sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

alb_break_llc() decides whether to break LLC preference during active
load balance. It does so by testing that every runnable fair task on the
source rq prefers its LLC:

	env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable

But the two counters cover different sets. nr_pref_llc_running is updated
in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued,
so it follows queued tasks. h_nr_runnable is updated in set_delayed()/
clear_delayed() and drops delay-dequeued tasks.

So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted
in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks,
alb_break_llc() returns false, and active balance is free to pull a task
off its preferred LLC. Active balance only moves runnable tasks, and this
is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB
skips the per-task test in can_migrate_task(). The runnable set is the one
we want.

Fix it on the counter side. A task should be counted in
nr_pref_llc_running exactly while it is both queued on its preferred LLC
(pref_llc_queued) and runnable (!sched_delayed). Define that membership
once in task_pref_llc_runnable(), and adjust the counter only through
pref_llc_running_inc()/pref_llc_running_dec() from the four sites that
change either input: account_llc_enqueue(), account_llc_dequeue(),
set_delayed() and clear_delayed(). Gating every update on the same
predicate keeps the delay, wake and dequeue paths from double-counting
or underflowing; see the comments at those sites for the ordering.

nr_llc_running and sd->llc_counts are not touched and stay on queued
semantics.

Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC")
Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/
Reported-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 76 insertions, 0 deletions

diff --git a/lib/crypto/arc4.c b/lib/crypto/arc4.c
new file mode 100644
index 000000000..4e950e1e6
--- /dev/null
+++ b/lib/crypto/arc4.c
@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Cryptographic API
+ *
+ * ARC4 Cipher Algorithm
+ *
+ * Jon Oberheide <jon@oberheide.org>
+ */
+
+#include <crypto/arc4.h>
+#include <linux/export.h>
+#include <linux/module.h>
+
+int arc4_setkey(struct arc4_ctx *ctx, const u8 *in_key, unsigned int key_len)
+{
+	int i, j = 0, k = 0;
+
+	ctx->x = 1;
+	ctx->y = 0;
+
+	for (i = 0; i < 256; i++)
+		ctx->S[i] = i;
+
+	for (i = 0; i < 256; i++) {
+		u32 a = ctx->S[i];
+
+		j = (j + in_key[k] + a) & 0xff;
+		ctx->S[i] = ctx->S[j];
+		ctx->S[j] = a;
+		if (++k >= key_len)
+			k = 0;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(arc4_setkey);
+
+void arc4_crypt(struct arc4_ctx *ctx, u8 *out, const u8 *in, unsigned int len)
+{
+	u32 *const S = ctx->S;
+	u32 x, y, a, b;
+	u32 ty, ta, tb;
+
+	if (len == 0)
+		return;
+
+	x = ctx->x;
+	y = ctx->y;
+
+	a = S[x];
+	y = (y + a) & 0xff;
+	b = S[y];
+
+	do {
+		S[y] = a;
+		a = (a + b) & 0xff;
+		S[x] = b;
+		x = (x + 1) & 0xff;
+		ta = S[x];
+		ty = (y + ta) & 0xff;
+		tb = S[ty];
+		*out++ = *in++ ^ S[a];
+		if (--len == 0)
+			break;
+		y = ty;
+		a = ta;
+		b = tb;
+	} while (true);
+
+	ctx->x = x;
+	ctx->y = y;
+}
+EXPORT_SYMBOL(arc4_crypt);
+
+MODULE_DESCRIPTION("ARC4 Cipher Algorithm");
+MODULE_LICENSE("GPL");