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, 37 insertions, 0 deletions

diff --git a/Documentation/devicetree/changesets.rst b/Documentation/devicetree/changesets.rst
new file mode 100644
index 000000000..2091468d4
--- /dev/null
+++ b/Documentation/devicetree/changesets.rst
@@ -0,0 +1,37 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+=====================
+Devicetree Changesets
+=====================
+
+A Devicetree changeset is a method which allows one to apply changes
+in the live tree in such a way that either the full set of changes
+will be applied, or none of them will be. If an error occurs partway
+through applying the changeset, then the tree will be rolled back to the
+previous state. A changeset can also be removed after it has been
+applied.
+
+When a changeset is applied, all of the changes get applied to the tree
+at once before emitting OF_RECONFIG notifiers. This is so that the
+receiver sees a complete and consistent state of the tree when it
+receives the notifier.
+
+The sequence of a changeset is as follows.
+
+1. of_changeset_init() - initializes a changeset
+
+2. A number of DT tree change calls, of_changeset_attach_node(),
+   of_changeset_detach_node(), of_changeset_add_property(),
+   of_changeset_remove_property, of_changeset_update_property() to prepare
+   a set of changes. No changes to the active tree are made at this point.
+   All the change operations are recorded in the of_changeset 'entries'
+   list.
+
+3. of_changeset_apply() - Apply the changes to the tree. Either the
+   entire changeset will get applied, or if there is an error the tree will
+   be restored to the previous state. The core ensures proper serialization
+   through locking. An unlocked version __of_changeset_apply is available,
+   if needed.
+
+If a successfully applied changeset needs to be removed, it can be done
+with of_changeset_revert().