Rediscover location of HDAT data to handle IO CM operations

PHYP modifies the location of the HDAT memory during a IO CM
operation.  During a mpipl, during dump collection, HB will
cache the location of the SPIRA.  This cache will be pointing
to the B side.  Later in the boot the FSP repopulates the SPIRA
using the A side.  HB needs to use the A side pointers instead.

Change-Id: I991fc731eb6292869e7280d29727ef270a6975a5
CQ: SW311349
Backport: release-fips830
Backport: release-fips820
Reviewed-on: http://gfw160.aus.stglabs.ibm.com:8080/gerrit/18836
Tested-by: Jenkins Server
Tested-by: Jenkins OP Build CI
Tested-by: Jenkins OP HW
Reviewed-by: Elizabeth Liner <eliner@us.ibm.com>
Reviewed-by: A. Patrick Williams III <iawillia@us.ibm.com>

1 files changed, 6 insertions, 0 deletions

diff --git a/src/usr/runtime/populate_attributes.C b/src/usr/runtime/populate_attributes.C
index 6a6024219..820dfc62a 100644
--- a/src/usr/runtime/populate_attributes.C
+++ b/src/usr/runtime/populate_attributes.C
@@ -245,6 +245,9 @@ errlHndl_t populate_system_attributes( uint64_t i_nodes )
     do {
         TRACDCOMP( g_trac_runtime, "-SYSTEM-" );
 
+        // Wipe out our cache of the NACA/SPIRA pointers
+        RUNTIME::rediscover_hdat();
+
         // find our memory range and fill it with some junk data
         uint64_t sys_data_addr = 0;
         uint64_t sys_data_size = 0;
@@ -415,6 +418,9 @@ errlHndl_t populate_node_attributes( uint64_t i_nodeNum )
         TRACDCOMP( g_trac_runtime, "-NODE-" );
         TRACFCOMP( g_trac_runtime, "populate_node_attributes(node=%d)", i_nodeNum );
 
+        // Wipe out our cache of the NACA/SPIRA pointers
+        RUNTIME::rediscover_hdat();
+
         // allocate memory and fill it with some junk data
         uint64_t node_data_addr = 0;
         size_t node_data_size = 0;