Triggering Hostboot Shutdown when PNOR is bad

Updating the doShutdown path to support receiving a reason
code as input.  Then changing PNOR RP to issue a shutdown
when problems are detected with the PNOR Partition table.

RTC: 44146
Change-Id: Ib4111d0a91f53d90fa100422a1463539897598e6
Reviewed-on: http://gfw160.austin.ibm.com:8080/gerrit/3024
Tested-by: Jenkins Server
Reviewed-by: Daniel M. Crowell <dcrowell@us.ibm.com>
Reviewed-by: A. Patrick Williams III <iawillia@us.ibm.com>

10 files changed, 113 insertions, 63 deletions

diff --git a/src/build/simics/hb-simdebug.py b/src/build/simics/hb-simdebug.py
index 12f4a2c17..048546b46 100755
--- a/src/build/simics/hb-simdebug.py
+++ b/src/build/simics/hb-simdebug.py
@@ -171,7 +171,7 @@ Examples: \n
 new_command("hb-errl",
     (lambda logid, logidStr, flg_l, flg_d:
         run_hb_debug_framework("Errl",
-                ("display="+(str(logid) if logid else logidStr) if flg_d else ""
+                ("display="+(("0x%x" % logid) if logid else logidStr) if flg_d else ""
                 ),
                 outputFile = "hb-errl.output")),
     [ arg(int_t, "logid", "?", None),
diff --git a/src/include/kernel/kernel_reasoncodes.H b/src/include/kernel/kernel_reasoncodes.H
index fe4748a19..f76d0993b 100644
--- a/src/include/kernel/kernel_reasoncodes.H
+++ b/src/include/kernel/kernel_reasoncodes.H
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2012                   */
+/* COPYRIGHT International Business Machines Corp. 2012,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -36,6 +36,9 @@
 
     enum KernelReasonCode
     {
+        //NOTE: RC_ASSERT and RC_SHUTDOWN must be used in one unique
+        //  locationto maintain uniqueness as they will be used
+        //  without a MODULE ID.
         RC_ASSERT            = KERNEL_COMP_ID | 0x01, /**< Assert */
         RC_SHUTDOWN          = KERNEL_COMP_ID | 0x02, /**< Shutdown */
     };
diff --git a/src/include/usr/errl/errlmanager.H b/src/include/usr/errl/errlmanager.H
index 3b9ad6572..75a5ae25e 100644
--- a/src/include/usr/errl/errlmanager.H
+++ b/src/include/usr/errl/errlmanager.H
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2011,2012              */
+/* COPYRIGHT International Business Machines Corp. 2011,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -176,6 +176,22 @@ private:
     };
 
     /**
+     * @enum  ERRORLOG_PLID_OFFSET
+     *
+     * Base ID of Hostboot PLIDs.  The hostboot plid range is 0x90 to 0x93
+     * (For each instance running on a multinode system.  Eventually the
+     * offset will need to be offset for the proper node and possibly some
+     * other initializer sent from FSP to keep PLIDs unique across IPLs.
+     *
+     * NOTE: Changes to this define may require changes to
+     *       CpuManager::requestShutdown
+     */
+    enum ERRORLOG_PLID_OFFSET
+    {
+        ERRLOG_PLID_BASE    = 0x90000000, /**< Hostboot Base PLID Offset */
+    };
+
+    /**
       * @brief   Disabled copy constructor and assignment operator
       */
     ErrlManager(const ErrlManager& i_right);
diff --git a/src/include/usr/pnor/pnor_reasoncodes.H b/src/include/usr/pnor/pnor_reasoncodes.H
index e8ad6341d..e580e2242 100644
--- a/src/include/usr/pnor/pnor_reasoncodes.H
+++ b/src/include/usr/pnor/pnor_reasoncodes.H
@@ -66,6 +66,9 @@ namespace PNOR
         RC_BAD_STARTUP_RC            = PNOR_COMP_ID | 0x09,
         RC_UNSUPORTED_HARDWARE       = PNOR_COMP_ID | 0x0A,
         RC_MICRON_INCOMPLETE         = PNOR_COMP_ID | 0x0B,
+        //RC_PARTITION_TABLE_INVALID Must be Unique by itself
+        //   for Shutdown path  
+        RC_PARTITION_TABLE_INVALID   = PNOR_COMP_ID | 0x0C,
     };
 };
 
