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, 120 insertions, 0 deletions

diff --git a/crypto/dh_helper.c b/crypto/dh_helper.c
new file mode 100644
index 000000000..2d4998793
--- /dev/null
+++ b/crypto/dh_helper.c
@@ -0,0 +1,120 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2016, Intel Corporation
+ * Authors: Salvatore Benedetto <salvatore.benedetto@intel.com>
+ */
+#include <linux/kernel.h>
+#include <linux/export.h>
+#include <linux/err.h>
+#include <linux/string.h>
+#include <crypto/dh.h>
+#include <crypto/kpp.h>
+
+#define DH_KPP_SECRET_MIN_SIZE (sizeof(struct kpp_secret) + 3 * sizeof(int))
+
+static inline u8 *dh_pack_data(u8 *dst, u8 *end, const void *src, size_t size)
+{
+	if (!dst || size > end - dst)
+		return NULL;
+	memcpy(dst, src, size);
+	return dst + size;
+}
+
+static inline const u8 *dh_unpack_data(void *dst, const void *src, size_t size)
+{
+	memcpy(dst, src, size);
+	return src + size;
+}
+
+static inline unsigned int dh_data_size(const struct dh *p)
+{
+	return p->key_size + p->p_size + p->g_size;
+}
+
+unsigned int crypto_dh_key_len(const struct dh *p)
+{
+	return DH_KPP_SECRET_MIN_SIZE + dh_data_size(p);
+}
+EXPORT_SYMBOL_GPL(crypto_dh_key_len);
+
+int crypto_dh_encode_key(char *buf, unsigned int len, const struct dh *params)
+{
+	u8 *ptr = buf;
+	u8 * const end = ptr + len;
+	struct kpp_secret secret = {
+		.type = CRYPTO_KPP_SECRET_TYPE_DH,
+		.len = len
+	};
+
+	if (unlikely(!len))
+		return -EINVAL;
+
+	ptr = dh_pack_data(ptr, end, &secret, sizeof(secret));
+	ptr = dh_pack_data(ptr, end, &params->key_size,
+			   sizeof(params->key_size));
+	ptr = dh_pack_data(ptr, end, &params->p_size, sizeof(params->p_size));
+	ptr = dh_pack_data(ptr, end, &params->g_size, sizeof(params->g_size));
+	ptr = dh_pack_data(ptr, end, params->key, params->key_size);
+	ptr = dh_pack_data(ptr, end, params->p, params->p_size);
+	ptr = dh_pack_data(ptr, end, params->g, params->g_size);
+	if (ptr != end)
+		return -EINVAL;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(crypto_dh_encode_key);
+
+int __crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
+{
+	const u8 *ptr = buf;
+	struct kpp_secret secret;
+
+	if (unlikely(!buf || len < DH_KPP_SECRET_MIN_SIZE))
+		return -EINVAL;
+
+	ptr = dh_unpack_data(&secret, ptr, sizeof(secret));
+	if (secret.type != CRYPTO_KPP_SECRET_TYPE_DH)
+		return -EINVAL;
+
+	ptr = dh_unpack_data(&params->key_size, ptr, sizeof(params->key_size));
+	ptr = dh_unpack_data(&params->p_size, ptr, sizeof(params->p_size));
+	ptr = dh_unpack_data(&params->g_size, ptr, sizeof(params->g_size));
+	if (secret.len != crypto_dh_key_len(params))
+		return -EINVAL;
+
+	/* Don't allocate memory. Set pointers to data within
+	 * the given buffer
+	 */
+	params->key = (void *)ptr;
+	params->p = (void *)(ptr + params->key_size);
+	params->g = (void *)(ptr + params->key_size + params->p_size);
+
+	return 0;
+}
+
+int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
+{
+	int err;
+
+	err = __crypto_dh_decode_key(buf, len, params);
+	if (err)
+		return err;
+
+	/*
+	 * Don't permit the buffer for 'key' or 'g' to be larger than 'p', since
+	 * some drivers assume otherwise.
+	 */
+	if (params->key_size > params->p_size ||
+	    params->g_size > params->p_size)
+		return -EINVAL;
+
+	/*
+	 * Don't permit 'p' to be 0.  It's not a prime number, and it's subject
+	 * to corner cases such as 'mod 0' being undefined or
+	 * crypto_kpp_maxsize() returning 0.
+	 */
+	if (memchr_inv(params->p, 0, params->p_size) == NULL)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(crypto_dh_decode_key);