Deconfig EC/EX/EQ at runtime

Applied the normal parent/child rules for EC/EX/EQ logic to be done in
the runtime deconfig logic.

After applying the deconfig logic, then reset the PM for the parent
proc of the EX/EX/EQ.

To facilitate the code changes, modified the return value of the
TARGETING::getParent(...) call.  It no longer returns a const * Target,
considering 5 times out of 7 of it's usage, users were just
staic_cast(ing) away the const.

Clear out the HB reset attribute in the mpipl path.

Modify PRD logic to use runtime deconfig call in malfunction alert
path.

CMVC-Prereq: 1066157
Change-Id: If81f2f15c6e84bf2adb6fd66c1321b8f1ddf1a86
CQ: SW443157
Reviewed-on: http://rchgit01.rchland.ibm.com/gerrit1/64873
Tested-by: Jenkins Server <pfd-jenkins+hostboot@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: Roland Veloz <rveloz@us.ibm.com>
Reviewed-by: Daniel M. Crowell <dcrowell@us.ibm.com>

src/include/usr/hwas/common/deconfigGard.H

@@ -488,7 +488,7 @@ public:
      *
      * NOTE: Currently only core targets are supported
      *
-     * @param[in]  i_pTarget   const pointer to the core target to deconfigure
+     * @param[in]  i_pTarget   pointer to the target to deconfigure
      * @param[in]  i_deconfigureAction Deconfigure options -
      *                                              FULLY_AT_RUNTIME
      * @param[in]  i_deconfigErrl - error log assocated with this deconfig
@@ -500,7 +500,7 @@ public:
      * @return errlHndl_t. Error log handle.
      */
     errlHndl_t deconfigureTargetAtRuntime(
-            TARGETING::ConstTargetHandle_t const i_pTarget,
+            TARGETING::TargetHandle_t i_pTarget,
             const DeconfigureFlags i_deconfigureAction,
             const errlHndl_t i_deconfigErrl);
 
@@ -515,7 +515,7 @@ public:
      * child targets. Any errors deconfiguring targets are committed and
      * the function continues. The only errors returned are usage errors.
      *
-     * @param[in]  i_pTarget   const pointer to Target to deconfigure.
+     * @param[in]  i_pTarget   pointer to Target to deconfigure.
      * @param[in]  i_deconfigureAction Deconfigure options -
      *                                              FULLY_AT_RUNTIME
      * @param[in]  i_deconfigErrl - error log assocated with this deconfig
@@ -529,7 +529,7 @@ public:
 
 #ifdef __HOSTBOOT_RUNTIME
     errlHndl_t platDeconfigureTargetAtRuntime(
-            TARGETING::ConstTargetHandle_t const i_pTarget,
+            TARGETING::TargetHandle_t i_pTarget,
             const DeconfigureFlags i_deconfigureAction,
             const errlHndl_t i_deconfigErrl);
 #endif

src/include/usr/targeting/common/utilFilter.H

@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2012,2016                        */
+/* Contributors Listed Below - COPYRIGHT 2012,2018                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
 /*                                                                        */
@@ -300,7 +300,7 @@ const Target * getParentChip( const Target * i_pChiplet );
  *
  * @return Parent target pointer or NULL if error
  */
-const Target * getParent( const Target * i_unit , TARGETING::TYPE &i_pType);
+Target * getParent( const Target * i_unit , TARGETING::TYPE &i_pType);
 
 /**
  * @brief return the EX target of the specified input core chiplet

src/usr/diag/prdf/plat/p9/prdfP9PmRecovery.C

@@ -67,7 +67,6 @@ int32_t pmRecovery( ExtensibleChip * i_chip, STEP_CODE_DATA_STRUCT & io_sc )
     // Get the Global Errorlog PLID and EID
     errlHndl_t globalErrl =
         ServiceGeneratorClass::ThisServiceGenerator().getErrl();
-    uint32_t eid  = globalErrl->eid();
     uint32_t plid = globalErrl->plid();
 
