Ignore wakeup failures on checkstopped cores

If a core (unit)checkstops, PHYP no longer allows wakeup operations
to take place.  However, HBRT has generic logic that always does a
wakeup before doing any ec/ex/eq scom at runtime.  Since PRD needs
to do those scoms to analyze the checkstop we have a problem.

The change here is for HBRT to handle a new return code from PHYP
that indicates this situation.  This RC will be treated as a
success, thus allowing the subsequent scom operation to pass.

Change-Id: I4ac62129399db50f48c938b649994423c99622a5
CQ: SW444076
Reviewed-on: http://rchgit01.rchland.ibm.com/gerrit1/66143
Tested-by: Jenkins Server <pfd-jenkins+hostboot@us.ibm.com>
Reviewed-by: Matt Derksen <mderkse1@us.ibm.com>
Tested-by: Jenkins OP Build CI <op-jenkins+hostboot@us.ibm.com>
Tested-by: Jenkins OP HW <op-hw-jenkins+hostboot@us.ibm.com>
Tested-by: FSP CI Jenkins <fsp-CI-jenkins+hostboot@us.ibm.com>
Reviewed-by: Corey V. Swenson <cswenson@us.ibm.com>
Reviewed-by: William G. Hoffa <wghoffa@us.ibm.com>

2 files changed, 20 insertions, 8 deletions

diff --git a/src/usr/scom/handleSpecialWakeup.C b/src/usr/scom/handleSpecialWakeup.C
index 11fbb94b1..87a775440 100644
--- a/src/usr/scom/handleSpecialWakeup.C
+++ b/src/usr/scom/handleSpecialWakeup.C
@@ -113,8 +113,8 @@ errlHndl_t callWakeupHyp(TARGETING::Target* i_target,
     if( (g_hostInterfaces == NULL) ||
         (g_hostInterfaces->wakeup == NULL) )
     {
-        TRACFCOMP( g_trac_scom,
-                   ERR_MRK"Hypervisor wakeup interface not linked");
+        TRACFCOMP( g_trac_scom,ERR_MRK
+                   "callWakeupHyp> Hypervisor wakeup interface not linked");
 
         /*@
          * @errortype
@@ -210,10 +210,19 @@ errlHndl_t callWakeupHyp(TARGETING::Target* i_target,
         // Do the special wakeup
         int l_rc = g_hostInterfaces->wakeup(rtTargetId,mode);
 
-        if(l_rc)
+        // Check for the specific case where PHYP has detected a core
+        //  checkstop and will fail no matter what
+        if( HBRT_RC_WAKEUP_INVALID_ON_CORE_XSTOP == l_rc )
+        {
+            TRACFCOMP( g_trac_scom,ERR_MRK
+                       "callWakeupHyp> Wakeup on %.8X failed due to core checkstop",
+                       TARGETING::get_huid(i_target) );
+        }
+        // Any other failure is valid
+        else if(l_rc)
         {
             TRACFCOMP( g_trac_scom,ERR_MRK
-                "Hypervisor wakeup failed. "
+                "callWakeupHyp> Hypervisor wakeup failed. "
                 "rc 0x%X target_huid 0x%llX rt_target_id 0x%llX mode %d",
                 l_rc, get_huid(*pCore_it), rtTargetId, mode );
 
@@ -295,7 +304,7 @@ errlHndl_t callWakeupHwp(TARGETING::Target* i_target,
     if((l_count==0) && (i_enable==WAKEUP::DISABLE))
     {
         TRACFCOMP( g_trac_scom,ERR_MRK
-            "Disabling special wakeup on target with SPCWKUP_COUNT=0");
+            "callWakeupHwp> Disabling special wakeup on target with SPCWKUP_COUNT=0");
 
         /*@
          * @errortype
@@ -413,7 +422,7 @@ errlHndl_t callWakeupHwp(TARGETING::Target* i_target,
         if(l_errl)
         {
             TRACFCOMP( g_trac_scom,
-                    "p9_cpu_special_wakeup ERROR :"
+                    "callWakeupHwp> p9_cpu_special_wakeup ERROR :"
                     " Returning errorlog, reason=0x%x",
                     l_errl->reasonCode() );
 
diff --git a/src/usr/scom/scomtrans.C b/src/usr/scom/scomtrans.C
index f100a5e3c..c2db14ff3 100644
--- a/src/usr/scom/scomtrans.C
+++ b/src/usr/scom/scomtrans.C
@@ -209,6 +209,11 @@ errlHndl_t startScomProcess(DeviceFW::OperationType i_opType,
             g_wakeupInProgress = true;
 
             l_err = WAKEUP::handleSpecialWakeup(i_target,WAKEUP::ENABLE);
+
+            // always clear the global flag, even on error, otherwise we'll
+            //  never call wakeup again even for different targets
+            g_wakeupInProgress = false;
+
             if(l_err)
             {
                 TRACFCOMP(g_trac_scom, "startScomProcess: "
@@ -220,8 +225,6 @@ errlHndl_t startScomProcess(DeviceFW::OperationType i_opType,
                 l_err->collectTrace(SCOM_COMP_NAME,1024);
                 break;
             }
-
-            g_wakeupInProgress = false;
         }
 #endif