sched: add sched_numa_find_nth_cpu()

The function finds Nth set CPU in a given cpumask starting from a given
node.

Leveraging the fact that each hop in sched_domains_numa_masks includes the
same or greater number of CPUs than the previous one, we can use binary
search on hops instead of linear walk, which makes the overall complexity
of O(log n) in terms of number of cpumask_weight() calls.

Signed-off-by: Yury Norov <yury.norov@gmail.com>
Acked-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Peter Lafreniere <peter@n8pjl.ca>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

1 files changed, 57 insertions, 0 deletions

diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 8739c2a5a54e..2bf89186a10f 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -3,6 +3,8 @@
  * Scheduler topology setup/handling methods
  */
 
+#include <linux/bsearch.h>
+
 DEFINE_MUTEX(sched_domains_mutex);
 
 /* Protected by sched_domains_mutex: */
@@ -2067,6 +2069,61 @@ unlock:
 	return found;
 }
 
+struct __cmp_key {
+	const struct cpumask *cpus;
+	struct cpumask ***masks;
+	int node;
+	int cpu;
+	int w;
+};
+
+static int hop_cmp(const void *a, const void *b)
+{
+	struct cpumask **prev_hop = *((struct cpumask ***)b - 1);
+	struct cpumask **cur_hop = *(struct cpumask ***)b;
+	struct __cmp_key *k = (struct __cmp_key *)a;
+
+	if (cpumask_weight_and(k->cpus, cur_hop[k->node]) <= k->cpu)
+		return 1;
+
+	k->w = (b == k->masks) ? 0 : cpumask_weight_and(k->cpus, prev_hop[k->node]);
+	if (k->w <= k->cpu)
+		return 0;
+
+	return -1;
+}
+
+/*
+ * sched_numa_find_nth_cpu() - given the NUMA topology, find the Nth next cpu
+ *                             closest to @cpu from @cpumask.
+ * cpumask: cpumask to find a cpu from
+ * cpu: Nth cpu to find
+ *
+ * returns: cpu, or nr_cpu_ids when nothing found.
+ */
+int sched_numa_find_nth_cpu(const struct cpumask *cpus, int cpu, int node)
+{
+	struct __cmp_key k = { .cpus = cpus, .node = node, .cpu = cpu };
+	struct cpumask ***hop_masks;
+	int hop, ret = nr_cpu_ids;
+
+	rcu_read_lock();
+
+	k.masks = rcu_dereference(sched_domains_numa_masks);
+	if (!k.masks)
+		goto unlock;
+
+	hop_masks = bsearch(&k, k.masks, sched_domains_numa_levels, sizeof(k.masks[0]), hop_cmp);
+	hop = hop_masks	- k.masks;
+
+	ret = hop ?
+		cpumask_nth_and_andnot(cpu - k.w, cpus, k.masks[hop][node], k.masks[hop-1][node]) :
+		cpumask_nth_and(cpu, cpus, k.masks[0][node]);
+unlock:
+	rcu_read_unlock();
+	return ret;
+}
+EXPORT_SYMBOL_GPL(sched_numa_find_nth_cpu);
 #endif /* CONFIG_NUMA */
 
 static int __sdt_alloc(const struct cpumask *cpu_map)