core: introduce NUMAPolicy and NUMAMask options

Make possible to set NUMA allocation policy for manager. Manager's
policy is by default inherited to all forked off processes. However, it
is possible to override the policy on per-service basis. Currently we
support, these policies: default, prefer, bind, interleave, local.
See man 2 set_mempolicy for details on each policy.

Overall NUMA policy actually consists of two parts. Policy itself and
bitmask representing NUMA nodes where is policy effective. Node mask can
be specified using related option, NUMAMask. Default mask can be
overwritten on per-service level.

1 files changed, 93 insertions, 0 deletions

diff --git a/src/shared/cpu-set-util.c b/src/shared/cpu-set-util.c
index b0036c7f61..f27543dfe2 100644
--- a/src/shared/cpu-set-util.c
+++ b/src/shared/cpu-set-util.c
@@ -7,12 +7,20 @@
 
 #include "alloc-util.h"
 #include "cpu-set-util.h"
+#include "dirent-util.h"
+#include "errno-util.h"
 #include "extract-word.h"
+#include "fd-util.h"
 #include "log.h"
 #include "macro.h"
 #include "memory-util.h"
+#include "missing_syscall.h"
 #include "parse-util.h"
+#include "stat-util.h"
 #include "string-util.h"
+#include "string-table.h"
+#include "strv.h"
+#include "util.h"
 
 char* cpu_set_to_string(const CPUSet *a) {
         _cleanup_free_ char *str = NULL;
@@ -287,3 +295,88 @@ int cpu_set_from_dbus(const uint8_t *bits, size_t size, CPUSet *set) {
         s = (CPUSet) {};
         return 0;
 }
+
+bool numa_policy_is_valid(const NUMAPolicy *policy) {
+        assert(policy);
+
+        if (!mpol_is_valid(numa_policy_get_type(policy)))
+                return false;
+
+        if (!policy->nodes.set &&
+            !IN_SET(numa_policy_get_type(policy), MPOL_DEFAULT, MPOL_LOCAL, MPOL_PREFERRED))
+                return false;
+
+        if (policy->nodes.set &&
+            numa_policy_get_type(policy) == MPOL_PREFERRED &&
+            CPU_COUNT_S(policy->nodes.allocated, policy->nodes.set) != 1)
+                return false;
+
+        return true;
+}
+
+static int numa_policy_to_mempolicy(const NUMAPolicy *policy, unsigned long *ret_maxnode, unsigned long **ret_nodes) {
+        unsigned node, bits = 0, ulong_bits;
+        _cleanup_free_ unsigned long *out = NULL;
+
+        assert(policy);
+        assert(ret_maxnode);
+        assert(ret_nodes);
+
+        if (IN_SET(numa_policy_get_type(policy), MPOL_DEFAULT, MPOL_LOCAL) ||
+            (numa_policy_get_type(policy) == MPOL_PREFERRED && !policy->nodes.set)) {
+                *ret_nodes = NULL;
+                *ret_maxnode = 0;
+                return 0;
+        }
+
+        bits = policy->nodes.allocated * 8;
+        ulong_bits = sizeof(unsigned long) * 8;
+
+        out = new0(unsigned long, DIV_ROUND_UP(policy->nodes.allocated, sizeof(unsigned long)));
+        if (!out)
+                return -ENOMEM;
+
+        /* We don't make any assumptions about internal type libc is using to store NUMA node mask.
+           Hence we need to convert the node mask to the representation expected by set_mempolicy() */
+        for (node = 0; node < bits; node++)
+                if (CPU_ISSET_S(node, policy->nodes.allocated, policy->nodes.set))
+                        out[node / ulong_bits] |= 1ul << (node % ulong_bits);
+
+        *ret_nodes = TAKE_PTR(out);
+        *ret_maxnode = bits + 1;
+        return 0;
+}
+
+int apply_numa_policy(const NUMAPolicy *policy) {
+        int r;
+        _cleanup_free_ unsigned long *nodes = NULL;
+        unsigned long maxnode;
+
+        assert(policy);
+
+        if (get_mempolicy(NULL, NULL, 0, 0, 0) < 0 && errno == ENOSYS)
+                return -EOPNOTSUPP;
+
+        if (!numa_policy_is_valid(policy))
+                return -EINVAL;
+
+        r = numa_policy_to_mempolicy(policy, &maxnode, &nodes);
+        if (r < 0)
+                return r;
+
+        r = set_mempolicy(numa_policy_get_type(policy), nodes, maxnode);
+        if (r < 0)
+                return -errno;
+
+        return 0;
+}
+
+static const char* const mpol_table[] = {
+        [MPOL_DEFAULT]    = "default",
+        [MPOL_PREFERRED]  = "preferred",
+        [MPOL_BIND]       = "bind",
+        [MPOL_INTERLEAVE] = "interleave",
+        [MPOL_LOCAL]      = "local",
+};
+
+DEFINE_STRING_TABLE_LOOKUP(mpol, int);