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, 63 insertions, 0 deletions

diff --git a/drivers/cdx/controller/mcdid.h b/drivers/cdx/controller/mcdid.h
new file mode 100644
index 000000000..7fc29f099
--- /dev/null
+++ b/drivers/cdx/controller/mcdid.h
@@ -0,0 +1,63 @@
+/* SPDX-License-Identifier: GPL-2.0
+ *
+ * Copyright 2008-2013 Solarflare Communications Inc.
+ * Copyright (C) 2022-2025, Advanced Micro Devices, Inc.
+ */
+
+#ifndef CDX_MCDID_H
+#define CDX_MCDID_H
+
+#include <linux/mutex.h>
+#include <linux/kref.h>
+#include <linux/rpmsg.h>
+
+#include "mc_cdx_pcol.h"
+
+#ifdef DEBUG
+#define CDX_WARN_ON_ONCE_PARANOID(x) WARN_ON_ONCE(x)
+#define CDX_WARN_ON_PARANOID(x) WARN_ON(x)
+#else
+#define CDX_WARN_ON_ONCE_PARANOID(x) do {} while (0)
+#define CDX_WARN_ON_PARANOID(x) do {} while (0)
+#endif
+
+#define MCDI_BUF_LEN (8 + MCDI_CTL_SDU_LEN_MAX)
+
+static inline struct cdx_mcdi_iface *cdx_mcdi_if(struct cdx_mcdi *cdx)
+{
+	return cdx->mcdi ? &cdx->mcdi->iface : NULL;
+}
+
+int cdx_mcdi_rpc_async(struct cdx_mcdi *cdx, unsigned int cmd,
+		       const struct cdx_dword *inbuf, size_t inlen,
+		       cdx_mcdi_async_completer *complete,
+		       unsigned long cookie);
+int cdx_mcdi_wait_for_quiescence(struct cdx_mcdi *cdx,
+				 unsigned int timeout_jiffies);
+
+/*
+ * We expect that 16- and 32-bit fields in MCDI requests and responses
+ * are appropriately aligned, but 64-bit fields are only
+ * 32-bit-aligned.
+ */
+#define MCDI_BYTE(_buf, _field)						\
+	((void)BUILD_BUG_ON_ZERO(MC_CMD_ ## _field ## _LEN != 1),	\
+	 *MCDI_PTR(_buf, _field))
+#define MCDI_WORD(_buf, _field)						\
+	((void)BUILD_BUG_ON_ZERO(MC_CMD_ ## _field ## _LEN != 2),	\
+	 le16_to_cpu(*(__force const __le16 *)MCDI_PTR(_buf, _field)))
+#define MCDI_POPULATE_DWORD_1(_buf, _field, _name1, _value1)		\
+	CDX_POPULATE_DWORD_1(*_MCDI_DWORD(_buf, _field),		\
+			     MC_CMD_ ## _name1, _value1)
+#define MCDI_SET_QWORD(_buf, _field, _value)				\
+	do {								\
+		CDX_POPULATE_DWORD_1(_MCDI_DWORD(_buf, _field)[0],	\
+				     CDX_DWORD, (u32)(_value));	\
+		CDX_POPULATE_DWORD_1(_MCDI_DWORD(_buf, _field)[1],	\
+				     CDX_DWORD, (u64)(_value) >> 32);	\
+	} while (0)
+#define MCDI_QWORD(_buf, _field)					\
+	(CDX_DWORD_FIELD(_MCDI_DWORD(_buf, _field)[0], CDX_DWORD) |	\
+	(u64)CDX_DWORD_FIELD(_MCDI_DWORD(_buf, _field)[1], CDX_DWORD) << 32)
+
+#endif /* CDX_MCDID_H */