Hostboot Serviceability Review Part 2

Resolving TODOs and ensuring error logs have correct callouts

Change-Id: Ic2e65427487fb91553ffe4ed6e3ed922004963ba
RTC: 92837
Reviewed-on: http://gfw160.aus.stglabs.ibm.com:8080/gerrit/8374
Tested-by: Jenkins Server
Reviewed-by: A. Patrick Williams III <iawillia@us.ibm.com>

3 files changed, 48 insertions, 97 deletions

diff --git a/src/usr/hwas/common/deconfigGard.C b/src/usr/hwas/common/deconfigGard.C
index fe51d0914..522ba02b6 100644
--- a/src/usr/hwas/common/deconfigGard.C
+++ b/src/usr/hwas/common/deconfigGard.C
@@ -40,24 +40,6 @@
 // Trace definition
 #define __COMP_TD__ g_trac_deconf
 
-// TODO The DeconfigGard code needs to trace a target. The current recommended
-// way is to get the Target's PHYS_PATH attribute and do a binary trace.
-// However, the size of a EntityPath is more than 16 bytes. This code
-// will trace only the first 16 bytes (which in most cases is enough) to avoid a
-// multi-line binary trace. This all seems a little convoluted. Is there a
-// better way to trace a Target
-#define DG_DBG_TARGET(string, pPath) \
-    HWAS_DBG_BIN(string, pPath, sizeof(EntityPath) - 1)
-#define DG_INF_TARGET(string, pPath) \
-    HWAS_INF_BIN(string, pPath, sizeof(EntityPath) - 1)
-#define DG_ERR_TARGET(string, pPath) \
-    HWAS_ERR_BIN(string, pPath, sizeof(EntityPath) - 1)
-
-// TODO There are a number of error logs created in this file. Most of them
-// should include the target identifier (PHYS_PATH). There is a plan in RTC
-// story 4110 to provide a way to easily add a target to an error log. When that
-// is done need to update the error logs
-
 namespace HWAS
 {
 
@@ -266,8 +248,10 @@ errlHndl_t DeconfigGard::deconfigureTargetsFromGardRecordsForIpl(
             {
                 // could be a platform specific target for the other
                 // ie, we are hostboot and this is an FSP target, or vice-versa
-                DG_INF_TARGET("Could not find Target for",
-                               &(l_gardRecord.iv_targetId));
+                // Binary trace the iv_targetId (EntityPath)
+                HWAS_INF_BIN("Could not find Target for:",
+                             &(l_gardRecord.iv_targetId),
+                             sizeof(l_gardRecord.iv_targetId));
                 continue;
             }
 
@@ -1343,7 +1327,7 @@ void DeconfigGard::_deconfigureTarget(Target & i_target,
 //******************************************************************************
 void DeconfigGard::_doDeconfigureActions(Target & i_target)
 {
-    // TODO
+    // Placeholder for any necessary deconfigure actions
 }
 
 //******************************************************************************
diff --git a/src/usr/hwas/hostbootIstep.C b/src/usr/hwas/hostbootIstep.C
index c7b5bff79..15d587357 100644
--- a/src/usr/hwas/hostbootIstep.C
+++ b/src/usr/hwas/hostbootIstep.C
@@ -108,50 +108,30 @@ void* host_discover_targets( void *io_pArgs )
     // Check whether we're in MPIPL mode
     TARGETING::Target* l_pTopLevel = NULL;
     targetService().getTopLevelTarget( l_pTopLevel );
+    HWAS_ASSERT(l_pTopLevel, "HWAS host_discover_targets: no TopLevelTarget");
 
-    if( l_pTopLevel == NULL )
+    if (l_pTopLevel->getAttr<ATTR_IS_MPIPL_HB>())
     {
-        TRACFCOMP( ISTEPS_TRACE::g_trac_isteps_trace,
-                   "Top level handle was NULL" );
-
-        /*@
-         * @errortype
-         * @severity     ERRORLOG::ERRL_SEV_UNRECOVERABLE
-         * @moduleid     HWAS::MOD_HOST_DISCOVER_TARGETS
-         * @reasoncode   HWAS::RC_TOP_LEVEL_TARGET_NULL
-         * @devdesc      Call to get top level targeting handle
-         *               returned NULL
-         */
-        errl =  hwasError( ERRORLOG::ERRL_SEV_UNRECOVERABLE,
-                           HWAS::MOD_HOST_DISCOVER_TARGETS,
-                           HWAS::RC_TOP_LEVEL_TARGET_NULL );
+        TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "MPIPL mode");
+
+        // Sync attributes from Fsp
+        errl = syncAllAttributesFromFsp();
     }
     else
     {
-        if (l_pTopLevel->getAttr<ATTR_IS_MPIPL_HB>())
-        {
-            TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "MPIPL mode");
+        TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "Normal IPL mode");
 
-            // Sync attributes from Fsp
-            errl = syncAllAttributesFromFsp();
+        errl = discoverTargets();
 
-        }
-        else
+        // also if SP doesn't support change detection, call
+        // function to do it here.
+        if (!errl &&
+            !l_pTopLevel->getAttr<ATTR_SP_FUNCTIONS>()
+                .hardwareChangeDetection)
         {
-            TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "Normal IPL mode");
-
-            errl = discoverTargets();
-
-            // also if SP doesn't support change detection, call
-            // function to do it here.
-            if (!errl &&
-                !l_pTopLevel->getAttr<ATTR_SP_FUNCTIONS>()
-                    .hardwareChangeDetection)
-            {
-                TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace,
-                    "calling hwasChangeDetection");
-                errl = hwasChangeDetection();
-            }
+            TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace,
+                "calling hwasChangeDetection");
+            errl = hwasChangeDetection();
         }
     }
 
