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, 62 insertions, 0 deletions

diff --git a/include/crypto/mldsa.h b/include/crypto/mldsa.h
new file mode 100644
index 000000000..3ef267678
--- /dev/null
+++ b/include/crypto/mldsa.h
@@ -0,0 +1,62 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Support for verifying ML-DSA signatures
+ *
+ * Copyright 2025 Google LLC
+ */
+#ifndef _CRYPTO_MLDSA_H
+#define _CRYPTO_MLDSA_H
+
+#include <linux/types.h>
+
+/* Identifier for an ML-DSA parameter set */
+enum mldsa_alg {
+	MLDSA44, /* ML-DSA-44 */
+	MLDSA65, /* ML-DSA-65 */
+	MLDSA87, /* ML-DSA-87 */
+};
+
+/* Lengths of ML-DSA public keys and signatures in bytes */
+#define MLDSA44_PUBLIC_KEY_SIZE 1312
+#define MLDSA65_PUBLIC_KEY_SIZE 1952
+#define MLDSA87_PUBLIC_KEY_SIZE 2592
+#define MLDSA44_SIGNATURE_SIZE 2420
+#define MLDSA65_SIGNATURE_SIZE 3309
+#define MLDSA87_SIGNATURE_SIZE 4627
+
+/**
+ * mldsa_verify() - Verify an ML-DSA signature
+ * @alg: The ML-DSA parameter set to use
+ * @sig: The signature
+ * @sig_len: Length of the signature in bytes.  Should match the
+ *	     MLDSA*_SIGNATURE_SIZE constant associated with @alg,
+ *	     otherwise -EBADMSG will be returned.
+ * @msg: The message
+ * @msg_len: Length of the message in bytes
+ * @pk: The public key
+ * @pk_len: Length of the public key in bytes.  Should match the
+ *	    MLDSA*_PUBLIC_KEY_SIZE constant associated with @alg,
+ *	    otherwise -EBADMSG will be returned.
+ *
+ * This verifies a signature using pure ML-DSA with the specified parameter set.
+ * The context string is assumed to be empty.  This corresponds to FIPS 204
+ * Algorithm 3 "ML-DSA.Verify" with the ctx parameter set to the empty string
+ * and the lengths of the signature and key given explicitly by the caller.
+ *
+ * Context: Might sleep
+ *
+ * Return:
+ * * 0 if the signature is valid
+ * * -EBADMSG if the signature and/or public key is malformed
+ * * -EKEYREJECTED if the signature is invalid but otherwise well-formed
+ * * -ENOMEM if out of memory so the validity of the signature is unknown
+ */
+int mldsa_verify(enum mldsa_alg alg, const u8 *sig, size_t sig_len,
+		 const u8 *msg, size_t msg_len, const u8 *pk, size_t pk_len);
+
+#if IS_ENABLED(CONFIG_CRYPTO_LIB_MLDSA_KUNIT_TEST)
+/* Internal function, exposed only for unit testing */
+s32 mldsa_use_hint(u8 h, s32 r, s32 gamma2);
+#endif
+
+#endif /* _CRYPTO_MLDSA_H */