailed. Using registers per processing unit within each module built-in test circuit allows the data collected within each module to be serially provided to the system built-in test circuitry and/or added diagnostic equipment. Full serial flow can be obtained by daisy-chaining all module built-in test register circuits or serial-parallel flow can be supported over the already existing error reporting paths between each module and the system built-in test circuitry.

Having thus shown and described what is at present considered to be the preferred embodiments of the invention, it is to be noted that all modifications, variations and changes coming within the spirit and scope of the invention are herein meant to be included.