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, 124 insertions, 0 deletions

diff --git a/arch/alpha/boot/head.S b/arch/alpha/boot/head.S
new file mode 100644
index 000000000..06a7c95fe
--- /dev/null
+++ b/arch/alpha/boot/head.S
@@ -0,0 +1,124 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * arch/alpha/boot/head.S
+ *
+ * initial bootloader stuff..
+ */
+
+#include <asm/pal.h>
+
+	.set noreorder
+	.globl	__start
+	.ent	__start
+__start:
+	br	$29,2f
+2:	ldgp	$29,0($29)
+	jsr	$26,start_kernel
+	call_pal PAL_halt
+	.end __start
+
+	.align 5
+	.globl	wrent
+	.ent	wrent
+wrent:
+	.prologue 0
+	call_pal PAL_wrent
+	ret ($26)
+	.end wrent
+
+	.align 5
+	.globl	wrkgp
+	.ent	wrkgp
+wrkgp:
+	.prologue 0
+	call_pal PAL_wrkgp
+	ret ($26)
+	.end wrkgp
+
+	.align 5
+	.globl	switch_to_osf_pal
+	.ent	switch_to_osf_pal
+switch_to_osf_pal:
+	subq	$30,128,$30
+	.frame	$30,128,$26
+	stq	$26,0($30)
+	stq	$1,8($30)
+	stq	$2,16($30)
+	stq	$3,24($30)
+	stq	$4,32($30)
+	stq	$5,40($30)
+	stq	$6,48($30)
+	stq	$7,56($30)
+	stq	$8,64($30)
+	stq	$9,72($30)
+	stq	$10,80($30)
+	stq	$11,88($30)
+	stq	$12,96($30)
+	stq	$13,104($30)
+	stq	$14,112($30)
+	stq	$15,120($30)
+	.prologue 0
+
+	stq	$30,0($17)	/* save KSP in PCB */
+
+	bis	$30,$30,$20	/* a4 = KSP */
+	br	$17,1f
+
+	ldq	$26,0($30)
+	ldq	$1,8($30)
+	ldq	$2,16($30)
+	ldq	$3,24($30)
+	ldq	$4,32($30)
+	ldq	$5,40($30)
+	ldq	$6,48($30)
+	ldq	$7,56($30)
+	ldq	$8,64($30)
+	ldq	$9,72($30)
+	ldq	$10,80($30)
+	ldq	$11,88($30)
+	ldq	$12,96($30)
+	ldq	$13,104($30)
+	ldq	$14,112($30)
+	ldq	$15,120($30)
+	addq	$30,128,$30
+	ret ($26)
+1:	call_pal PAL_swppal
+	.end	switch_to_osf_pal
+
+	.align 3
+	.globl	tbi
+	.ent	tbi
+tbi:
+	.prologue 0
+	call_pal PAL_tbi
+	ret	($26)
+	.end tbi
+
+	.align 3
+	.globl	halt
+	.ent	halt
+halt:
+	.prologue 0
+	call_pal PAL_halt
+	.end halt
+
+/* $16 - new stack page */
+	.align 3
+	.globl	move_stack
+	.ent	move_stack
+move_stack:
+	.prologue 0
+	lda	$0, 0x1fff($31)
+	and	$0, $30, $1			/* Stack offset */
+	or	$1, $16, $16			/* New stack pointer */
+	mov	$30, $1
+	mov	$16, $2
+1:	ldq	$3, 0($1)			/* Move the stack */
+	addq	$1, 8, $1
+	stq	$3, 0($2)
+	and	$0, $1, $4
+	addq	$2, 8, $2
+	bne	$4, 1b
+	mov	$16, $30
+	ret	($26)
+	.end move_stack