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, 127 insertions, 0 deletions

diff --git a/include/crypto/internal/des.h b/include/crypto/internal/des.h
new file mode 100644
index 000000000..723fe5bf1
--- /dev/null
+++ b/include/crypto/internal/des.h
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * DES & Triple DES EDE key verification helpers
+ */
+
+#ifndef __CRYPTO_INTERNAL_DES_H
+#define __CRYPTO_INTERNAL_DES_H
+
+#include <linux/crypto.h>
+#include <linux/fips.h>
+#include <crypto/des.h>
+#include <crypto/aead.h>
+#include <crypto/skcipher.h>
+
+/**
+ * crypto_des_verify_key - Check whether a DES key is weak
+ * @tfm: the crypto algo
+ * @key: the key buffer
+ *
+ * Returns -EINVAL if the key is weak and the crypto TFM does not permit weak
+ * keys. Otherwise, 0 is returned.
+ *
+ * It is the job of the caller to ensure that the size of the key equals
+ * DES_KEY_SIZE.
+ */
+static inline int crypto_des_verify_key(struct crypto_tfm *tfm, const u8 *key)
+{
+	struct des_ctx tmp;
+	int err;
+
+	err = des_expand_key(&tmp, key, DES_KEY_SIZE);
+	if (err == -ENOKEY) {
+		if (crypto_tfm_get_flags(tfm) & CRYPTO_TFM_REQ_FORBID_WEAK_KEYS)
+			err = -EINVAL;
+		else
+			err = 0;
+	}
+	memzero_explicit(&tmp, sizeof(tmp));
+	return err;
+}
+
+/*
+ * RFC2451:
+ *
+ *   For DES-EDE3, there is no known need to reject weak or
+ *   complementation keys.  Any weakness is obviated by the use of
+ *   multiple keys.
+ *
+ *   However, if the first two or last two independent 64-bit keys are
+ *   equal (k1 == k2 or k2 == k3), then the DES3 operation is simply the
+ *   same as DES.  Implementers MUST reject keys that exhibit this
+ *   property.
+ *
+ */
+static inline int des3_ede_verify_key(const u8 *key, unsigned int key_len,
+				      bool check_weak)
+{
+	int ret = fips_enabled ? -EINVAL : -ENOKEY;
+	u32 K[6];
+
+	memcpy(K, key, DES3_EDE_KEY_SIZE);
+
+	if ((!((K[0] ^ K[2]) | (K[1] ^ K[3])) ||
+	     !((K[2] ^ K[4]) | (K[3] ^ K[5]))) &&
+	    (fips_enabled || check_weak))
+		goto bad;
+
+	if ((!((K[0] ^ K[4]) | (K[1] ^ K[5]))) && fips_enabled)
+		goto bad;
+
+	ret = 0;
+bad:
+	memzero_explicit(K, DES3_EDE_KEY_SIZE);
+
+	return ret;
+}
+
+/**
+ * crypto_des3_ede_verify_key - Check whether a DES3-EDE key is weak
+ * @tfm: the crypto algo
+ * @key: the key buffer
+ *
+ * Returns -EINVAL if the key is weak and the crypto TFM does not permit weak
+ * keys or when running in FIPS mode. Otherwise, 0 is returned. Note that some
+ * keys are rejected in FIPS mode even if weak keys are permitted by the TFM
+ * flags.
+ *
+ * It is the job of the caller to ensure that the size of the key equals
+ * DES3_EDE_KEY_SIZE.
+ */
+static inline int crypto_des3_ede_verify_key(struct crypto_tfm *tfm,
+					     const u8 *key)
+{
+	return des3_ede_verify_key(key, DES3_EDE_KEY_SIZE,
+				   crypto_tfm_get_flags(tfm) &
+				   CRYPTO_TFM_REQ_FORBID_WEAK_KEYS);
+}
+
+static inline int verify_skcipher_des_key(struct crypto_skcipher *tfm,
+					  const u8 *key)
+{
+	return crypto_des_verify_key(crypto_skcipher_tfm(tfm), key);
+}
+
+static inline int verify_skcipher_des3_key(struct crypto_skcipher *tfm,
+					   const u8 *key)
+{
+	return crypto_des3_ede_verify_key(crypto_skcipher_tfm(tfm), key);
+}
+
+static inline int verify_aead_des_key(struct crypto_aead *tfm, const u8 *key,
+				      int keylen)
+{
+	if (keylen != DES_KEY_SIZE)
+		return -EINVAL;
+	return crypto_des_verify_key(crypto_aead_tfm(tfm), key);
+}
+
+static inline int verify_aead_des3_key(struct crypto_aead *tfm, const u8 *key,
+				       int keylen)
+{
+	if (keylen != DES3_EDE_KEY_SIZE)
+		return -EINVAL;
+	return crypto_des3_ede_verify_key(crypto_aead_tfm(tfm), key);
+}
+
+#endif /* __CRYPTO_INTERNAL_DES_H */