Fix rollover of PLID numbers

Need to avoid incrementing into the processor node
field of the PLID.

Decided to use the highest bit in the last 3 bytes as a dividing bit.
[0x##800000 - 0x##FFFFFF] -- for pre-boot setup errors.
[0x##000000 - 0x##7FFFFF] -- for post setup errors

When setting up from PNOR, we will look for the max EID in the post-setup range.
If no max is found, restart at 0.

Change-Id: I4a8811d9c93039090504ca9cf8c0eebc6812a4ca
CQ:SW416592
Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/54362
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: Daniel M. Crowell <dcrowell@us.ibm.com>

4 files changed, 71 insertions, 15 deletions

diff --git a/src/include/usr/errl/errlmanager.H b/src/include/usr/errl/errlmanager.H
index 93165a821..f7407d8c4 100644
--- a/src/include/usr/errl/errlmanager.H
+++ b/src/include/usr/errl/errlmanager.H
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2011,2017                        */
+/* Contributors Listed Below - COPYRIGHT 2011,2018                        */
 /* [+] Google Inc.                                                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
@@ -276,7 +276,9 @@ private:
 #endif
         ERRLOG_PLID_NODE_SHIFT  = 24,        // shift to put node number 0x9#
         ERRLOG_PLID_MASK        = 0x00FFFFFF,// mask to find log number
-        ERRLOG_PLID_INITIAL     = 0x00FF0000,// initial big id number
+        ERRLOG_PLID_INITIAL     = 0x00800000,// min number for pre-boot
+        ERRLOG_PLID_PRE_MAX     = 0x00FFFFFF,// max number during pre-boot
+        ERRLOG_PLID_POST_MAX    = 0x007FFFFF,// max number after pre-boot
     };
 
     /**
@@ -311,7 +313,7 @@ private:
     /**
      * @brief Access PNOR and get the address and size of the HBEL section in
      * PNOR; sets the iv_pnorAddr, iv_maxErrlInPnor, iv_pnorOpenSlot variables;
-     * parsers the error logs in PNOR and determines the new iv_currLogId.
+     * parses the error logs in PNOR and determines the new iv_currLogId.
      *
      * @param[in/out] NONE
      * @return NONE.
@@ -409,6 +411,19 @@ private:
     uint32_t    iv_currLogId;
 
     /**
+      * @brief    Base & Node part of log ID.
+      *  This value will be used as a common ID among this node's error logs
+      */
+    uint32_t    iv_baseNodeId;
+
+    /**
+     *  @brief   Indicates if PNOR is ready for handling error logs
+     *  Will cause a switch of MIN/MAX values used in calculating the
+     *  next iv_currLogId.
+     */
+    bool        iv_pnorReadyForErrorLogs;
+
+    /**
       * @brief
       * Pointer to the header that precedes the error log storage buffer
       * in L3 RAM.  This may go away when we adopt PNOR, or else become
diff --git a/src/usr/errl/errlmanager.C b/src/usr/errl/errlmanager.C
index 2f465a049..724c03a63 100644
--- a/src/usr/errl/errlmanager.C
+++ b/src/usr/errl/errlmanager.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2011,2017                        */
+/* Contributors Listed Below - COPYRIGHT 2011,2018                        */
 /* [+] Google Inc.                                                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
@@ -117,6 +117,7 @@ AtLoadFunctions atLoadFunction;
 ///////////////////////////////////////////////////////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////
 ErrlManager::ErrlManager() :
+    iv_pnorReadyForErrorLogs(false),
     iv_hwasProcessCalloutFn(NULL),
     iv_msgQ(NULL),
     iv_pnorAddr(NULL),
@@ -162,8 +163,8 @@ ErrlManager::ErrlManager() :
                                     masterCpu.groupId :
                                     masterCpu.groupId + 4;
 
-    iv_currLogId = ERRLOG_PLID_BASE + ERRLOG_PLID_INITIAL +
-                        (l_eid_id << ERRLOG_PLID_NODE_SHIFT);
+    iv_baseNodeId = ERRLOG_PLID_BASE + (l_eid_id << ERRLOG_PLID_NODE_SHIFT);
+    iv_currLogId = iv_baseNodeId | ERRLOG_PLID_INITIAL;
 
