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, 254 insertions, 0 deletions

diff --git a/crypto/algboss.c b/crypto/algboss.c
new file mode 100644
index 000000000..8b936cae6
--- /dev/null
+++ b/crypto/algboss.c
@@ -0,0 +1,254 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Create default crypto algorithm instances.
+ *
+ * Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
+ */
+
+#include <crypto/internal/aead.h>
+#include <linux/completion.h>
+#include <linux/ctype.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/kthread.h>
+#include <linux/module.h>
+#include <linux/notifier.h>
+#include <linux/rtnetlink.h>
+#include <linux/sched/signal.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+
+#include "internal.h"
+
+struct cryptomgr_param {
+	struct rtattr *tb[CRYPTO_MAX_ATTRS + 2];
+
+	struct {
+		struct rtattr attr;
+		struct crypto_attr_type data;
+	} type;
+
+	struct {
+		struct rtattr attr;
+		struct crypto_attr_alg data;
+	} attrs[CRYPTO_MAX_ATTRS];
+
+	char template[CRYPTO_MAX_ALG_NAME];
+
+	struct crypto_larval *larval;
+
+	u32 otype;
+	u32 omask;
+};
+
+struct crypto_test_param {
+	char driver[CRYPTO_MAX_ALG_NAME];
+	char alg[CRYPTO_MAX_ALG_NAME];
+	u32 type;
+};
+
+static int cryptomgr_probe(void *data)
+{
+	struct cryptomgr_param *param = data;
+	struct crypto_template *tmpl;
+	int err = -ENOENT;
+
+	tmpl = crypto_lookup_template(param->template);
+	if (!tmpl)
+		goto out;
+
+	do {
+		err = tmpl->create(tmpl, param->tb);
+	} while (err == -EAGAIN && !signal_pending(current));
+
+	crypto_tmpl_put(tmpl);
+
+out:
+	param->larval->adult = ERR_PTR(err);
+	param->larval->alg.cra_flags |= CRYPTO_ALG_DEAD;
+	complete_all(&param->larval->completion);
+	crypto_alg_put(&param->larval->alg);
+	kfree(param);
+	module_put_and_kthread_exit(0);
+}
+
+static int cryptomgr_schedule_probe(struct crypto_larval *larval)
+{
+	struct task_struct *thread;
+	struct cryptomgr_param *param;
+	const char *name = larval->alg.cra_name;
+	const char *p;
+	unsigned int len;
+	int i;
+
+	if (!try_module_get(THIS_MODULE))
+		goto err;
+
+	param = kzalloc_obj(*param);
+	if (!param)
+		goto err_put_module;
+
+	for (p = name; isalnum(*p) || *p == '-' || *p == '_'; p++)
+		;
+
+	len = p - name;
+	if (!len || *p != '(')
+		goto err_free_param;
+
+	memcpy(param->template, name, len);
+
+	i = 0;
+	for (;;) {
+		name = ++p;
+
+		for (; isalnum(*p) || *p == '-' || *p == '_'; p++)
+			;
+
+		if (*p == '(') {
+			int recursion = 0;
+
+			for (;;) {
+				if (!*++p)
+					goto err_free_param;
+				if (*p == '(')
+					recursion++;
+				else if (*p == ')' && !recursion--)
+					break;
+			}
+
+			p++;
+		}
+
+		len = p - name;
+		if (!len)
+			goto err_free_param;
+
+		param->attrs[i].attr.rta_len = sizeof(param->attrs[i]);
+		param->attrs[i].attr.rta_type = CRYPTOA_ALG;
+		memcpy(param->attrs[i].data.name, name, len);
+
+		param->tb[i + 1] = &param->attrs[i].attr;
+		i++;
+
+		if (i >= CRYPTO_MAX_ATTRS)
+			goto err_free_param;
+
+		if (*p == ')')
+			break;
+
+		if (*p != ',')
+			goto err_free_param;
+	}
+
+	param->tb[i + 1] = NULL;
+
+	param->type.attr.rta_len = sizeof(param->type);
+	param->type.attr.rta_type = CRYPTOA_TYPE;
+	param->type.data.type = larval->alg.cra_flags & ~CRYPTO_ALG_TESTED;
+	param->type.data.mask = larval->mask & ~CRYPTO_ALG_TESTED;
+	param->tb[0] = &param->type.attr;
+
+	param->otype = larval->alg.cra_flags;
+	param->omask = larval->mask;
+
+	crypto_alg_get(&larval->alg);
+	param->larval = larval;
+
+	thread = kthread_run(cryptomgr_probe, param, "cryptomgr_probe");
+	if (IS_ERR(thread))
+		goto err_put_larval;
+
+	return NOTIFY_STOP;
+
+err_put_larval:
+	crypto_alg_put(&larval->alg);
+err_free_param:
+	kfree(param);
+err_put_module:
+	module_put(THIS_MODULE);
+err:
+	return NOTIFY_OK;
+}
+
+static int cryptomgr_test(void *data)
+{
+	struct crypto_test_param *param = data;
+	u32 type = param->type;
+	int err;
+
+	err = alg_test(param->driver, param->alg, type, CRYPTO_ALG_TESTED);
+
+	crypto_alg_tested(param->driver, err);
+
+	kfree(param);
+	module_put_and_kthread_exit(0);
+}
+
+static int cryptomgr_schedule_test(struct crypto_alg *alg)
+{
+	struct task_struct *thread;
+	struct crypto_test_param *param;
+
+	if (!IS_ENABLED(CONFIG_CRYPTO_SELFTESTS))
+		return NOTIFY_DONE;
+
+	if (!try_module_get(THIS_MODULE))
+		goto err;
+
+	param = kzalloc_obj(*param);
+	if (!param)
+		goto err_put_module;
+
+	memcpy(param->driver, alg->cra_driver_name, sizeof(param->driver));
+	memcpy(param->alg, alg->cra_name, sizeof(param->alg));
+	param->type = alg->cra_flags;
+
+	thread = kthread_run(cryptomgr_test, param, "cryptomgr_test");
+	if (IS_ERR(thread))
+		goto err_free_param;
+
+	return NOTIFY_STOP;
+
+err_free_param:
+	kfree(param);
+err_put_module:
+	module_put(THIS_MODULE);
+err:
+	return NOTIFY_OK;
+}
+
+static int cryptomgr_notify(struct notifier_block *this, unsigned long msg,
+			    void *data)
+{
+	switch (msg) {
+	case CRYPTO_MSG_ALG_REQUEST:
+		return cryptomgr_schedule_probe(data);
+	case CRYPTO_MSG_ALG_REGISTER:
+		return cryptomgr_schedule_test(data);
+	case CRYPTO_MSG_ALG_LOADED:
+		break;
+	}
+
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block cryptomgr_notifier = {
+	.notifier_call = cryptomgr_notify,
+};
+
+static int __init cryptomgr_init(void)
+{
+	return crypto_register_notifier(&cryptomgr_notifier);
+}
+
+static void __exit cryptomgr_exit(void)
+{
+	int err = crypto_unregister_notifier(&cryptomgr_notifier);
+	BUG_ON(err);
+}
+
+module_init(cryptomgr_init);
+module_exit(cryptomgr_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Crypto Algorithm Manager");