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, 86 insertions, 0 deletions

diff --git a/arch/alpha/lib/strncat.S b/arch/alpha/lib/strncat.S
new file mode 100644
index 000000000..a913a7c84
--- /dev/null
+++ b/arch/alpha/lib/strncat.S
@@ -0,0 +1,86 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * arch/alpha/lib/strncat.S
+ * Contributed by Richard Henderson (rth@tamu.edu)
+ *
+ * Append no more than COUNT characters from the null-terminated string SRC
+ * to the null-terminated string DST.  Always null-terminate the new DST.
+ *
+ * This differs slightly from the semantics in libc in that we never write
+ * past count, whereas libc may write to count+1.  This follows the generic
+ * implementation in lib/string.c and is, IMHO, more sensible.
+ */
+#include <linux/export.h>
+	.text
+
+	.align 3
+	.globl strncat
+	.ent strncat
+strncat:
+	.frame $30, 0, $26
+	.prologue 0
+
+	mov	$16, $0		# set up return value
+	beq	$18, $zerocount
+
+	/* Find the end of the string.  */
+
+	ldq_u   $1, 0($16)	# load first quadword ($16 may be misaligned)
+	lda     $2, -1($31)
+	insqh   $2, $16, $2
+	andnot  $16, 7, $16
+	or      $2, $1, $1
+	cmpbge  $31, $1, $2	# bits set iff byte == 0
+	bne     $2, $found
+
+$loop:	ldq     $1, 8($16)
+	addq    $16, 8, $16
+	cmpbge  $31, $1, $2
+	beq     $2, $loop
+
+$found:	negq    $2, $3		# clear all but least set bit
+	and     $2, $3, $2
+
+	and     $2, 0xf0, $3	# binary search for that set bit
+	and	$2, 0xcc, $4
+	and	$2, 0xaa, $5
+	cmovne	$3, 4, $3
+	cmovne	$4, 2, $4
+	cmovne	$5, 1, $5
+	addq	$3, $4, $3
+	addq	$16, $5, $16
+	addq	$16, $3, $16
+
+	/* Now do the append.  */
+
+	bsr	$23, __stxncpy
+
+	/* Worry about the null termination.  */
+
+	zapnot	$1, $27, $2	# was last byte a null?
+	bne	$2, 0f
+	ret
+
+0:	cmplt	$27, $24, $2	# did we fill the buffer completely?
+	or	$2, $18, $2
+	bne	$2, 2f
+
+	and	$24, 0x80, $2	# no zero next byte
+	bne	$2, 1f
+
+	/* Here there are bytes left in the current word.  Clear one.  */
+	addq	$24, $24, $24	# end-of-count bit <<= 1
+2:	zap	$1, $24, $1
+	stq_u	$1, 0($16)
+	ret
+
+1:	/* Here we must read the next DST word and clear the first byte.  */
+	ldq_u	$1, 8($16)
+	zap	$1, 1, $1
+	stq_u	$1, 8($16)
+
+$zerocount:
+	ret
+
+	.end strncat
+	EXPORT_SYMBOL(strncat)