     // Runtime deconfig lost cores
@@ -86,8 +85,10 @@ int32_t pmRecovery( ExtensibleChip * i_chip, STEP_CODE_DATA_STRUCT & io_sc )
 
             // Call Deconfig
             errlHndl_t errl = nullptr;
-            errl = HWAS::theDeconfigGard().deconfigureTarget( *coreTgt, eid,
-                                 NULL, HWAS::DeconfigGard::FULLY_AT_RUNTIME );
+            errl = HWAS::theDeconfigGard().deconfigureTargetAtRuntime(
+                       coreTgt,
+                       HWAS::DeconfigGard::FULLY_AT_RUNTIME,
+                       globalErrl );
 
             if (errl)
             {

src/usr/errl/runtime/makefile

@@ -37,6 +37,8 @@ EXTRAINCDIR += ${ROOTPATH}/src/import/chips/common/utils/imageProcs
 EXTRAINCDIR += ${ROOTPATH}/src/import/chips/p9/procedures/hwp/ffdc
 EXTRAINCDIR += ${ROOTPATH}/src/import/chips/p9/common/include
 
+##      support for pm_common
+EXTRAINCDIR += ${ROOTPATH}/src/usr/isteps
 
 # need find the source for this hwp in runtime env
 VPATH += ${ROOTPATH}/src/import/chips/p9/procedures/hwp/pm

src/usr/hwas/common/deconfigGard.C

@@ -1640,15 +1640,24 @@ void DeconfigGard::_deconfigureByAssoc(
         _deconfigureByAssoc(*pChild, i_errlEid, i_deconfigRule);
     } // for CHILD
 
-    if ((i_deconfigRule == NOT_AT_RUNTIME)||
-            (i_deconfigRule == SPEC_DECONFIG))
+    // Retrieve the target type from the given target
+    TYPE l_targetType = i_target.getAttr<ATTR_TYPE>();
+
+    if ((i_deconfigRule == NOT_AT_RUNTIME) ||
+        (i_deconfigRule == SPEC_DECONFIG)  ||
+        (l_targetType == TYPE_EQ)          ||
+        (l_targetType == TYPE_EX)          ||
+        (l_targetType == TYPE_CORE))
     {
         // if the rule is NOT_AT_RUNTIME and we got here, then we are
         // not at runtime.
         // only do these 'by association' checks if we are NOT at runtime
         // reason is, we're not really deconfigureing anything, we're just
         // marking them as non-functional. we only want to do that for the
         // desired target and it's CHILD
+        // Except for target EQ, EX, CORE.  Deconfigure these regardless of the
+        // runtime status
+
 
         // find all CHILD_BY_AFFINITY targets and deconfigure them
         targetService().getAssociated(pChildList, &i_target,
@@ -1673,7 +1682,7 @@ void DeconfigGard::_deconfigureByAssoc(
         // chip  (TYPE_EQ, TYPE_EX, TYPE_CORE)
         // deconfigureByAssociation rules
 
-        switch (i_target.getAttr<ATTR_TYPE>())
+        switch (l_targetType)
         {
             case TYPE_CORE:
             {
@@ -3069,7 +3078,7 @@ bool DeconfigGard::anyChildFCO (Target & i_parent)
 // deconfigureTargetAtRuntime
 /******************************************************************************/
 errlHndl_t DeconfigGard::deconfigureTargetAtRuntime(
-        TARGETING::ConstTargetHandle_t const i_pTarget,
+        TARGETING::TargetHandle_t i_pTarget,
         const DeconfigGard::DeconfigureFlags i_deconfigureAction,
         const errlHndl_t i_deconfigErrl)
 

src/usr/hwas/common/hwas.C

@@ -1500,8 +1500,7 @@ bool isDescFunctional(const TARGETING::TargetHandle_t &i_desc,
     if (i_desc->getAttr<ATTR_TYPE>() == TYPE_OBUS_BRICK)
     {
         auto obusType = TARGETING::TYPE_OBUS;
-        TARGETING::Target* l_obus_ptr = const_cast<TARGETING::Target*>(
-                                            getParent(i_desc, obusType));
+        TARGETING::Target* l_obus_ptr = getParent(i_desc, obusType);
 
