[PATCH 08 of 10 v2] libxl: automatic placement deals with node-affinity

[Dario Faggioli <dario.faggioli@citrix.com>]

Which basically means the following two things:
 1) during domain creation, it is the node-affinity of
    the domain --rather than the vcpu-affinities of its
    VCPUs-- that is affected by automatic placement;
 2) during automatic placement, when counting how many
    VCPUs are already "bound" to a placement candidate
    (as part of the process of choosing the best
    candidate), both vcpu-affinity and node-affinity
    are considered.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Juergen Gross <juergen.gross@ts.fujitsu.com>

diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -133,13 +133,13 @@ static int numa_place_domain(libxl__gc *
 {
     int found;
     libxl__numa_candidate candidate;
-    libxl_bitmap candidate_nodemap;
+    libxl_bitmap cpupool_nodemap;
     libxl_cpupoolinfo cpupool_info;
     int i, cpupool, rc = 0;
     uint32_t memkb;
 
     libxl__numa_candidate_init(&candidate);
-    libxl_bitmap_init(&candidate_nodemap);
+    libxl_bitmap_init(&cpupool_nodemap);
 
     /*
      * Extract the cpumap from the cpupool the domain belong to. In fact,
@@ -156,7 +156,7 @@ static int numa_place_domain(libxl__gc *
     rc = libxl_domain_need_memory(CTX, info, &memkb);
     if (rc)
         goto out;
-    if (libxl_node_bitmap_alloc(CTX, &candidate_nodemap, 0)) {
+    if (libxl_node_bitmap_alloc(CTX, &cpupool_nodemap, 0)) {
         rc = ERROR_FAIL;
         goto out;
     }
@@ -174,17 +174,19 @@ static int numa_place_domain(libxl__gc *
     if (found == 0)
         goto out;
 
-    /* Map the candidate's node map to the domain's info->cpumap */
-    libxl__numa_candidate_get_nodemap(gc, &candidate, &candidate_nodemap);
-    rc = libxl_nodemap_to_cpumap(CTX, &candidate_nodemap, &info->cpumap);
+    /* Map the candidate's node map to the domain's info->nodemap */
+    libxl__numa_candidate_get_nodemap(gc, &candidate, &info->nodemap);
+
+    /* Avoid trying to set the affinity to nodes that might be in the
+     * candidate's nodemap but out of our cpupool. */
+    rc = libxl_cpumap_to_nodemap(CTX, &cpupool_info.cpumap,
+                                 &cpupool_nodemap);
     if (rc)
         goto out;
 
-    /* Avoid trying to set the affinity to cpus that might be in the
-     * nodemap but not in our cpupool. */
-    libxl_for_each_set_bit(i, info->cpumap) {
-        if (!libxl_bitmap_test(&cpupool_info.cpumap, i))
-            libxl_bitmap_reset(&info->cpumap, i);
+    libxl_for_each_set_bit(i, info->nodemap) {
+        if (!libxl_bitmap_test(&cpupool_nodemap, i))
+            libxl_bitmap_reset(&info->nodemap, i);
     }
 
     LOG(DETAIL, "NUMA placement candidate with %d nodes, %d cpus and "
@@ -193,7 +195,7 @@ static int numa_place_domain(libxl__gc *
 
  out:
     libxl__numa_candidate_dispose(&candidate);
-    libxl_bitmap_dispose(&candidate_nodemap);
+    libxl_bitmap_dispose(&cpupool_nodemap);
     libxl_cpupoolinfo_dispose(&cpupool_info);
     return rc;
 }
@@ -211,10 +213,10 @@ int libxl__build_pre(libxl__gc *gc, uint
     /*
      * Check if the domain has any CPU affinity. If not, try to build
      * up one. In case numa_place_domain() find at least a suitable
-     * candidate, it will affect info->cpumap accordingly; if it
+     * candidate, it will affect info->nodemap accordingly; if it
      * does not, it just leaves it as it is. This means (unless
      * some weird error manifests) the subsequent call to
-     * libxl_set_vcpuaffinity_all() will do the actual placement,
+     * libxl_domain_set_nodeaffinity() will do the actual placement,
      * whatever that turns out to be.
      */
     if (libxl_defbool_val(info->numa_placement)) {
diff --git a/tools/libxl/libxl_numa.c b/tools/libxl/libxl_numa.c
--- a/tools/libxl/libxl_numa.c
+++ b/tools/libxl/libxl_numa.c
@@ -184,7 +184,7 @@ static int nr_vcpus_on_nodes(libxl__gc *
                              int vcpus_on_node[])
 {
     libxl_dominfo *dinfo = NULL;
-    libxl_bitmap vcpu_nodemap;
+    libxl_bitmap dom_nodemap, vcpu_nodemap;
     int nr_doms, nr_cpus;
     int i, j, k;
 
@@ -197,6 +197,12 @@ static int nr_vcpus_on_nodes(libxl__gc *
         return ERROR_FAIL;
     }
 
+    if (libxl_node_bitmap_alloc(CTX, &dom_nodemap, 0) < 0) {
+        libxl_bitmap_dispose(&vcpu_nodemap);
+        libxl_dominfo_list_free(dinfo, nr_doms);
+        return ERROR_FAIL;
+    }
+
     for (i = 0; i < nr_doms; i++) {
         libxl_vcpuinfo *vinfo;
         int nr_dom_vcpus;
@@ -205,14 +211,21 @@ static int nr_vcpus_on_nodes(libxl__gc *
         if (vinfo == NULL)
             continue;
 
+        /* Retrieve the domain's node-affinity map */
+        libxl_domain_get_nodeaffinity(CTX, dinfo[i].domid, &dom_nodemap);
+
         for (j = 0; j < nr_dom_vcpus; j++) {
-            /* For each vcpu of each domain, increment the elements of
-             * the array corresponding to the nodes where the vcpu runs */
+            /*
+             * For each vcpu of each domain, it must have both vcpu-affinity
+             * and node-affinity to (a pcpu belonging to) a certain node to
+             * cause an increment in the corresponding element of the array.
+             */
             libxl_bitmap_set_none(&vcpu_nodemap);
             libxl_for_each_set_bit(k, vinfo[j].cpumap) {
                 int node = tinfo[k].node;
 
                 if (libxl_bitmap_test(suitable_cpumap, k) &&
+                    libxl_bitmap_test(&dom_nodemap, node) &&
                     !libxl_bitmap_test(&vcpu_nodemap, node)) {
                     libxl_bitmap_set(&vcpu_nodemap, node);
                     vcpus_on_node[node]++;
@@ -223,6 +236,7 @@ static int nr_vcpus_on_nodes(libxl__gc *
         libxl_vcpuinfo_list_free(vinfo, nr_dom_vcpus);
     }
 
+    libxl_bitmap_dispose(&dom_nodemap);
     libxl_bitmap_dispose(&vcpu_nodemap);
     libxl_dominfo_list_free(dinfo, nr_doms);
     return 0;