@@ -173,47 +153,27 @@ void* host_gard( void *io_pArgs )
     // Check whether we're in MPIPL mode
     TARGETING::Target* l_pTopLevel = NULL;
     targetService().getTopLevelTarget( l_pTopLevel );
+    HWAS_ASSERT(l_pTopLevel, "HWAS host_gard: no TopLevelTarget");
 
-    if( l_pTopLevel == NULL )
+    if (l_pTopLevel->getAttr<ATTR_IS_MPIPL_HB>())
     {
-        TRACFCOMP( ISTEPS_TRACE::g_trac_isteps_trace,
-                   "Top level handle was NULL" );
-
-        /*@
-         * @errortype
-         * @severity     ERRORLOG::ERRL_SEV_UNRECOVERABLE
-         * @moduleid     HWAS::MOD_HOST_GARD
-         * @reasoncode   HWAS::RC_TOP_LEVEL_TARGET_NULL
-         * @devdesc      Call to get top level targeting handle
-         *               returned NULL
-         */
-        errl =  hwasError( ERRORLOG::ERRL_SEV_UNRECOVERABLE,
-                           HWAS::MOD_HOST_GARD,
-                           HWAS::RC_TOP_LEVEL_TARGET_NULL );
+        TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "MPIPL mode");
+
+        // we only want EX units to be processed
+        TARGETING::PredicateCTM l_exFilter(TARGETING::CLASS_UNIT,
+                                           TARGETING::TYPE_EX);
+        errl = collectGard(&l_exFilter);
     }
     else
     {
-        if (l_pTopLevel->getAttr<ATTR_IS_MPIPL_HB>())
-        {
-            TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "MPIPL mode");
-
-            // we only want EX units to be processed
-            TARGETING::PredicateCTM l_exFilter(TARGETING::CLASS_UNIT,
-                                                TARGETING::TYPE_EX);
-            errl = collectGard(&l_exFilter);
-        }
-        else
-        {
-            TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "Normal IPL mode");
+        TRACFCOMP(ISTEPS_TRACE::g_trac_isteps_trace, "Normal IPL mode");
 
-            errl = collectGard();
+        errl = collectGard();
 
-            // if no error,
-            if (errl == NULL)
-            {
-                //  check and see if we still have enough hardware to continue
-                errl = checkMinimumHardware();
-            }
+        if (errl == NULL)
+        {
+            //  check and see if we still have enough hardware to continue
+            errl = checkMinimumHardware();
         }
         // If targets are deconfigured as a result of host_gard, they are
         // done so using the PLID as the reason for deconfiguration.  This
diff --git a/src/usr/hwas/hwasPlatDeconfigGard.C b/src/usr/hwas/hwasPlatDeconfigGard.C
index 551373a15..99c679bbb 100644
--- a/src/usr/hwas/hwasPlatDeconfigGard.C
+++ b/src/usr/hwas/hwasPlatDeconfigGard.C
@@ -213,12 +213,12 @@ errlHndl_t DeconfigGard::platCreateGardRecord(
                 (static_cast<uint64_t>(get_huid(i_pTarget)) << 32) | i_errlEid;
             const uint64_t userdata2 =
                 (static_cast<uint64_t>(lDeconfigGardable) << 32) | lPresent;
-            l_pErr = hwasError(
-                ERRL_SEV_UNRECOVERABLE,
+            const bool hbSwError = true;
+            l_pErr = new ERRORLOG::ErrlEntry(
+                ERRORLOG::ERRL_SEV_UNRECOVERABLE,
                 HWAS::MOD_PLAT_DECONFIG_GARD,
                 HWAS::RC_TARGET_NOT_GARDABLE,
-                userdata1,
-                userdata2);
+                userdata1, userdata2, hbSwError);
             break;
         }
 
@@ -308,6 +308,13 @@ errlHndl_t DeconfigGard::platCreateGardRecord(
         {
             HWAS_ERR("GARD Record Repository full");
 
+            // TODO RTC 96397
+            // Hostboot will only write GARD Records to PNOR when it is the
+            // gardRecordMaster. An error will be logged if GARD Record storage
+            // exceeds 90% and the GARD Record will not be written if full. The
+            // error will have a new procedure callout requesting that the
+            // machine be serviced. Right now, this error log has no callouts.
+
             /*@
              * @errortype
              * @moduleid     HWAS::MOD_PLAT_DECONFIG_GARD
@@ -319,8 +326,8 @@ errlHndl_t DeconfigGard::platCreateGardRecord(
             const uint64_t userdata1 =
                 (static_cast<uint64_t> (get_huid(i_pTarget)) << 32) |
                 i_errlEid;
-            l_pErr = hwasError(
-                ERRL_SEV_UNRECOVERABLE,
+            l_pErr = new ERRORLOG::ErrlEntry(
+                ERRORLOG::ERRL_SEV_UNRECOVERABLE,
                 HWAS::MOD_PLAT_DECONFIG_GARD,
                 HWAS::RC_GARD_REPOSITORY_FULL,
                 userdata1);