         //If NPU is bad and OBUS is non-SMP, then mark them bad
         // Bit does not matter unless not in SMP mode
@@ -1590,8 +1589,7 @@ void forceEcExEqDeconfig(const TARGETING::TargetHandle_t i_core,
     auto exType = TARGETING::TYPE_EX;
     auto eqType = TARGETING::TYPE_EQ;
 
-    TARGETING::Target* l_ex = const_cast<TARGETING::Target*>(
-                                    getParent(i_core, exType));
+    TARGETING::Target* l_ex = getParent(i_core, exType);
     getChildChiplets(pECList, l_ex, TYPE_CORE, true);
     if(pECList.size() == 0)
     {
@@ -1600,8 +1598,7 @@ void forceEcExEqDeconfig(const TARGETING::TargetHandle_t i_core,
                  l_ex->getAttr<ATTR_HUID>(), i_present ? "" : "NOT ");
 
         //Now get the parent EQ and check to see if it should be deconfigured
-        TARGETING::Target* l_eq = const_cast<TARGETING::Target*>(
-                                         getParent(l_ex, eqType));
+        TARGETING::Target* l_eq = getParent(l_ex, eqType);
         getChildChiplets(pEXList, l_eq, TYPE_EX, true);
         if(pEXList.size() == 0)
         {

src/usr/hwas/hwasPlatDeconfigGard.C

@@ -46,6 +46,7 @@
 #include <initservice/initserviceif.H>
 
 #include <pnor/pnorif.H>
+#include <pm/pm_common.H>                  // HBPM::resetPMComplex
 
 #ifdef __HOSTBOOT_RUNTIME
 #include <runtime/interface.h>             // g_hostInterfaces
@@ -733,7 +734,7 @@ errlHndl_t hwasError(const uint8_t i_sev,
 // platDeconfigureTargetAtRuntime
 /******************************************************************************/
 errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
-        TARGETING::ConstTargetHandle_t const i_pTarget,
+        TARGETING::TargetHandle_t i_pTarget,
         const DeconfigureFlags i_deconfigureAction,
         const errlHndl_t i_deconfigErrl)
 {
@@ -762,16 +763,23 @@ errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
             break;
         }
 
-        if(i_pTarget->getAttr<TARGETING::ATTR_TYPE>() !=
-                TARGETING::TYPE_CORE)
+        // Retrieve the target type from the given target
+        TYPE l_targetType = i_pTarget->getAttr<ATTR_TYPE>();
+
+        // Make sure we are only working with the following types
+        if( (l_targetType != TYPE_EQ) &&
+            (l_targetType != TYPE_EX) &&
+            (l_targetType != TYPE_CORE) )
         {
+            HWAS_ERR("Caller passed invalid type: 0x%08X", l_targetType);
+
             // only supporting cores
             /*@
              * @errortype
              * @moduleid     MOD_RUNTIME_DECONFIG
              * @reasoncode   RC_INVALID_TARGET
-             * @devdesc      Target is not a TYPE_CORE
-             * userdata1     target huid
+             * @devdesc      Target is neiter TYPE_EQ, TYPE_EX nor TYPE_CORE
+             * @userdata1    target huid
              * @custdesc     Host Firmware encountered an internal
              *               error
              */
@@ -786,7 +794,7 @@ errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
         {
             case DeconfigGard::FULLY_AT_RUNTIME:
 
-                HWAS_INF(" Deconfig action FULLY_AT_RUNTIME :0x%08X",
+                HWAS_INF("Deconfig action FULLY_AT_RUNTIME :0x%08X",
                         DeconfigGard::FULLY_AT_RUNTIME);
 
                 break;
@@ -813,10 +821,12 @@ errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
                 break;
         }
 
-    }while(0);
+        // If an error then exit while loop
+        if (l_errl)
+        {
+           break;
+        }
 
