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, 72 insertions, 0 deletions

diff --git a/arch/alpha/lib/strchr.S b/arch/alpha/lib/strchr.S
new file mode 100644
index 000000000..68c54ff50
--- /dev/null
+++ b/arch/alpha/lib/strchr.S
@@ -0,0 +1,72 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * arch/alpha/lib/strchr.S
+ * Contributed by Richard Henderson (rth@tamu.edu)
+ *
+ * Return the address of a given character within a null-terminated
+ * string, or null if it is not found.
+ */
+#include <linux/export.h>
+#include <asm/regdef.h>
+
+	.set noreorder
+	.set noat
+
+	.align 3
+	.globl strchr
+	.ent strchr
+strchr:
+	.frame sp, 0, ra
+	.prologue 0
+
+	zapnot	a1, 1, a1	# e0    : zero extend the search character
+	ldq_u   t0, 0(a0)	# .. e1 : load first quadword
+	sll	a1, 8, t5	# e0    : replicate the search character
+	andnot  a0, 7, v0	# .. e1 : align our loop pointer
+	or	t5, a1, a1	# e0    :
+	lda	t4, -1		# .. e1 : build garbage mask
+	sll	a1, 16, t5	# e0    :
+	cmpbge  zero, t0, t2	# .. e1 : bits set iff byte == zero
+	mskqh	t4, a0, t4	# e0    :
+	or	t5, a1, a1	# .. e1 :
+	sll	a1, 32, t5	# e0    :
+	cmpbge	zero, t4, t4	# .. e1 : bits set iff byte is garbage
+	or	t5, a1, a1	# e0    :
+	xor	t0, a1, t1	# .. e1 : make bytes == c zero
+	cmpbge  zero, t1, t3	# e0    : bits set iff byte == c
+	or	t2, t3, t0	# e1    : bits set iff char match or zero match
+	andnot	t0, t4, t0	# e0    : clear garbage bits
+	bne	t0, $found	# .. e1 (zdb)
+
+$loop:	ldq	t0, 8(v0)	# e0    :
+	addq	v0, 8, v0	# .. e1 :
+	nop			# e0    :
+	xor	t0, a1, t1	# .. e1 (ev5 data stall)
+	cmpbge	zero, t0, t2	# e0    : bits set iff byte == 0
+	cmpbge	zero, t1, t3	# .. e1 : bits set iff byte == c
+	or	t2, t3, t0	# e0    :
+	beq	t0, $loop	# .. e1 (zdb)
+
+$found:	negq    t0, t1		# e0    : clear all but least set bit
+	and     t0, t1, t0	# e1 (stall)
+
+	and	t0, t3, t1	# e0    : bit set iff byte was the char
+	beq	t1, $retnull	# .. e1 (zdb)
+
+	and     t0, 0xf0, t2	# e0    : binary search for that set bit
+	and	t0, 0xcc, t3	# .. e1 :
+	and	t0, 0xaa, t4	# e0    :
+	cmovne	t2, 4, t2	# .. e1 :
+	cmovne	t3, 2, t3	# e0    :
+	cmovne	t4, 1, t4	# .. e1 :
+	addq	t2, t3, t2	# e0    :
+	addq	v0, t4, v0	# .. e1 :
+	addq	v0, t2, v0	# e0    :
+	ret			# .. e1 :
+
+$retnull:
+	mov	zero, v0	# e0    :
+	ret			# .. e1 :
+
+	.end strchr
+	EXPORT_SYMBOL(strchr)