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, 198 insertions, 0 deletions

diff --git a/include/net/flow.h b/include/net/flow.h
new file mode 100644
index 000000000..ae9481c40
--- /dev/null
+++ b/include/net/flow.h
@@ -0,0 +1,198 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ *
+ *	Generic internet FLOW.
+ *
+ */
+
+#ifndef _NET_FLOW_H
+#define _NET_FLOW_H
+
+#include <linux/in6.h>
+#include <linux/atomic.h>
+#include <linux/container_of.h>
+#include <linux/uidgid.h>
+#include <net/inet_dscp.h>
+
+struct flow_keys;
+
+/*
+ * ifindex generation is per-net namespace, and loopback is
+ * always the 1st device in ns (see net_dev_init), thus any
+ * loopback device should get ifindex 1
+ */
+
+#define LOOPBACK_IFINDEX	1
+
+struct flowi_tunnel {
+	__be64			tun_id;
+};
+
+struct flowi_common {
+	int	flowic_oif;
+	int	flowic_iif;
+	int     flowic_l3mdev;
+	__u32	flowic_mark;
+	dscp_t	flowic_dscp;
+	__u8	flowic_scope;
+	__u8	flowic_proto;
+	__u8	flowic_flags;
+#define FLOWI_FLAG_ANYSRC		0x01
+#define FLOWI_FLAG_KNOWN_NH		0x02
+#define FLOWI_FLAG_L3MDEV_OIF		0x04
+#define FLOWI_FLAG_ANY_SPORT		0x08
+	__u32	flowic_secid;
+	kuid_t  flowic_uid;
+	__u32		flowic_multipath_hash;
+	struct flowi_tunnel flowic_tun_key;
+};
+
+union flowi_uli {
+	struct {
+		__be16	dport;
+		__be16	sport;
+	} ports;
+
+	struct {
+		__u8	type;
+		__u8	code;
+	} icmpt;
+
+	__be32		gre_key;
+
+	struct {
+		__u8	type;
+	} mht;
+};
+
+struct flowi4 {
+	struct flowi_common	__fl_common;
+#define flowi4_oif		__fl_common.flowic_oif
+#define flowi4_iif		__fl_common.flowic_iif
+#define flowi4_l3mdev		__fl_common.flowic_l3mdev
+#define flowi4_mark		__fl_common.flowic_mark
+#define flowi4_dscp		__fl_common.flowic_dscp
+#define flowi4_scope		__fl_common.flowic_scope
+#define flowi4_proto		__fl_common.flowic_proto
+#define flowi4_flags		__fl_common.flowic_flags
+#define flowi4_secid		__fl_common.flowic_secid
+#define flowi4_tun_key		__fl_common.flowic_tun_key
+#define flowi4_uid		__fl_common.flowic_uid
+#define flowi4_multipath_hash	__fl_common.flowic_multipath_hash
+
+	/* (saddr,daddr) must be grouped, same order as in IP header */
+	__be32			saddr;
+	__be32			daddr;
+
+	union flowi_uli		uli;
+#define fl4_sport		uli.ports.sport
+#define fl4_dport		uli.ports.dport
+#define fl4_icmp_type		uli.icmpt.type
+#define fl4_icmp_code		uli.icmpt.code
+#define fl4_mh_type		uli.mht.type
+#define fl4_gre_key		uli.gre_key
+} __attribute__((__aligned__(BITS_PER_LONG/8)));
+
+static inline void flowi4_init_output(struct flowi4 *fl4, int oif,
+				      __u32 mark, __u8 tos, __u8 scope,
+				      __u8 proto, __u8 flags,
+				      __be32 daddr, __be32 saddr,
+				      __be16 dport, __be16 sport,
+				      kuid_t uid)
+{
+	fl4->flowi4_oif = oif;
+	fl4->flowi4_iif = LOOPBACK_IFINDEX;
+	fl4->flowi4_l3mdev = 0;
+	fl4->flowi4_mark = mark;
