int32_t analyzeIplCEStats( TargetHandle_t i_mba, bool &o_calloutMade )
{
#define PRDF_FUNC "PRDF::analyzeIplCEStats"
PRDF_ENTER( PRDF_FUNC"(0x%08x)", getHuid(i_mba) );
// will unlock when going out of scope
PRDF_SYSTEM_SCOPELOCK;
int32_t o_rc = SUCCESS;
o_calloutMade = false;
ExtensibleChip * mbaChip = (ExtensibleChip *)systemPtr->GetChip( i_mba );
CenMbaDataBundle * mbadb = getMbaDataBundle( mbaChip );
o_rc = mbadb->getIplCeStats()->analyzeStats( o_calloutMade );
if ( SUCCESS != o_rc )
{
PRDF_ERR( "["PRDF_FUNC"] analyzeStats() failed");
// Get user data
uint64_t ud12 = PRDF_GET_UINT64_FROM_UINT32( getHuid(i_mba), 0 );
uint64_t ud34 = PRDF_GET_UINT64_FROM_UINT32( PRDFSIG_MnfgIplFail, 0 );
// Create error log
errlHndl_t errl = new ERRORLOG::ErrlEntry(
ERRORLOG::ERRL_SEV_PREDICTIVE, // severity
PRDF_MNFG_IPL_CE_ANALYSIS, // module ID
PRDF_DETECTED_FAIL_SOFTWARE, // reason code
ud12, ud34 ); // user data 1-4
// Add 2nd level support
errl->addProcedureCallout( EPUB_PRC_LVL_SUPP, SRCI_PRIORITY_HIGH );
// Add traces
errl->collectTrace( PRDF_COMP_NAME, 512 );
// Commit the error log
ERRORLOG::errlCommit( errl, PRDF_COMP_ID );
}
PRDF_EXIT( PRDF_FUNC"(0x%08x), o_calloutMade:%u",
getHuid(i_mba), o_calloutMade );
return o_rc;
#undef PRDF_FUNC
}
int32_t CenMbaTdCtlr::analyzeTpsPhase2( STEP_CODE_DATA_STRUCT & io_sc,
const CenAddr & i_stopAddr,
const CenAddr & i_endAddr )
{
#define PRDF_FUNC "[CenMbaTdCtlr::analyzeTpsPhase2] "
int32_t o_rc = SUCCESS;
do
{
if ( TPS_PHASE_2 != iv_tdState )
{
PRDF_ERR( PRDF_FUNC"Invalid state machine configuration" );
o_rc = FAIL; break;
}
CenMbaDataBundle * mbadb = getMbaDataBundle( iv_mbaChip );
o_rc = mbadb->getIplCeStats()->calloutHardCes( iv_rank );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"calloutHardCes() failed");
break;
}
// Get error condition which caused command to stop
uint16_t eccErrorMask = NO_ERROR;
o_rc = checkEccErrors( eccErrorMask, io_sc );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"checkEccErrors() failed" );
break;
}
if ( ( eccErrorMask & UE ) || ( eccErrorMask & RETRY_CTE ))
{
// Handle UE. Highest priority
o_rc = handleUE( io_sc );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"handleUE() failed" );
break;
}
}
else if ( eccErrorMask & MPE )
{
o_rc = handleMPE( io_sc );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"handleMPE() failed");
break;
}
}
else
{
// No error found so add rank to callout list, just in case.
MemoryMru memmru (iv_mbaTrgt, iv_rank, MemoryMruData::CALLOUT_RANK);
io_sc.service_data->SetCallout( memmru );
io_sc.service_data->AddSignatureList( iv_mbaTrgt,
PRDFSIG_EndTpsPhase2 );
iv_tdState = NO_OP;
}
} while (0);
return o_rc;
#undef PRDF_FUNC
}
int32_t CenMbaTdCtlr::handleUE( STEP_CODE_DATA_STRUCT & io_sc )
{
#define PRDF_FUNC "[CenMbaTdCtlr::handleUE] "
using namespace CalloutUtil;
int32_t o_rc = SUCCESS;
iv_tdState = NO_OP; // Abort the TD procedure.
setTdSignature( io_sc, PRDFSIG_MaintUE );
io_sc.service_data->SetServiceCall();
CenMbaDataBundle * mbadb = getMbaDataBundle( iv_mbaChip );
do
{
// Clean up the maintenance command. This is needed just in case the UE
// isolation procedure is modified to use maintenance commands.
o_rc = cleanupPrevCmd();
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"cleanupPrevCmd() failed" );
break;
}
// Look for all failing bits on this rank.
CenDqBitmap bitmap;
o_rc = mssIplUeIsolation( iv_mbaTrgt, iv_rank, bitmap );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"mssIplUeIsolation() failed" );
break;
}
// Add UE data to capture data.
bitmap.getCaptureData( io_sc.service_data->GetCaptureData() );
// Callout the failing DIMMs.
TargetHandleList callouts;
for ( int32_t ps = 0; ps < PORT_SLCT_PER_MBA; ps++ )
{
bool badDqs = false;
o_rc = bitmap.badDqs( ps, badDqs );
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC"badDqs(%d) failed", ps );
break;
}
if ( !badDqs ) continue; // nothing to do.
TargetHandleList dimms = getConnectedDimms(iv_mbaTrgt, iv_rank, ps);
if ( 0 == dimms.size() )
{
PRDF_ERR( PRDF_FUNC"getConnectedDimms(%d) failed", ps );
o_rc = FAIL; break;
}
callouts.insert( callouts.end(), dimms.begin(), dimms.end() );
if ( isMfgCeCheckingEnabled() )
{
// As we are doing callout for UE, we dont need to do callout
// during CE for this rank on given port
mbadb->getIplCeStats()->banAnalysis( iv_rank, ps );
}
}
if ( SUCCESS != o_rc ) break;
if ( 0 == callouts.size() )
{
// It is possible the scrub counters have rolled over to zero due to
// a known DD1.0 hardware bug. In this case, the best we can do is
// callout both DIMMs, because at minimum we know there was a UE, we
// just don't know where.
// NOTE: If this condition happens because of a DD2.0+ bug, the
// mssIplUeIsolation procedure will callout the Centaur.
callouts = getConnectedDimms( iv_mbaTrgt, iv_rank );
if ( 0 == callouts.size() )
{
PRDF_ERR( PRDF_FUNC"getConnectedDimms() failed" );
o_rc = FAIL; break;
}
if ( isMfgCeCheckingEnabled() )
{
// As we are doing callout for UE, we dont need to do callout
// during CE for this rank on both port
mbadb->getIplCeStats()->banAnalysis( iv_rank);
}
}
// Callout all DIMMs in the list.
for ( TargetHandleList::iterator i = callouts.begin();
i != callouts.end(); i++ )
{
io_sc.service_data->SetCallout( *i, MRU_HIGH );
}
} while(0);
return o_rc;
#undef PRDF_FUNC
}