diff --git a/src/kernel/cpumgr.C b/src/kernel/cpumgr.C
index 6e6b59860..954232b2d 100644
--- a/src/kernel/cpumgr.C
+++ b/src/kernel/cpumgr.C
@@ -111,11 +111,21 @@ void CpuManager::requestShutdown(uint64_t i_status)
     cv_shutdown_requested = true;
 
     // If the shutdown was not called with a Good shutdown status
-    // then we know we are shutting down due to error and the
-    // status passed back is instead a PLID
+    // then we know we are shutting down due to error.  We need to
+    // figure out if the error provided is a PLID or reasoncode
+    // and write it appropriately.
+    // Hostboot PLIDs always start with 0x9 (32-bit)
+    static const uint64_t PLID_MASK = 0x0000000090000000;
     if (i_status != SHUTDOWN_STATUS_GOOD)
     {
-        termWritePlid(TI_SHUTDOWN, i_status);
+        if ((i_status & 0x00000000F0000000) == PLID_MASK)
+        {
+            termWritePlid(TI_SHUTDOWN, i_status);
+        }
+        else
+        {
+            termWriteSRC(TI_SHUTDOWN,i_status, 0);
+        }
 
         printk("TI initiated on all threads (shutdown)\n");
     }
diff --git a/src/kernel/terminate.C b/src/kernel/terminate.C
index 10daec717..2c103c5af 100644
--- a/src/kernel/terminate.C
+++ b/src/kernel/terminate.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2012                   */
+/* COPYRIGHT International Business Machines Corp. 2012,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -64,7 +64,7 @@ void termWriteSRC(uint16_t i_source, uint16_t i_reasoncode,uint64_t i_failAddr)
     kernel_TIDataArea.src.ID = 0xBC;
     kernel_TIDataArea.src.subsystem = 0x11;
     kernel_TIDataArea.src.reasoncode = i_reasoncode;
-    kernel_TIDataArea.src.moduleID = MOD_KERNEL_TERMINATE;
+    kernel_TIDataArea.src.moduleID = 0;
     kernel_TIDataArea.src.iType = TI_WITH_SRC;
 
     // Update User Data with address of fail location
diff --git a/src/usr/errl/errlmanager.C b/src/usr/errl/errlmanager.C
index e95b6d2fa..4cce80016 100644
--- a/src/usr/errl/errlmanager.C
+++ b/src/usr/errl/errlmanager.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2011,2012              */
+/* COPYRIGHT International Business Machines Corp. 2011,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -74,13 +74,12 @@ char* g_ErrlStorage = new char[ ERRL_STORAGE_SIZE ];
 #define POINTER2OFFSET(p) ((reinterpret_cast<char*>(p))-(g_ErrlStorage))
 
 
-
 ///////////////////////////////////////////////////////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////
 ErrlManager::ErrlManager()
 {
     // PNOR will be reinitialized every time hostboot runs
-    iv_currLogId = 0;
+    iv_currLogId = ERRLOG_PLID_BASE;
 
     iv_hwasProcessCalloutFn = NULL;
 
