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, 84 insertions, 0 deletions

diff --git a/include/uapi/asm-generic/statfs.h b/include/uapi/asm-generic/statfs.h
new file mode 100644
index 000000000..bd35c7619
--- /dev/null
+++ b/include/uapi/asm-generic/statfs.h
@@ -0,0 +1,84 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_GENERIC_STATFS_H
+#define _UAPI_GENERIC_STATFS_H
+
+#include <linux/types.h>
+
+
+/*
+ * Most 64-bit platforms use 'long', while most 32-bit platforms use '__u32'.
+ * Yes, they differ in signedness as well as size.
+ * Special cases can override it for themselves -- except for S390x, which
+ * is just a little too special for us. And MIPS, which I'm not touching
+ * with a 10' pole.
+ */
+#ifndef __statfs_word
+#if __BITS_PER_LONG == 64
+#define __statfs_word __kernel_long_t
+#else
+#define __statfs_word __u32
+#endif
+#endif
+
+struct statfs {
+	__statfs_word f_type;
+	__statfs_word f_bsize;
+	__statfs_word f_blocks;
+	__statfs_word f_bfree;
+	__statfs_word f_bavail;
+	__statfs_word f_files;
+	__statfs_word f_ffree;
+	__kernel_fsid_t f_fsid;
+	__statfs_word f_namelen;
+	__statfs_word f_frsize;
+	__statfs_word f_flags;
+	__statfs_word f_spare[4];
+};
+
+/*
+ * ARM needs to avoid the 32-bit padding at the end, for consistency
+ * between EABI and OABI 
+ */
+#ifndef ARCH_PACK_STATFS64
+#define ARCH_PACK_STATFS64
+#endif
+
+struct statfs64 {
+	__statfs_word f_type;
+	__statfs_word f_bsize;
+	__u64 f_blocks;
+	__u64 f_bfree;
+	__u64 f_bavail;
+	__u64 f_files;
+	__u64 f_ffree;
+	__kernel_fsid_t f_fsid;
+	__statfs_word f_namelen;
+	__statfs_word f_frsize;
+	__statfs_word f_flags;
+	__statfs_word f_spare[4];
+} ARCH_PACK_STATFS64;
+
+/* 
+ * IA64 and x86_64 need to avoid the 32-bit padding at the end,
+ * to be compatible with the i386 ABI
+ */
+#ifndef ARCH_PACK_COMPAT_STATFS64
+#define ARCH_PACK_COMPAT_STATFS64
+#endif
+
+struct compat_statfs64 {
+	__u32 f_type;
+	__u32 f_bsize;
+	__u64 f_blocks;
+	__u64 f_bfree;
+	__u64 f_bavail;
+	__u64 f_files;
+	__u64 f_ffree;
+	__kernel_fsid_t f_fsid;
+	__u32 f_namelen;
+	__u32 f_frsize;
+	__u32 f_flags;
+	__u32 f_spare[4];
+} ARCH_PACK_COMPAT_STATFS64;
+
+#endif /* _UAPI_GENERIC_STATFS_H */