PRD: Isolation for HWP Failure

Change-Id: I384c7874d1acb583afa0c2d11492e83acacddee7
RTC: 172287
Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/39724
Tested-by: Jenkins Server <pfd-jenkins+hostboot@us.ibm.com>
Reviewed-by: Benjamin J. Weisenbeck <bweisenb@us.ibm.com>
Reviewed-by: Brian J. Stegmiller <bjs@us.ibm.com>
Reviewed-by: Zane C. Shelley <zshelle@us.ibm.com>
Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/40597
Tested-by: Jenkins OP Build CI <op-jenkins+hostboot@us.ibm.com>
Tested-by: FSP CI Jenkins <fsp-CI-jenkins+hostboot@us.ibm.com>

4 files changed, 72 insertions, 2 deletions

diff --git a/src/usr/diag/prdf/common/plat/p9/p9_mca.rule b/src/usr/diag/prdf/common/plat/p9/p9_mca.rule
index 1494622db..4dd5677ec 100644
--- a/src/usr/diag/prdf/common/plat/p9/p9_mca.rule
+++ b/src/usr/diag/prdf/common/plat/p9/p9_mca.rule
@@ -247,7 +247,7 @@ group gMCACALFIR filter priority( 13 ), cs_root_cause( 4, 13, 14 )
     /** MCACALFIR[2]
      *  Excessive refreshes to a single rank.
      */
-    (rMCACALFIR, bit(2)) ? self_th_32perDay;
+    (rMCACALFIR, bit(2)) ? excessive_refreshes;
 
     /** MCACALFIR[3]
      *  Err detected in the MBA debug WAT logic
@@ -262,7 +262,7 @@ group gMCACALFIR filter priority( 13 ), cs_root_cause( 4, 13, 14 )
     /** MCACALFIR[5]
      *  Calibration complete indication xout
      */
-    (rMCACALFIR, bit(5)) ? self_th_1;
+    (rMCACALFIR, bit(5)) ? calibration_complete;
 
     /** MCACALFIR[6]
      *  Emergency Throttle
diff --git a/src/usr/diag/prdf/common/plat/p9/p9_mca_actions.rule b/src/usr/diag/prdf/common/plat/p9/p9_mca_actions.rule
index a18372c0f..196a95485 100644
--- a/src/usr/diag/prdf/common/plat/p9/p9_mca_actions.rule
+++ b/src/usr/diag/prdf/common/plat/p9/p9_mca_actions.rule
@@ -115,6 +115,7 @@ actionclass maintenance_iaue_handling
 /** MCA/UE algroithm, threshold 5 per day */
 actionclass mca_ue_algorithm_th_5perDay
 {
+    funccall("HwpErrorIsolation");
     calloutSelfMed;
     threshold5pday;
     funccall("mcaUeAlgorithm"); # must be called last
@@ -128,3 +129,17 @@ actionclass mca_ue_algorithm_th_1
     funccall("mcaUeAlgorithm"); # must be called last
 };
 
+/** Excessive refreshes to a single rank */
+actionclass excessive_refreshes
+{
+    funccall("HwpErrorIsolation");
+    self_th_32perDay;
+};
+
+/** Calibration complete indication xout */
+actionclass calibration_complete
+{
+    funccall("HwpErrorIsolation");
+    self_th_1;
+};
+
diff --git a/src/usr/diag/prdf/common/plat/p9/prdfCommonPlugins.C b/src/usr/diag/prdf/common/plat/p9/prdfCommonPlugins.C
index ebf4f8ca7..eeedc505b 100644
--- a/src/usr/diag/prdf/common/plat/p9/prdfCommonPlugins.C
+++ b/src/usr/diag/prdf/common/plat/p9/prdfCommonPlugins.C
@@ -31,6 +31,10 @@
 #include <iipServiceDataCollector.h>
 #include <prdfExtensibleChip.H>
 #include <prdfPluginMap.H>
+#include <prdfPlatServices.H>
+#include <xspprdService.h>
+
+using namespace TARGETING;
 
 namespace PRDF
 {
@@ -92,6 +96,44 @@ int32_t ClearServiceCallFlag_mnfgInfo( ExtensibleChip * i_chip,
 }
 PRDF_PLUGIN_DEFINE_NS(p9_nimbus, CommonPlugins, ClearServiceCallFlag_mnfgInfo);
 
+
+/**
+ * @brief   PRD will perform error isolation for certain errors that may cause
+ *          a HWP to fail.
+ * @param   i_chip PROC or MCA
+ * @param   i_sc   Step code data struct
+ * @returns SUCCESS always
+ */
+int32_t HwpErrorIsolation( ExtensibleChip * i_chip,
+                           STEP_CODE_DATA_STRUCT & io_sc )
+{
+    #if defined (__HOSTBOOT_MODULE) && !defined(__HOSTBOOT_RUNTIME)
+
+    TargetHandle_t trgt = i_chip->getTrgt();
+    uint32_t plid = trgt->getAttr<ATTR_PRD_HWP_PLID>();
+
+    // Check for non-zero value in PLID attribute
+    if ( 0 != plid )
+    {
+        // Link HWP PLID to PRD error log
+        errlHndl_t errl =
+            ServiceGeneratorClass::ThisServiceGenerator().getErrl();
+        errl->plid(plid);
+
+        // Make the error log and callouts predictive
+        io_sc.service_data->setServiceCall();
+
+        // Clear PRD_HWP_PLID attribute
+        trgt->setAttr<ATTR_PRD_HWP_PLID>( 0 );
+    }
+
+    #endif
+
+    return SUCCESS;
+}
+PRDF_PLUGIN_DEFINE_NS(p9_nimbus, CommonPlugins, HwpErrorIsolation);
+PRDF_PLUGIN_DEFINE_NS(p9_mca, CommonPlugins, HwpErrorIsolation);
+
 } // namespace CommonPlugins ends
 
 }// namespace PRDF ends
diff --git a/src/usr/diag/prdf/common/plat/p9/prdfP9PllDomain.C b/src/usr/diag/prdf/common/plat/p9/prdfP9PllDomain.C
index 4819869ac..472eaae1f 100644
--- a/src/usr/diag/prdf/common/plat/p9/prdfP9PllDomain.C
+++ b/src/usr/diag/prdf/common/plat/p9/prdfP9PllDomain.C
@@ -162,11 +162,24 @@ int32_t PllDomain::Analyze(STEP_CODE_DATA_STRUCT & serviceData,
             }
         }
 
+        // In the case of a PLL_UNLOCK error, we want to do additional isolation
+        // in case of a HWP failure
+        ExtensibleChipFunction * l_hwpErrIsolation =
+            l_chip->getExtensibleFunction("HwpErrorIsolation");
+
         // Update error lists
         if (l_errType & SYS_PLL_UNLOCK)
+        {
             sysRefList.push_back( l_chip );
+            (*l_hwpErrIsolation)(l_chip,
+                PluginDef::bindParm<STEP_CODE_DATA_STRUCT&>(serviceData));
+        }
         if (l_errType & PCI_PLL_UNLOCK)
+        {
             pciList.push_back( l_chip );
+            (*l_hwpErrIsolation)(l_chip,
+                PluginDef::bindParm<STEP_CODE_DATA_STRUCT&>(serviceData));
+        }
         if (l_errType & SYS_OSC_FAILOVER)
             mfFoList.push_back( l_chip );
         if (l_errType & PCI_OSC_FAILOVER)