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, 133 insertions, 0 deletions

diff --git a/crypto/blowfish_generic.c b/crypto/blowfish_generic.c
new file mode 100644
index 000000000..f3c5f9b09
--- /dev/null
+++ b/crypto/blowfish_generic.c
@@ -0,0 +1,133 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Cryptographic API.
+ *
+ * Blowfish Cipher Algorithm, by Bruce Schneier.
+ * http://www.counterpane.com/blowfish.html
+ *
+ * Adapted from Kerneli implementation.
+ *
+ * Copyright (c) Herbert Valerio Riedel <hvr@hvrlab.org>
+ * Copyright (c) Kyle McMartin <kyle@debian.org>
+ * Copyright (c) 2002 James Morris <jmorris@intercode.com.au>
+ */
+
+#include <crypto/algapi.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/unaligned.h>
+#include <linux/types.h>
+#include <crypto/blowfish.h>
+
+/*
+ * Round loop unrolling macros, S is a pointer to a S-Box array
+ * organized in 4 unsigned longs at a row.
+ */
+#define GET32_3(x) (((x) & 0xff))
+#define GET32_2(x) (((x) >> (8)) & (0xff))
+#define GET32_1(x) (((x) >> (16)) & (0xff))
+#define GET32_0(x) (((x) >> (24)) & (0xff))
+
+#define bf_F(x) (((S[GET32_0(x)] + S[256 + GET32_1(x)]) ^ \
+		S[512 + GET32_2(x)]) + S[768 + GET32_3(x)])
+
+#define ROUND(a, b, n) ({ b ^= P[n]; a ^= bf_F(b); })
+
+static void bf_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	struct bf_ctx *ctx = crypto_tfm_ctx(tfm);
+	const u32 *P = ctx->p;
+	const u32 *S = ctx->s;
+	u32 yl = get_unaligned_be32(src);
+	u32 yr = get_unaligned_be32(src + 4);
+
+	ROUND(yr, yl, 0);
+	ROUND(yl, yr, 1);
+	ROUND(yr, yl, 2);
+	ROUND(yl, yr, 3);
+	ROUND(yr, yl, 4);
+	ROUND(yl, yr, 5);
+	ROUND(yr, yl, 6);
+	ROUND(yl, yr, 7);
+	ROUND(yr, yl, 8);
+	ROUND(yl, yr, 9);
+	ROUND(yr, yl, 10);
+	ROUND(yl, yr, 11);
+	ROUND(yr, yl, 12);
+	ROUND(yl, yr, 13);
+	ROUND(yr, yl, 14);
+	ROUND(yl, yr, 15);
+
+	yl ^= P[16];
+	yr ^= P[17];
+
+	put_unaligned_be32(yr, dst);
+	put_unaligned_be32(yl, dst + 4);
+}
+
+static void bf_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	struct bf_ctx *ctx = crypto_tfm_ctx(tfm);
+	const u32 *P = ctx->p;
+	const u32 *S = ctx->s;
+	u32 yl = get_unaligned_be32(src);
+	u32 yr = get_unaligned_be32(src + 4);
+
+	ROUND(yr, yl, 17);
+	ROUND(yl, yr, 16);
+	ROUND(yr, yl, 15);
+	ROUND(yl, yr, 14);
+	ROUND(yr, yl, 13);
+	ROUND(yl, yr, 12);
+	ROUND(yr, yl, 11);
+	ROUND(yl, yr, 10);
+	ROUND(yr, yl, 9);
+	ROUND(yl, yr, 8);
+	ROUND(yr, yl, 7);
+	ROUND(yl, yr, 6);
+	ROUND(yr, yl, 5);
+	ROUND(yl, yr, 4);
+	ROUND(yr, yl, 3);
+	ROUND(yl, yr, 2);
+
+	yl ^= P[1];
+	yr ^= P[0];
+
+	put_unaligned_be32(yr, dst);
+	put_unaligned_be32(yl, dst + 4);
+}
+
+static struct crypto_alg alg = {
+	.cra_name		=	"blowfish",
+	.cra_driver_name	=	"blowfish-generic",
+	.cra_priority		=	100,
+	.cra_flags		=	CRYPTO_ALG_TYPE_CIPHER,
+	.cra_blocksize		=	BF_BLOCK_SIZE,
+	.cra_ctxsize		=	sizeof(struct bf_ctx),
+	.cra_module		=	THIS_MODULE,
+	.cra_u			=	{ .cipher = {
+	.cia_min_keysize	=	BF_MIN_KEY_SIZE,
+	.cia_max_keysize	=	BF_MAX_KEY_SIZE,
+	.cia_setkey		=	blowfish_setkey,
+	.cia_encrypt		=	bf_encrypt,
+	.cia_decrypt		=	bf_decrypt } }
+};
+
+static int __init blowfish_mod_init(void)
+{
+	return crypto_register_alg(&alg);
+}
+
+static void __exit blowfish_mod_fini(void)
+{
+	crypto_unregister_alg(&alg);
+}
+
+module_init(blowfish_mod_init);
+module_exit(blowfish_mod_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Blowfish Cipher Algorithm");
+MODULE_ALIAS_CRYPTO("blowfish");
+MODULE_ALIAS_CRYPTO("blowfish-generic");