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, 170 insertions, 0 deletions

diff --git a/include/net/strparser.h b/include/net/strparser.h
new file mode 100644
index 000000000..0ed73e364
--- /dev/null
+++ b/include/net/strparser.h
@@ -0,0 +1,170 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Stream Parser
+ *
+ * Copyright (c) 2016 Tom Herbert <tom@herbertland.com>
+ */
+
+#ifndef __NET_STRPARSER_H_
+#define __NET_STRPARSER_H_
+
+#include <linux/skbuff.h>
+#include <net/sock.h>
+
+#define STRP_STATS_ADD(stat, count) ((stat) += (count))
+#define STRP_STATS_INCR(stat) ((stat)++)
+
+struct strp_stats {
+	unsigned long long msgs;
+	unsigned long long bytes;
+	unsigned int mem_fail;
+	unsigned int need_more_hdr;
+	unsigned int msg_too_big;
+	unsigned int msg_timeouts;
+	unsigned int bad_hdr_len;
+};
+
+struct strp_aggr_stats {
+	unsigned long long msgs;
+	unsigned long long bytes;
+	unsigned int mem_fail;
+	unsigned int need_more_hdr;
+	unsigned int msg_too_big;
+	unsigned int msg_timeouts;
+	unsigned int bad_hdr_len;
+	unsigned int aborts;
+	unsigned int interrupted;
+	unsigned int unrecov_intr;
+};
+
+struct strparser;
+
+/* Callbacks are called with lock held for the attached socket */
+struct strp_callbacks {
+	int (*parse_msg)(struct strparser *strp, struct sk_buff *skb);
+	void (*rcv_msg)(struct strparser *strp, struct sk_buff *skb);
+	int (*read_sock)(struct strparser *strp, read_descriptor_t *desc,
+			 sk_read_actor_t recv_actor);
+	int (*read_sock_done)(struct strparser *strp, int err);
+	void (*abort_parser)(struct strparser *strp, int err);
+	void (*lock)(struct strparser *strp);
+	void (*unlock)(struct strparser *strp);
+};
+
+struct strp_msg {
+	int full_len;
+	int offset;
+};
+
+struct _strp_msg {
+	/* Internal cb structure. struct strp_msg must be first for passing
+	 * to upper layer.
+	 */
+	struct strp_msg strp;
+	int accum_len;
+};
+
+struct sk_skb_cb {
+#define SK_SKB_CB_PRIV_LEN 20
+	unsigned char data[SK_SKB_CB_PRIV_LEN];
+	/* align strp on cache line boundary within skb->cb[] */
+	unsigned char pad[4];
+	struct _strp_msg strp;
+
+	/* strp users' data follows */
+	struct tls_msg {
+		u8 control;
+	} tls;
+	/* temp_reg is a temporary register used for bpf_convert_data_end_access
+	 * when dst_reg == src_reg.
+	 */
+	u64 temp_reg;
+};
+
+static inline struct strp_msg *strp_msg(struct sk_buff *skb)
+{
+	return (struct strp_msg *)((void *)skb->cb +
+		offsetof(struct sk_skb_cb, strp));
+}
+
+/* Structure for an attached lower socket */
+struct strparser {
+	struct sock *sk;
+
+	u32 stopped : 1;
+	u32 paused : 1;
+	u32 aborted : 1;
+	u32 interrupted : 1;
+	u32 unrecov_intr : 1;
+
+	struct sk_buff **skb_nextp;
+	struct sk_buff *skb_head;
+	unsigned int need_bytes;
+	struct delayed_work msg_timer_work;
+	struct work_struct work;
+	struct strp_stats stats;
+	struct strp_callbacks cb;
+};
+
+/* Must be called with lock held for attached socket */
+static inline void strp_pause(struct strparser *strp)
+{
+	strp->paused = 1;
+}
+
+/* May be called without holding lock for attached socket */
+void strp_unpause(struct strparser *strp);
+
+static inline void save_strp_stats(struct strparser *strp,
+				   struct strp_aggr_stats *agg_stats)
+{
+	/* Save psock statistics in the mux when psock is being unattached. */
+
+#define SAVE_PSOCK_STATS(_stat) (agg_stats->_stat +=		\
+				 strp->stats._stat)
+	SAVE_PSOCK_STATS(msgs);
+	SAVE_PSOCK_STATS(bytes);
+	SAVE_PSOCK_STATS(mem_fail);
+	SAVE_PSOCK_STATS(need_more_hdr);
+	SAVE_PSOCK_STATS(msg_too_big);
+	SAVE_PSOCK_STATS(msg_timeouts);
+	SAVE_PSOCK_STATS(bad_hdr_len);
+#undef SAVE_PSOCK_STATS
+
+	if (strp->aborted)
+		agg_stats->aborts++;
+	if (strp->interrupted)
+		agg_stats->interrupted++;
+	if (strp->unrecov_intr)
+		agg_stats->unrecov_intr++;
+}
+
+static inline void aggregate_strp_stats(struct strp_aggr_stats *stats,
+					struct strp_aggr_stats *agg_stats)
+{
+#define SAVE_PSOCK_STATS(_stat) (agg_stats->_stat += stats->_stat)
+	SAVE_PSOCK_STATS(msgs);
+	SAVE_PSOCK_STATS(bytes);
+	SAVE_PSOCK_STATS(mem_fail);
+	SAVE_PSOCK_STATS(need_more_hdr);
+	SAVE_PSOCK_STATS(msg_too_big);
+	SAVE_PSOCK_STATS(msg_timeouts);
+	SAVE_PSOCK_STATS(bad_hdr_len);
+	SAVE_PSOCK_STATS(aborts);
+	SAVE_PSOCK_STATS(interrupted);
+	SAVE_PSOCK_STATS(unrecov_intr);
+#undef SAVE_PSOCK_STATS
+
+}
+
+void strp_done(struct strparser *strp);
+void strp_stop(struct strparser *strp);
+void strp_check_rcv(struct strparser *strp);
+int strp_init(struct strparser *strp, struct sock *sk,
+	      const struct strp_callbacks *cb);
+void strp_data_ready(struct strparser *strp);
+int strp_process(struct strparser *strp, struct sk_buff *orig_skb,
+		 unsigned int orig_offset, size_t orig_len,
+		 size_t max_msg_size, long timeo);
+
+#endif /* __NET_STRPARSER_H_ */