int32_t checkMcsChannelFail( ExtensibleChip * i_mcsChip,
STEP_CODE_DATA_STRUCT & io_sc )
{
#define PRDF_FUNC "[MemUtils::checkMcsChannelFail] "
int32_t o_rc = SUCCESS;
do
{
// Skip if already handling unit checkstop.
if ( io_sc.service_data->GetFlag(ServiceDataCollector::UNIT_CS) )
break;
// Must be an MCS.
if ( TYPE_MCS != getTargetType(i_mcsChip->GetChipHandle()) )
{
PRDF_ERR( PRDF_FUNC "i_mcsChip is not TYPE_MCS" );
o_rc = FAIL; break;
}
// Check MCIFIR[31] for presence of channel fail.
SCAN_COMM_REGISTER_CLASS * mcifir = i_mcsChip->getRegister("MCIFIR");
o_rc = mcifir->Read();
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC "Read() failed on MCIFIR" );
break;
}
if ( !mcifir->IsBitSet(31) ) break; // No channel fail, so exit.
// Set unit checkstop flag and cause attention type.
io_sc.service_data->SetFlag(ServiceDataCollector::UNIT_CS);
io_sc.service_data->setSecondaryAttnType(UNIT_CS);
io_sc.service_data->SetThresholdMaskId(0);
// Indicate that cleanup is required.
P8McsDataBundle * mcsdb = getMcsDataBundle( i_mcsChip );
ExtensibleChip * membChip = mcsdb->getMembChip();
if ( NULL == membChip )
{
PRDF_ERR( PRDF_FUNC "getMembChip() returned NULL" );
o_rc = FAIL; break;
}
CenMembufDataBundle * mbdb = getMembufDataBundle( membChip );
mbdb->iv_doChnlFailCleanup = true;
} while (0);
if ( SUCCESS != o_rc )
{
PRDF_ERR( PRDF_FUNC "Failed: i_mcsChip=0x%08x", i_mcsChip->GetId() );
}
return o_rc;
#undef PRDF_FUNC
}
/**
* @brief Handles Max Spares Exceeded attentions on the MCS.
*
* This function will first check for channel fail conditions on the MCS side of
* the bus, handle the attention, then do channel fail cleanup if needed. It
* would be preferred that the channel fail handling would be done in a more
* generic way, like it is done on the MCS and MEMBUF pre/post analysis
* functions. Unfortunately, the IOMCFIRs are not on the MCS chiplet, which
* complicates things. Fortunately, the only attentions on the IOMCFIRs that are
* hardwired to channel fail are the Max Spares Exceeded attentions. Therefore,
* we can deal with the channel fail handling within this attention.
*
* @param i_procChip A PROC chip.
* @param i_sc The step code data struct.
* @param i_mcsPos The MCS position.
* @return SUCCESS always.
*/
int32_t maxSparesExceeded_MCS( ExtensibleChip * i_procChip,
STEP_CODE_DATA_STRUCT & i_sc,
uint32_t i_mcsPos )
{
#define PRDF_FUNC "[Proc::maxSparesExceeded_MCS] "
int32_t l_rc = SUCCESS;
TargetHandle_t procTrgt = i_procChip->GetChipHandle();
TargetHandle_t mcsTrgt = NULL;
ExtensibleChip * mcsChip = NULL;
ExtensibleChip * membChip = NULL;
do
{
// Get the connected MCS chip
mcsTrgt = getConnectedChild( procTrgt, TYPE_MCS, i_mcsPos );
if ( NULL == mcsTrgt )
{
PRDF_ERR( PRDF_FUNC"getConnectedChild() returned NULL" );
l_rc = FAIL; break;
}
mcsChip = (ExtensibleChip *)systemPtr->GetChip( mcsTrgt );
if ( NULL == mcsChip )
{
PRDF_ERR( PRDF_FUNC"GetChip() returned NULL" );
l_rc = FAIL; break;
}
// Check for channel fails on the MCS side of this bus.
l_rc = MemUtils::checkMcsChannelFail( mcsChip, i_sc );
if ( SUCCESS != l_rc )
{
PRDF_ERR( PRDF_FUNC"checkMcsChannelFail() failed" );
break;
}
// Do additional bus analysis.
l_rc = handleLaneRepairEvent( i_procChip, TYPE_MCS, i_mcsPos, i_sc,
false );
if ( SUCCESS != l_rc )
{
PRDF_ERR( PRDF_FUNC"handleLaneRepairEvent() failed" );
break;
}
// Get the connected MEMBUF chip.
P8McsDataBundle * mcsdb = getMcsDataBundle( mcsChip );
membChip = mcsdb->getMembChip();
if ( NULL == membChip )
{
PRDF_ERR( PRDF_FUNC"getMembChip() returned NULL" );
l_rc = FAIL; break;
}
// Do channel fail cleanup.
l_rc = MemUtils::chnlCsCleanup( membChip, i_sc );
if ( SUCCESS != l_rc )
{
PRDF_ERR( PRDF_FUNC"chnlCsCleanup() failed" );
break;
}
} while (0);
if ( SUCCESS != l_rc )
{
PRDF_ERR( PRDF_FUNC"Failed: i_procChip=0x%08x i_mcsPos=%d",
i_procChip->GetId(), i_mcsPos );
CalloutUtil::defaultError( i_sc );
}
return SUCCESS; // Always return SUCCESS
#undef PRDF_FUNC
}