diff --git a/src/usr/errl/parser/errlparser.C b/src/usr/errl/parser/errlparser.C
index 801da1635..71d71aab5 100644
--- a/src/usr/errl/parser/errlparser.C
+++ b/src/usr/errl/parser/errlparser.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2011,2012              */
+/* COPYRIGHT International Business Machines Corp. 2011,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -622,10 +622,10 @@ int main( int argc,  char *argv[] )
             }
             else
             {
-                int c = sscanf( argv[i], "%d", &ulLogId );
+                int c = sscanf( argv[i], "%x", &ulLogId );
                 if( c != 1 )
                 {
-                  fprintf( stdout, "Provide -d <decimal log ID>\n" );
+                  fprintf( stdout, "Provide -d <hex log ID>\n" );
                   exit( 2 );
                 }
             }
@@ -893,7 +893,7 @@ int main( int argc,  char *argv[] )
                 printf( "%s\n", szDivider );
                 fListHead  = 1;
             }
-            printf( "%-16s %8d\n",
+            printf( "%-16s 0X%8X\n",
                     FindComp(pPrivateHdr->sectionheader.compId),
                     pPrivateHdr->eid );
         }
@@ -901,7 +901,7 @@ int main( int argc,  char *argv[] )
         {
             // Write the native PEL to a temporary file
             // for x86 errl tool to display.
-            sprintf( szTmpFilename, "/tmp/pel%d.bin", pPrivateHdr->eid );
+            sprintf( szTmpFilename, "/tmp/pel%X.bin", pPrivateHdr->eid );
 
             int fd = open( szTmpFilename,  O_RDWR | O_CREAT , 0664 );
             if( -1 == fd )
@@ -974,7 +974,7 @@ int main( int argc,  char *argv[] )
         if( fExtractPEL )
         {
             // Write the native PEL to a temporary file for debug later.
-            sprintf( szTmpFilename, "%s/pel%d.bin", pszOutputDir, pPrivateHdr->eid );
+            sprintf( szTmpFilename, "%s/pel%X.bin", pszOutputDir, pPrivateHdr->eid );
 
             int fd = open( szTmpFilename,  O_RDWR | O_CREAT , 0664 );
             if( -1 == fd )
diff --git a/src/usr/hwpf/test/hwpisteperrortest.H b/src/usr/hwpf/test/hwpisteperrortest.H
index ba75d18ea..0af67a776 100644
--- a/src/usr/hwpf/test/hwpisteperrortest.H
+++ b/src/usr/hwpf/test/hwpisteperrortest.H
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2012                   */
+/* COPYRIGHT International Business Machines Corp. 2012,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -72,30 +72,34 @@ public:
 
                 uint64_t l_data0 = new_errl->getUserData1();
 
-                uint32_t eid = ( l_data0  & 0xFFFFFFF00000000) >> 32;
+                uint32_t eid = ( l_data0  & 0xFFFFFFFF00000000) >> 32;
                 uint32_t reason = (uint32_t)(l_data0 & 0x00000000FFFFFFFF);
 
                 if( eid != l_errl->eid() )
                 {
-                    TS_FAIL("expected eid == l_errl->eid() "
-                              "eid rebuilt from user data of "
-                              "IStepError did not match original error eid");
+                    TS_FAIL("testIstepError1: expected"
+                            "eid[0x%.8x] == l_errl->eid()[0x%.8x] "
+                            "eid rebuilt from user data of "
+                            "IStepError did not match original error eid",
+                            eid, l_errl->eid());
                 }
                 else
                 {
-                    TS_TRACE("passed: eid == l_errl->eid()");
+                    TS_TRACE("testIstepError1: passed: eid == l_errl->eid()");
                 }
 
                 if( reason != l_errl->reasonCode() )
                 {
-                    TS_FAIL("expected reasonCode == l_errl->reasonCode() \
+                    TS_FAIL("testIstepError1: "
+                            "expected reasonCode == l_errl->reasonCode() \
                               reasonCode rebuilt from user data of \
                               IStepError did not match original reasoncode");
 
                 }
                 else
                 {
-                    TS_TRACE("passed: reason == l_errl->reasonCode()");
+                    TS_TRACE("testIstepError1: passed: "
+                             "reason == l_errl->reasonCode()");
                 }
 
