pmalloc: tolerate move_pages failure on numa node #0

Type: fix
Change-Id: I65dbc4f3db47c745a0187e015845e07b3b0568e5
Signed-off-by: Klement Sekera <ksekera@cisco.com>
diff --git a/src/vppinfra/pmalloc.c b/src/vppinfra/pmalloc.c
index 27738e1..e5fdc26 100644
--- a/src/vppinfra/pmalloc.c
+++ b/src/vppinfra/pmalloc.c
@@ -309,7 +309,8 @@
      to support non-numa kernels */
   clib_mem_get_page_stats (va, CLIB_MEM_PAGE_SZ_DEFAULT, 1, &stats);
 
-  if (stats.per_numa[numa_node] != 1)
+  if (stats.per_numa[numa_node] != 1 &&
+      !(numa_node == 0 && stats.unknown == 1))
     {
       u16 allocated_at = ~0;
       if (stats.unknown)