Data Sheet
Configuration Options
OnceMPEhasexecuted,instructionsfor
furthertests,bothlocallyandsystem-wide,
aretakenfromalistofsystemdescriptor
itemsstoredinnon-volatilememory.This
listcanbemanipulatedfromtheinteractive
SystemDescriptionEnvironment(SDE).
Systemdescriptoritemscanbeinserted
ordeleted,orexistingitemsmodified.
Eachitemdescribesaboardtobetested
(includingthelocalcard),alongwithtest
andreportingcharacteristicsandatest
maskspecifyingwhichindividualtests
shouldberunonanyparticularcard.
Descriptoritemsforunintelligentcardsand
alsoitemstoinvokecustomtestscanbe
created.AMemoryDescriptorEnvironment
(MDE)performsasimilarfunctiontothe
SDE,butallowssystem-widedescriptor
itemsofmemoryareastobetested.
Thisflexibilityallowsteststoeasilybetailored
toindividualcardhardwareoptionsand
particularsystemmixesofcards.Deployed
testofGEIntelligentPlatformhardware
genuinelyfallswithinthescopeofatrue
COTSsoftwaretestpackage,avoidingmuch
ofthe‘testintegration’overheadsofnon-modular
packages,eveneasingsystemtestintegration
whennon-GEIntelligentPlatformelements
areincludedaspartofthetotalsystem.
BCS Operation
Tosatisfytheneedsofbothdeployedsystems
anddevelopmentscenarios,BCSisprovided
inaflexibleform.Forinstance,inthe
VxWorkstargetedversiontheBCSobject
modulecanbedownloadedfroma
VxWorkstargetshell,oralternativelyitcan
belinkedtotheVxWorksoperatingsystem
image.
BCScanalsobelaunchedfromthetarget
shellorfromanapplicationviainsertionof
anappropriatelineofcodewhichcallsthe
startpoint.Itcanbestaticallyconfigured
byaltering‘.h’filesandrecompiling,or
dynamicallyconfiguredatruntimeviaan
interactivemenu.Useofalltheseoptionsis
coveredintheuserdocumentation.
Configurationoptionsincludetestmasks
toenableordisablespecifictests,error
loggingactions(includingenableofvisual
indication)andBCStaskthreadpriority.
Typically,BCSwouldbesettothelowest
prioritysothatprocessorcyclesareonly
usedingainingtestconfidencewhenthere
isnodirectapplicationworktodo.
BCSisengineeredforminimumsystem
impact,atversion2.0imposingonly1.2uS
interruptlatency(measuredonaPPC4A
with750processor@250MHzand83
MHzbus).Periodicityofbackgroundtests
isconfigurable.BackgroundFlashintegrity
testsareaccomplishedsectorbysector,
butittypicallytakesabout0.5secondsper
megabytetoverifychecksums(dependson
processorspeedchosenetc.).SomeFlash
accesslatencyisimposedonthesector
BCSischecking.BCSprovidesacontinuous
warningofsystemproblemswiththemini-
malintrusionnecessaryforusewithopen
marketCOTSsoftwarecomponents.
BCS Features
• DownloadableorlinkabletoVxWorks
OSimage
• HLaunchedfromshellorapplication
• Staticconfigurationvia.hfile
• Dynamicconfigurationviainteractivemenu
• Configurablethreadpriorityandother
parameters
• ErrorloginFlashandvisualindication
• Call-backmenuforimmediateinvocation
ofindividualtests,inadditiontorunning
testsinbackgroundmode
• Logsandscrubssinglebiterrors
• Comprehensivemainmemorytest.By
dedicatingsmallsegmentsperbankfor
exclusiveBCSusage,andinconjunction
withECCcircuitry,allfailuremodes
throughoutallthememorycanbe
detectedwithoutdestructiveaction
outsidetheBCSsegments
• SystemanduserbackgroundFlash
checksumming
• NVRAMchecksum
• PCIbuserrorconditionmonitoring
• PresetPCIconfigurationverification
• Temperaturemonitoring
• Temperaturethrottling
• Networkconnectivity
• SCSIconnectivity
• Busmemoryprobing
• Real-timeClocktest
• Globalhardwareregisterverification
• Testsof8250-compatibleCOMport
devices
• AltiVecandFPUtests
• Customtestscanbeintegrated
DeployedTestSolutions
BIT Features
• Automaticinvocationofpre-configured
test set
• Layeredtestsforbestsubsystemisolation
• Interactiveconfigurationenvironment
• Interactivetestinvocationenvironment
• Objectproductrecognizesandruns
customPMCtestswrittentorecom-
mendedformat
• Visualindicationanddetailedresults
storedinFlashorsystemRAMforlater
analysis
• BITco-ordinatestestexecutionand
resultsthroughoutallGEIntelligentPlat-
formboardsinasystem
• 95%orgreaterfunctionallyverifiable
coverage
•Inadditiontointernalnodetests,
optionaledgenodetestsallowthe
useofBITforATPaswellasdeployment
Bit Coverage Verification
Standardreliabilityanalysisfiguresare
traditionallyusedtohelpdeterminetest
firmwarecoverage.Eitherapartscountor
partsstressfailureratepredictionwillproduce
atableofthenumberoffailurespermillion
hoursforeachcomponent.Everyfigure
inthistablecanbegivenaweightingthat
wouldcontributetoanoverallcoveragefigure
forallpartsthatthefirmwareactually
tests.However,thismethodhasanumber
ofdrawbacks.First,datafromactualfield
failuresdoesnotalwaysalignwellwith
failureratepredictions.Second,sincethe
fullacceptanceofCOTStechnology,thereis
nowabroadrangeofsourcesofdataused
forreliabilityprediction.Anytestfirmware
coveragebasedonlyuponreliabilityanalysis
tendstobemoredependentonthesources
ofthefailureratedatausedthanonthe
qualityofthefirmwareimplementation.
Last,onlytheverylatestfailurerateprediction
methodologiesarebeginningtotakeinto
accountthesignificanteffectofprinted
circuitboardandsolderjointreliability.
GEIntelligentPlatformstillprovidestraditional
reliabilityanalysisfigures(generallytoMIL-
STD-217),whichiscertainlyusefuldatato
havewhenconsideringtheoverallsuitabil-
ityofparticularboardsformission-critical
applications.However,inthehighlycom-
mercialworldofCOTSsilicon,thismethod
nolongerfurnishesareliableorobjective
measureoftestfirmwarecoverage.More-
over,itdoesnotpredictfailuredueto
mis-handlingproblemsorinterfaceabuse.
Thiscanbeimportant,consideringthe
adventofCOTShasmeantanincreasing