-    if(l_errl == nullptr)
-    {
         uint32_t  l_deconfigReason =  (i_deconfigErrl) ? i_deconfigErrl->eid() :
             DeconfigGard::DECONFIGURED_BY_PRD;
 
@@ -826,7 +836,6 @@ errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
                 get_huid(i_pTarget),i_deconfigureAction,
                 l_deconfigReason);
 
-
         bool l_isTargetDeconfigured = false;
         // deconfigureTarget() checks for targets that can be deconfigured at
         // runtime
@@ -836,38 +845,90 @@ errlHndl_t DeconfigGard::platDeconfigureTargetAtRuntime(
                                     &l_isTargetDeconfigured,
                                     i_deconfigureAction);
 
-        if(l_errl == nullptr && l_isTargetDeconfigured)
+        // If there was an error OR the target was not deconfigured, then
+        // do not continue and exit
+        if (l_errl || !l_isTargetDeconfigured)
         {
-            HWAS_INF("platDeconfigureTargetAtRuntime() - "
-                    "deconfigure successful");
+            HWAS_INF("platDeconfigureTargetAtRuntime() - deconfigure failed");
+            break;
+        }
+
+        HWAS_INF("platDeconfigureTargetAtRuntime() - "
+                 "deconfigure successful");
 
-            TARGETING::TYPE  l_type = TARGETING::TYPE_PROC;
-            const TARGETING::Target * l_parent =
-                                TARGETING::getParent(i_pTarget,l_type);
-            // get the parent proc and call the hwp to alert
-            // pm not to attempt to manage this core anymore
-            const fapi2::Target<fapi2::TARGET_TYPE_PROC_CHIP>
-                l_proc(const_cast<TARGETING::Target *>(l_parent));
+        TARGETING::TYPE  l_type = TARGETING::TYPE_PROC;
+        TARGETING::Target * l_parent =
+                            TARGETING::getParent(i_pTarget,l_type);
 
-            HWAS_INF("platDeconfigureTargetAtRuntime() - "
-                    "calling p9_update_ec_eq_state");
-            FAPI_INVOKE_HWP( l_errl,p9_update_ec_eq_state,l_proc);
+        // There might be a recursion possible with PRD and PM Reset so
+        //  don't call into reset twice
+        auto l_pmResetInProgress =
+          l_parent->getAttr<ATTR_HB_INITIATED_PM_RESET>();
+        if( HB_INITIATED_PM_RESET_IN_PROGRESS != l_pmResetInProgress )
+        {
+            // set ATTR_HB_INITIATED_PM_RESET to IN_PROGRESS to allow
+            //  special handling for PRD
+            l_parent->setAttr<ATTR_HB_INITIATED_PM_RESET>
+              (HB_INITIATED_PM_RESET_IN_PROGRESS);
+
+            HWAS_INF("platDeconfigureTargetAtRuntime() - Calling resetPMComplex");
+            l_errl = HBPM::resetPMComplex(l_parent);
 
             if(l_errl)
             {
-                HWAS_ERR("platDeconfigureTargetAtRuntime() - "
-                        "call to p9_update_ec_eq_state() failed on proc "
-                        "with HUID : %d",TARGETING::get_huid(l_proc));
-                ERRORLOG::ErrlUserDetailsTarget(l_proc).addToLog(l_errl);
-            }
+                HWAS_ERR("Failed call to HBPM::resetPMComplex on target "
+                         "with HUID : %d",TARGETING::get_huid(l_parent));
+                ERRORLOG::ErrlUserDetailsTarget(l_parent).addToLog(l_errl);
+
+                // set ATTR_HB_INITIATED_PM_RESET back to INACTIVE to allow
+                //  future recoveries to completely run
+                l_parent->setAttr<ATTR_HB_INITIATED_PM_RESET>
+                  (HB_INITIATED_PM_RESET_INACTIVE);
 
