sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

1 files changed, 118 insertions, 0 deletions

diff --git a/tools/lib/perf/threadmap.c b/tools/lib/perf/threadmap.c
new file mode 100644
index 000000000..db431b036
--- /dev/null
+++ b/tools/lib/perf/threadmap.c
@@ -0,0 +1,118 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <perf/threadmap.h>
+#include <stdlib.h>
+#include <linux/refcount.h>
+#include <internal/threadmap.h>
+#include <string.h>
+#include <asm/bug.h>
+#include <stdio.h>
+
+static void perf_thread_map__reset(struct perf_thread_map *map, int start, int nr)
+{
+	size_t size = (nr - start) * sizeof(map->map[0]);
+
+	memset(&map->map[start], 0, size);
+	map->err_thread = -1;
+}
+
+struct perf_thread_map *perf_thread_map__realloc(struct perf_thread_map *map, int nr)
+{
+	size_t size = sizeof(*map) + sizeof(map->map[0]) * nr;
+	int start = map ? map->nr : 0;
+
+	map = realloc(map, size);
+	/*
+	 * We only realloc to add more items, let's reset new items.
+	 */
+	if (map)
+		perf_thread_map__reset(map, start, nr);
+
+	return map;
+}
+
+#define thread_map__alloc(__nr) perf_thread_map__realloc(NULL, __nr)
+
+void perf_thread_map__set_pid(struct perf_thread_map *map, int idx, pid_t pid)
+{
+	map->map[idx].pid = pid;
+}
+
+char *perf_thread_map__comm(struct perf_thread_map *map, int idx)
+{
+	return map->map[idx].comm;
+}
+
+struct perf_thread_map *perf_thread_map__new_array(int nr_threads, pid_t *array)
+{
+	struct perf_thread_map *threads = thread_map__alloc(nr_threads);
+	int i;
+
+	if (!threads)
+		return NULL;
+
+	for (i = 0; i < nr_threads; i++)
+		perf_thread_map__set_pid(threads, i, array ? array[i] : -1);
+
+	threads->nr = nr_threads;
+	refcount_set(&threads->refcnt, 1);
+
+	return threads;
+}
+
+struct perf_thread_map *perf_thread_map__new_dummy(void)
+{
+	return perf_thread_map__new_array(1, NULL);
+}
+
+static void perf_thread_map__delete(struct perf_thread_map *threads)
+{
+	if (threads) {
+		int i;
+
+		WARN_ONCE(refcount_read(&threads->refcnt) != 0,
+			  "thread map refcnt unbalanced\n");
+		for (i = 0; i < threads->nr; i++)
+			free(perf_thread_map__comm(threads, i));
+		free(threads);
+	}
+}
+
+struct perf_thread_map *perf_thread_map__get(struct perf_thread_map *map)
+{
+	if (map)
+		refcount_inc(&map->refcnt);
+	return map;
+}
+
+void perf_thread_map__put(struct perf_thread_map *map)
+{
+	if (map && refcount_dec_and_test(&map->refcnt))
+		perf_thread_map__delete(map);
+}
+
+int perf_thread_map__nr(struct perf_thread_map *threads)
+{
+	return threads ? threads->nr : 1;
+}
+
+pid_t perf_thread_map__pid(struct perf_thread_map *map, int idx)
+{
+	if (!map) {
+		assert(idx == 0);
+		return -1;
+	}
+
+	return map->map[idx].pid;
+}
+
+int perf_thread_map__idx(struct perf_thread_map *threads, pid_t pid)
+{
+	if (!threads)
+		return pid == -1 ? 0 : -1;
+
+	for (int i = 0; i < threads->nr; ++i) {
+		if (threads->map[i].pid == pid)
+			return i;
+	}
+	return -1;
+}