 
@@ -179,30 +183,34 @@ public:
 
                 uint64_t l_data0 = new_errl->getUserData1();
 
-                uint32_t eid = ( l_data0  & 0xFFFFFFF00000000) >> 32;
+                uint32_t eid = ( l_data0  & 0xFFFFFFFF00000000) >> 32;
                 uint32_t reason = (uint32_t)(l_data0 & 0x00000000FFFFFFFF);
 
                 if( eid != l_errl->eid() )
                 {
-                    TS_FAIL("expected eid == l_errl->eid() "
-                              "eid rebuilt from user data of "
-                              "IStepError did not match original error eid");
+                    TS_FAIL("testIStepError2: expected "
+                            "eid[0x%.8x] == l_errl->eid()[0x%.8x] "
+                            "eid rebuilt from user data of "
+                            "IStepError did not match original error eid",
+                            eid, l_errl->eid());
                 }
                 else
                 {
-                    TS_TRACE("passed: eid == l_errl->eid()");
+                    TS_TRACE("testIStepError2: passed: eid == l_errl->eid()");
                 }
 
                 if( reason != l_errl->reasonCode() )
                 {
-                    TS_FAIL("expected reasonCode == l_errl->reasonCode()"
-                              "reasonCode rebuilt from user data of"
-                              "IStepError did not match original reasoncode");
+                    TS_FAIL("testIStepError2: "
+                            "expected reasonCode == l_errl->reasonCode()"
+                            "reasonCode rebuilt from user data of"
+                            "IStepError did not match original reasoncode");
 
                 }
                 else
                 {
-                    TS_TRACE("passed: reason == l_errl->reasonCode()");
+                    TS_TRACE("testIStepError2: passed: "
+                             "reason == l_errl->reasonCode()");
                 }
 
                 // verify that we counted the error we added
diff --git a/src/usr/pnor/pnorrp.C b/src/usr/pnor/pnorrp.C
index 94e98ba99..03159fb15 100644
--- a/src/usr/pnor/pnorrp.C
+++ b/src/usr/pnor/pnorrp.C
@@ -5,7 +5,7 @@
 /*                                                                        */
 /* IBM CONFIDENTIAL                                                       */
 /*                                                                        */
-/* COPYRIGHT International Business Machines Corp. 2011,2012              */
+/* COPYRIGHT International Business Machines Corp. 2011,2013              */
 /*                                                                        */
 /* p1                                                                     */
 /*                                                                        */
@@ -371,54 +371,63 @@ errlHndl_t PnorRP::readTOC()
 
         //TODO: verify checksum of header  RTC: 44147
 
-        TRACFCOMP(g_trac_pnor, "PnorRp::readTOC:  FFS Block size = 0x%.8X, Partition Table Size = 0x%.8x, entry_count=%d", l_ffs_hdr->block_size, l_ffs_hdr->size, l_ffs_hdr->entry_count);
+        TRACFCOMP(g_trac_pnor, "PnorRp::readTOC:  FFS Block size = 0x%.8X, Partition Table Size = 0x%.8x, entry_count=%d",
+                  l_ffs_hdr->block_size, l_ffs_hdr->size, l_ffs_hdr->entry_count);
+
+        uint64_t spaceUsed = (sizeof(ffs_entry))*l_ffs_hdr->entry_count;
 
