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, 94 insertions, 0 deletions

diff --git a/net/ceph/msgpool.c b/net/ceph/msgpool.c
new file mode 100644
index 000000000..e3ecb80cd
--- /dev/null
+++ b/net/ceph/msgpool.c
@@ -0,0 +1,94 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/err.h>
+#include <linux/sched.h>
+#include <linux/types.h>
+#include <linux/vmalloc.h>
+
+#include <linux/ceph/messenger.h>
+#include <linux/ceph/msgpool.h>
+
+static void *msgpool_alloc(gfp_t gfp_mask, void *arg)
+{
+	struct ceph_msgpool *pool = arg;
+	struct ceph_msg *msg;
+
+	msg = ceph_msg_new2(pool->type, pool->front_len, pool->max_data_items,
+			    gfp_mask, true);
+	if (!msg) {
+		dout("msgpool_alloc %s failed\n", pool->name);
+	} else {
+		dout("msgpool_alloc %s %p\n", pool->name, msg);
+		msg->pool = pool;
+	}
+	return msg;
+}
+
+static void msgpool_free(void *element, void *arg)
+{
+	struct ceph_msgpool *pool = arg;
+	struct ceph_msg *msg = element;
+
+	dout("msgpool_release %s %p\n", pool->name, msg);
+	msg->pool = NULL;
+	ceph_msg_put(msg);
+}
+
+int ceph_msgpool_init(struct ceph_msgpool *pool, int type,
+		      int front_len, int max_data_items, int size,
+		      const char *name)
+{
+	dout("msgpool %s init\n", name);
+	pool->type = type;
+	pool->front_len = front_len;
+	pool->max_data_items = max_data_items;
+	pool->pool = mempool_create(size, msgpool_alloc, msgpool_free, pool);
+	if (!pool->pool)
+		return -ENOMEM;
+	pool->name = name;
+	return 0;
+}
+
+void ceph_msgpool_destroy(struct ceph_msgpool *pool)
+{
+	dout("msgpool %s destroy\n", pool->name);
+	mempool_destroy(pool->pool);
+}
+
+struct ceph_msg *ceph_msgpool_get(struct ceph_msgpool *pool, int front_len,
+				  int max_data_items)
+{
+	struct ceph_msg *msg;
+
+	if (front_len > pool->front_len ||
+	    max_data_items > pool->max_data_items) {
+		pr_warn_ratelimited("%s need %d/%d, pool %s has %d/%d\n",
+		    __func__, front_len, max_data_items, pool->name,
+		    pool->front_len, pool->max_data_items);
+		WARN_ON_ONCE(1);
+
+		/* try to alloc a fresh message */
+		return ceph_msg_new2(pool->type, front_len, max_data_items,
+				     GFP_NOFS, false);
+	}
+
+	msg = mempool_alloc(pool->pool, GFP_NOFS);
+	dout("msgpool_get %s %p\n", pool->name, msg);
+	return msg;
+}
+
+void ceph_msgpool_put(struct ceph_msgpool *pool, struct ceph_msg *msg)
+{
+	dout("msgpool_put %s %p\n", pool->name, msg);
+
+	/* reset msg front_len; user may have changed it */
+	msg->front.iov_len = pool->front_len;
+	msg->hdr.front_len = cpu_to_le32(pool->front_len);
+
+	msg->data_length = 0;
+	msg->num_data_items = 0;
+
+	kref_init(&msg->kref);  /* retake single ref */
+	mempool_free(msg, pool->pool);
+}