+	fl4->flowi4_dscp = inet_dsfield_to_dscp(tos);
+	fl4->flowi4_scope = scope;
+	fl4->flowi4_proto = proto;
+	fl4->flowi4_flags = flags;
+	fl4->flowi4_secid = 0;
+	fl4->flowi4_tun_key.tun_id = 0;
+	fl4->flowi4_uid = uid;
+	fl4->daddr = daddr;
+	fl4->saddr = saddr;
+	fl4->fl4_dport = dport;
+	fl4->fl4_sport = sport;
+	fl4->flowi4_multipath_hash = 0;
+}
+
+/* Reset some input parameters after previous lookup */
+static inline void flowi4_update_output(struct flowi4 *fl4, int oif,
+					__be32 daddr, __be32 saddr)
+{
+	fl4->flowi4_oif = oif;
+	fl4->daddr = daddr;
+	fl4->saddr = saddr;
+}
+
+
+struct flowi6 {
+	struct flowi_common	__fl_common;
+#define flowi6_oif		__fl_common.flowic_oif
+#define flowi6_iif		__fl_common.flowic_iif
+#define flowi6_l3mdev		__fl_common.flowic_l3mdev
+#define flowi6_mark		__fl_common.flowic_mark
+#define flowi6_scope		__fl_common.flowic_scope
+#define flowi6_proto		__fl_common.flowic_proto
+#define flowi6_flags		__fl_common.flowic_flags
+#define flowi6_secid		__fl_common.flowic_secid
+#define flowi6_tun_key		__fl_common.flowic_tun_key
+#define flowi6_uid		__fl_common.flowic_uid
+	struct in6_addr		daddr;
+	struct in6_addr		saddr;
+	/* Note: flowi6_dscp is encoded in flowlabel, too. */
+	__be32			flowlabel;
+	union flowi_uli		uli;
+#define fl6_sport		uli.ports.sport
+#define fl6_dport		uli.ports.dport
+#define fl6_icmp_type		uli.icmpt.type
+#define fl6_icmp_code		uli.icmpt.code
+#define fl6_mh_type		uli.mht.type
+#define fl6_gre_key		uli.gre_key
+	__u32			mp_hash;
+} __attribute__((__aligned__(BITS_PER_LONG/8)));
+
+struct flowi {
+	union {
+		struct flowi_common	__fl_common;
+		struct flowi4		ip4;
+		struct flowi6		ip6;
+	} u;
+#define flowi_oif	u.__fl_common.flowic_oif
+#define flowi_iif	u.__fl_common.flowic_iif
+#define flowi_l3mdev	u.__fl_common.flowic_l3mdev
+#define flowi_mark	u.__fl_common.flowic_mark
+#define flowi_dscp	u.__fl_common.flowic_dscp
+#define flowi_scope	u.__fl_common.flowic_scope
+#define flowi_proto	u.__fl_common.flowic_proto
+#define flowi_flags	u.__fl_common.flowic_flags
+#define flowi_secid	u.__fl_common.flowic_secid
+#define flowi_tun_key	u.__fl_common.flowic_tun_key
+#define flowi_uid	u.__fl_common.flowic_uid
+} __attribute__((__aligned__(BITS_PER_LONG/8)));
+
+static inline struct flowi *flowi4_to_flowi(struct flowi4 *fl4)
+{
+	return container_of(fl4, struct flowi, u.ip4);
+}
+
+static inline struct flowi_common *flowi4_to_flowi_common(struct flowi4 *fl4)
+{
+	return &(fl4->__fl_common);
+}
+
+static inline struct flowi *flowi6_to_flowi(struct flowi6 *fl6)
+{
+	return container_of(fl6, struct flowi, u.ip6);
+}
+
+static inline struct flowi_common *flowi6_to_flowi_common(struct flowi6 *fl6)
+{
+	return &(fl6->__fl_common);
+}
+
+__u32 __get_hash_from_flowi6(const struct flowi6 *fl6, struct flow_keys *keys);
+
+#endif