         /* Checking FFS Header to make sure it looks valid */
-        //TODO: Leave FFDC Breadcrumbs before issuing critical assert in the checks below..  RTC: 44146.
+        bool header_good = true;
         if(l_ffs_hdr->magic != FFS_MAGIC)
         {
-            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Invalid magic number in FFS header: 0x%.4X", l_ffs_hdr->magic);
-            crit_assert(0);
+            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Invalid magic number in FFS header: 0x%.4X",
+                      l_ffs_hdr->magic);
+            header_good = false;
         }
-
-        if(l_ffs_hdr->version != SUPPORTED_FFS_VERSION)
+        else if(l_ffs_hdr->version != SUPPORTED_FFS_VERSION)
         {
-            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported FFS Header version: 0x%.4X", l_ffs_hdr->version);
-            crit_assert(0);
+            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported FFS Header version: 0x%.4X",
+                      l_ffs_hdr->version);
+            header_good = false;
         }
-
-        if(l_ffs_hdr->entry_size != sizeof(ffs_entry))
+        else if(l_ffs_hdr->entry_size != sizeof(ffs_entry))
         {
             TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unexpected entry_size(0x%.8x) in FFS header: 0x%.4X", l_ffs_hdr->entry_size);
-            crit_assert(0);
+            header_good = false;
         }
-
-        if(l_ffs_hdr->entries == NULL)
+        else if(l_ffs_hdr->entries == NULL)
         {
             TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  FFS Header pointer to entries is NULL.");
-            crit_assert(0);
+            header_good = false;
         }
-
-        if(l_ffs_hdr->block_size != PAGESIZE)
+        else if(l_ffs_hdr->block_size != PAGESIZE)
         {
-            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported Block Size(0x%.4X). PNOR Blocks must be 4k", l_ffs_hdr->block_size);
-            crit_assert(0);
+            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported Block Size(0x%.4X). PNOR Blocks must be 4k",
+                      l_ffs_hdr->block_size);
+            header_good = false;
         }
-
-        if(l_ffs_hdr->block_count == 0)
+        else if(l_ffs_hdr->block_count == 0)
         {
-            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported BLock COunt(0x%.4X). Device cannot be zero blocks in length.", l_ffs_hdr->block_count);
-            crit_assert(0);
+            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Unsupported BLock COunt(0x%.4X). Device cannot be zero blocks in length.",
+                      l_ffs_hdr->block_count);
+            header_good = false;
         }
-
         //Make sure all the entries fit in specified partition table size.
-        uint64_t spaceUsed = (sizeof(ffs_entry))*l_ffs_hdr->entry_count;
-        if(spaceUsed > ((l_ffs_hdr->block_size * l_ffs_hdr->size) - sizeof(ffs_hdr)))
+        else if(spaceUsed >
+                ((l_ffs_hdr->block_size * l_ffs_hdr->size) - sizeof(ffs_hdr)))
         {
-            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  FFS Entries (0x%.16X) go past end of FFS Table.", spaceUsed);
+            TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  FFS Entries (0x%.16X) go past end of FFS Table.",
+                      spaceUsed);
             crit_assert(0);
         }
 
+        if(!header_good)
+        {
+            //Shutdown if we detected a partition table issue for any reason
+            INITSERVICE::doShutdown( PNOR::RC_PARTITION_TABLE_INVALID);
+            while(1) { task_yield(); }
+        }
+
         ffs_hb_user_t* ffsUserData = NULL;
 
         //Walk through all the entries in the table and parse the data.
@@ -480,8 +489,10 @@ errlHndl_t PnorRP::readTOC()
 
             if((iv_TOC[cur_side][secId].flashAddr + iv_TOC[cur_side][secId].size) > (l_ffs_hdr->block_count*PAGESIZE))
             {
-                TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Partition(%s) at base address (0x%.8x) extends past end of flash device", cur_entry->name, iv_TOC[cur_side][secId].flashAddr);
-                crit_assert(0);
+                TRACFCOMP(g_trac_pnor, "E>PnorRp::readTOC:  Partition(%s) at base address (0x%.8x) extends past end of flash device",
+                          cur_entry->name, iv_TOC[cur_side][secId].flashAddr);
+                INITSERVICE::doShutdown( PNOR::RC_PARTITION_TABLE_INVALID);
+                while(1) { task_yield(); }
             }
 
             cur_entry++;