+                // If an error then exit while loop
+                break;
+            }
+            else
+            {
+                HWAS_INF("Successful call to HBPM::resetPMComplex on target"
+                         " with HUID : %.8X, setting ATTR_HB_INITIATED_PM_RESET to"
+                         " COMPLETE(%d)",
+                         TARGETING::get_huid(l_parent),
+                         HB_INITIATED_PM_RESET_COMPLETE);
+                // set ATTR_HB_INITIATED_PM_RESET to ACTIVE (reset IS in progress)
+                // and continue
+                l_parent->setAttr<ATTR_HB_INITIATED_PM_RESET>
+                  (HB_INITIATED_PM_RESET_COMPLETE);
+            }
         }
         else
         {
-            HWAS_INF("platDeconfigureTargetAtRuntime() - deconfigure failed");
+            HWAS_INF("platDeconfigureTargetAtRuntime() - skippign call to resetPMComplex - ATTR_HB_INITIATED_PM_RESET=%d", l_pmResetInProgress );
         }
-    }
-    HWAS_INF(">>>platDeconfigureTargetAtRuntime()" );
+
+        // get the parent proc and call the hwp to alert
+        // pm not to attempt to manage this core anymore
+        const fapi2::Target<fapi2::TARGET_TYPE_PROC_CHIP>
+            l_proc(l_parent);
+
+        HWAS_INF("platDeconfigureTargetAtRuntime() - "
+                "calling p9_update_ec_eq_state");
+        FAPI_INVOKE_HWP( l_errl,p9_update_ec_eq_state,
+                         l_proc,true/*skip qssr*/);
+
+        if(l_errl)
+        {
+            HWAS_ERR("platDeconfigureTargetAtRuntime() - "
+                    "call to p9_update_ec_eq_state() failed on proc "
+                    "with HUID : %d",TARGETING::get_huid(l_proc));
+            ERRORLOG::ErrlUserDetailsTarget(l_proc).addToLog(l_errl);
+
+            // If an error then exit while loop
+            break;
+        }
+    }while(0);
+
+    HWAS_INF("<<<platDeconfigureTargetAtRuntime()" );
 
     return l_errl ;
 }

src/usr/hwas/makefile

@@ -5,7 +5,7 @@
 #
 # OpenPOWER HostBoot Project
 #
-# Contributors Listed Below - COPYRIGHT 2011,2016
+# Contributors Listed Below - COPYRIGHT 2011,2018
 # [+] International Business Machines Corp.
 #
 #
@@ -31,6 +31,13 @@ EXTRAINCDIR += ${ROOTPATH}/src/include/usr/hwas/common
 EXTRAINCDIR += ${ROOTPATH}/src/include/usr/ecmddatabuffer
 EXTRAINCDIR += ${ROOTPATH}/src/usr/sbe
 
+##      support for pm_common
+EXTRAINCDIR += ${ROOTPATH}/src/usr/isteps
+EXTRAINCDIR += ${ROOTPATH}/src/import/hwpf/fapi2/include/
+EXTRAINCDIR += ${ROOTPATH}/src/include/usr/fapi2/
+EXTRAINCDIR += ${ROOTPATH}/src/import/chips/common/utils/imageProcs/
+EXTRAINCDIR += ${ROOTPATH}/src/import/chips/p9/procedures/hwp/ffdc
+
 VPATH += ${ROOTPATH}/src/usr/hwas/plat
 VPATH += ${ROOTPATH}/src/usr/hwas/common
 

src/usr/hwas/test/hwas1test.H

@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2011,2017                        */
+/* Contributors Listed Below - COPYRIGHT 2011,2018                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
 /*                                                                        */
@@ -1660,9 +1660,7 @@ public:
 
                             // get parent OBUS and check OPTICS_CONFIG_MODE
                             auto obusType = TARGETING::TYPE_OBUS;
-                            TARGETING::Target* l_obus_ptr =
-                                const_cast<TARGETING::Target*>(
-                                            getParent(pDesc, obusType));
+                            TARGETING::Target* l_obus_ptr = getParent(pDesc, obusType);
                             ATTR_OPTICS_CONFIG_MODE_type config_mode =
                                 l_obus_ptr->getAttr<ATTR_OPTICS_CONFIG_MODE>();