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, 154 insertions, 0 deletions

diff --git a/drivers/base/map.c b/drivers/base/map.c
new file mode 100644
index 000000000..bde415f93
--- /dev/null
+++ b/drivers/base/map.c
@@ -0,0 +1,154 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/drivers/base/map.c
+ *
+ * (C) Copyright Al Viro 2002,2003
+ *
+ * NOTE: data structure needs to be changed.  It works, but for large dev_t
+ * it will be too slow.  It is isolated, though, so these changes will be
+ * local to that file.
+ */
+
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/mutex.h>
+#include <linux/kdev_t.h>
+#include <linux/kobject.h>
+#include <linux/kobj_map.h>
+
+struct kobj_map {
+	struct probe {
+		struct probe *next;
+		dev_t dev;
+		unsigned long range;
+		struct module *owner;
+		kobj_probe_t *get;
+		int (*lock)(dev_t, void *);
+		void *data;
+	} *probes[255];
+	struct mutex *lock;
+};
+
+int kobj_map(struct kobj_map *domain, dev_t dev, unsigned long range,
+	     struct module *module, kobj_probe_t *probe,
+	     int (*lock)(dev_t, void *), void *data)
+{
+	unsigned int n = MAJOR(dev + range - 1) - MAJOR(dev) + 1;
+	unsigned int index = MAJOR(dev);
+	unsigned int i;
+	struct probe *p;
+
+	if (n > 255)
+		n = 255;
+
+	p = kmalloc_objs(struct probe, n);
+	if (p == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < n; i++, p++) {
+		p->owner = module;
+		p->get = probe;
+		p->lock = lock;
+		p->dev = dev;
+		p->range = range;
+		p->data = data;
+	}
+	mutex_lock(domain->lock);
+	for (i = 0, p -= n; i < n; i++, p++, index++) {
+		struct probe **s = &domain->probes[index % 255];
+		while (*s && (*s)->range < range)
+			s = &(*s)->next;
+		p->next = *s;
+		*s = p;
+	}
+	mutex_unlock(domain->lock);
+	return 0;
+}
+
+void kobj_unmap(struct kobj_map *domain, dev_t dev, unsigned long range)
+{
+	unsigned int n = MAJOR(dev + range - 1) - MAJOR(dev) + 1;
+	unsigned int index = MAJOR(dev);
+	unsigned int i;
+	struct probe *found = NULL;
+
+	if (n > 255)
+		n = 255;
+
+	mutex_lock(domain->lock);
+	for (i = 0; i < n; i++, index++) {
+		struct probe **s;
+		for (s = &domain->probes[index % 255]; *s; s = &(*s)->next) {
+			struct probe *p = *s;
+			if (p->dev == dev && p->range == range) {
+				*s = p->next;
+				if (!found)
+					found = p;
+				break;
+			}
+		}
+	}
+	mutex_unlock(domain->lock);
+	kfree(found);
+}
+
+struct kobject *kobj_lookup(struct kobj_map *domain, dev_t dev, int *index)
+{
+	struct kobject *kobj;
+	struct probe *p;
+	unsigned long best = ~0UL;
+
+retry:
+	mutex_lock(domain->lock);
+	for (p = domain->probes[MAJOR(dev) % 255]; p; p = p->next) {
+		struct kobject *(*probe)(dev_t, int *, void *);
+		struct module *owner;
+		void *data;
+
+		if (p->dev > dev || p->dev + p->range - 1 < dev)
+			continue;
+		if (p->range - 1 >= best)
+			break;
+		if (!try_module_get(p->owner))
+			continue;
+		owner = p->owner;
+		data = p->data;
+		probe = p->get;
+		best = p->range - 1;
+		*index = dev - p->dev;
+		if (p->lock && p->lock(dev, data) < 0) {
+			module_put(owner);
+			continue;
+		}
+		mutex_unlock(domain->lock);
+		kobj = probe(dev, index, data);
+		/* Currently ->owner protects _only_ ->probe() itself. */
+		module_put(owner);
+		if (kobj)
+			return kobj;
+		goto retry;
+	}
+	mutex_unlock(domain->lock);
+	return NULL;
+}
+
+struct kobj_map *kobj_map_init(kobj_probe_t *base_probe, struct mutex *lock)
+{
+	struct kobj_map *p = kmalloc_obj(struct kobj_map);
+	struct probe *base = kzalloc_obj(*base);
+	int i;
+
+	if ((p == NULL) || (base == NULL)) {
+		kfree(p);
+		kfree(base);
+		return NULL;
+	}
+
+	base->dev = 1;
+	base->range = ~0;
+	base->get = base_probe;
+	for (i = 0; i < 255; i++)
+		p->probes[i] = base;
+	p->lock = lock;
+	return p;
+}