     // next, we need to look thru PNOR and see what error records are there;
     //   ours will be 1 after the highest found.
diff --git a/src/usr/errl/errlmanager_common.C b/src/usr/errl/errlmanager_common.C
index 832b7748e..a9982d67e 100644
--- a/src/usr/errl/errlmanager_common.C
+++ b/src/usr/errl/errlmanager_common.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2015,2017                        */
+/* Contributors Listed Below - COPYRIGHT 2015,2018                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
 /*                                                                        */
@@ -79,7 +79,37 @@ uint8_t getHiddenLogsEnable( )
 // Atomically increment log id and return it.
 uint32_t ErrlManager::getUniqueErrId()
 {
-    return (__sync_add_and_fetch(&iv_currLogId, 1));
+    uint32_t l_logId = 0;
+    uint32_t l_nextLogId = 1;
+
+    while (1)
+    {
+        l_logId = iv_currLogId;
+
+        if (!iv_pnorReadyForErrorLogs)
+        {
+            // Range [__800000 - __FFFFFF]
+            l_nextLogId = ( iv_baseNodeId |
+                             (((l_logId + 1) & ERRLOG_PLID_MASK) |
+                                ERRLOG_PLID_INITIAL) );
+        }
+        else
+        {
+            // Range [__000000 - __7FFFFFF]
+            l_nextLogId = ( iv_baseNodeId |
+                            ((l_logId + 1) & ERRLOG_PLID_POST_MAX) );
+        }
+
+        if (__sync_bool_compare_and_swap(&iv_currLogId, l_logId, l_nextLogId))
+        {
+            // To maintain thread-safeness (do not use iv_currLogId here)
+            // Do the same operation to l_logId as this is the same value
+            // that iv_currLogID had before its operation
+            l_logId = l_nextLogId;
+            break;
+        }
+    }
+    return l_logId;
 }
 
 // ------------------------------------------------------------------
@@ -117,7 +147,8 @@ void ErrlManager::setupPnorInfo()
         // so that our first save will increment and wrap correctly
         iv_pnorOpenSlot = (iv_maxErrlInPnor - 1);
 
-        // walk thru memory, finding error logs and determine the highest ID
+        // walk thru memory, finding error logs and
+        // determine the highest ID within the lower POST range of EIDs
         uint32_t l_maxId = 0;
         for (uint32_t i = 0; i < iv_maxErrlInPnor; i++)
         {
@@ -134,7 +165,7 @@ void ErrlManager::setupPnorInfo()
                 }
                 // if this is 'my' type of plid (HB or HBRT) see if it's max
                 if (((l_id & FIRST_BYTE_ERRLOG) == ERRLOG_PLID_BASE ) &&
-                    (l_id > l_maxId ))
+                    (l_id > l_maxId ) && (l_id <= ERRLOG_PLID_POST_MAX))
                 {
                     l_maxId = l_id;
 
@@ -203,10 +234,18 @@ void ErrlManager::setupPnorInfo()
         } // for
 
         // bump the current eid to 1 past the max eid found
-        while (!__sync_bool_compare_and_swap(&iv_currLogId, iv_currLogId,
-                    (iv_currLogId & ERRLOG_PLID_BASE_MASK) +
-                    (l_maxId & ERRLOG_PLID_MASK) + 1));
-        TRACFCOMP( g_trac_errl, INFO_MRK"setupPnorInfo reseting LogId 0x%X", iv_currLogId);
+        // Stay within the non-preboot range
+        while ( !__sync_bool_compare_and_swap(&iv_currLogId, iv_currLogId,
+                 (iv_baseNodeId |
+                 ((l_maxId + 1) & ERRLOG_PLID_POST_MAX))) );
+
+        // set this to change new max/min values for iv_currLogId
+        // new max = ERRLOG_PLID_POST_MAX, new min = 0
+        iv_pnorReadyForErrorLogs = true;
+        TRACFCOMP( g_trac_errl,
+            INFO_MRK"setupPnorInfo resetting LogId number 0x%X",
+            iv_currLogId);
+
 
         // if error(s) came in before PNOR was ready,
         // the error log(s) would be on this list. save now.
diff --git a/src/usr/errl/runtime/rt_errlmanager.C b/src/usr/errl/runtime/rt_errlmanager.C
index 4361a691d..10b521a32 100644
--- a/src/usr/errl/runtime/rt_errlmanager.C
+++ b/src/usr/errl/runtime/rt_errlmanager.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* OpenPOWER HostBoot Project                                             */
 /*                                                                        */
-/* Contributors Listed Below - COPYRIGHT 2013,2017                        */
+/* Contributors Listed Below - COPYRIGHT 2013,2018                        */
 /* [+] International Business Machines Corp.                              */
 /*                                                                        */
 /*                                                                        */
@@ -66,6 +66,7 @@ bool rt_processCallout(errlHndl_t &io_errl,
 ///////////////////////////////////////////////////////////////////////////////
 ErrlManager::ErrlManager() :
         iv_currLogId(0),
+        iv_pnorReadyForErrorLogs(true),
         iv_pStorage(NULL),
         iv_hwasProcessCalloutFn(NULL),
         iv_pnorAddr(NULL),