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, 204 insertions, 0 deletions

diff --git a/fs/nfs/callback.h b/fs/nfs/callback.h
new file mode 100644
index 000000000..2a721c422
--- /dev/null
+++ b/fs/nfs/callback.h
@@ -0,0 +1,204 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * linux/fs/nfs/callback.h
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback definitions
+ */
+#ifndef __LINUX_FS_NFS_CALLBACK_H
+#define __LINUX_FS_NFS_CALLBACK_H
+#include <linux/sunrpc/svc.h>
+
+#define NFS4_CALLBACK 0x40000000
+#define NFS4_CALLBACK_XDRSIZE 2048
+#define NFS4_CALLBACK_BUFSIZE (1024 + NFS4_CALLBACK_XDRSIZE)
+
+enum nfs4_callback_procnum {
+	CB_NULL = 0,
+	CB_COMPOUND = 1,
+};
+
+struct nfs4_slot;
+struct cb_process_state {
+	struct nfs_client	*clp;
+	struct nfs4_slot	*slot;
+	struct net		*net;
+	u32			minorversion;
+	__be32			drc_status;
+	unsigned int		referring_calls;
+};
+
+struct cb_compound_hdr_arg {
+	unsigned int taglen;
+	const char *tag;
+	unsigned int minorversion;
+	unsigned int cb_ident; /* v4.0 callback identifier */
+	unsigned nops;
+};
+
+struct cb_compound_hdr_res {
+	__be32 *status;
+	unsigned int taglen;
+	const char *tag;
+	__be32 *nops;
+};
+
+struct cb_getattrargs {
+	struct nfs_fh fh;
+	uint32_t bitmap[3];
+};
+
+struct cb_getattrres {
+	__be32 status;
+	uint32_t bitmap[3];
+	uint64_t size;
+	uint64_t change_attr;
+	struct timespec64 atime;
+	struct timespec64 ctime;
+	struct timespec64 mtime;
+};
+
+struct cb_recallargs {
+	struct nfs_fh fh;
+	nfs4_stateid stateid;
+	uint32_t truncate;
+};
+
+struct referring_call {
+	uint32_t			rc_sequenceid;
+	uint32_t			rc_slotid;
+};
+
+struct referring_call_list {
+	struct nfs4_sessionid		rcl_sessionid;
+	uint32_t			rcl_nrefcalls;
+	struct referring_call 		*rcl_refcalls;
+};
+
+struct cb_sequenceargs {
+	struct sockaddr			*csa_addr;
+	struct nfs4_sessionid		csa_sessionid;
+	uint32_t			csa_sequenceid;
+	uint32_t			csa_slotid;
+	uint32_t			csa_highestslotid;
+	uint32_t			csa_cachethis;
+	uint32_t			csa_nrclists;
+	struct referring_call_list	*csa_rclists;
+};
+
+struct cb_sequenceres {
+	__be32				csr_status;
+	struct nfs4_sessionid		csr_sessionid;
+	uint32_t			csr_sequenceid;
+	uint32_t			csr_slotid;
+	uint32_t			csr_highestslotid;
+	uint32_t			csr_target_highestslotid;
+};
+
+extern __be32 nfs4_callback_sequence(void *argp, void *resp,
+				       struct cb_process_state *cps);
+
+#define RCA4_TYPE_MASK_RDATA_DLG	0
+#define RCA4_TYPE_MASK_WDATA_DLG	1
+#define RCA4_TYPE_MASK_DIR_DLG         2
+#define RCA4_TYPE_MASK_FILE_LAYOUT     3
+#define RCA4_TYPE_MASK_BLK_LAYOUT      4
+#define RCA4_TYPE_MASK_OBJ_LAYOUT_MIN  8
+#define RCA4_TYPE_MASK_OBJ_LAYOUT_MAX  9
+#define RCA4_TYPE_MASK_OTHER_LAYOUT_MIN 12
+#define RCA4_TYPE_MASK_OTHER_LAYOUT_MAX 15
+#define PNFS_FF_RCA4_TYPE_MASK_READ 16
+#define PNFS_FF_RCA4_TYPE_MASK_RW 17
+#define RCA4_TYPE_MASK_ALL 0x3f31f
+
+struct cb_recallanyargs {
+	uint32_t	craa_objs_to_keep;
+	uint32_t	craa_type_mask;
+};
+
+extern __be32 nfs4_callback_recallany(void *argp, void *resp,
+					struct cb_process_state *cps);
+
+struct cb_recallslotargs {
+	uint32_t	crsa_target_highest_slotid;
+};
+extern __be32 nfs4_callback_recallslot(void *argp, void *resp,
+					 struct cb_process_state *cps);
+
+struct cb_layoutrecallargs {
+	uint32_t		cbl_recall_type;
+	uint32_t		cbl_layout_type;
+	uint32_t		cbl_layoutchanged;
+	union {
+		struct {
+			struct nfs_fh		cbl_fh;
+			struct pnfs_layout_range cbl_range;
+			nfs4_stateid		cbl_stateid;
+		};
+		struct nfs_fsid		cbl_fsid;
+	};
+};
+
+extern __be32 nfs4_callback_layoutrecall(void *argp, void *resp,
+		struct cb_process_state *cps);
+
+struct cb_devicenotifyitem {
+	uint32_t		cbd_notify_type;
+	uint32_t		cbd_layout_type;
+	struct nfs4_deviceid	cbd_dev_id;
+	uint32_t		cbd_immediate;
+};
+
+struct cb_devicenotifyargs {
+	uint32_t			 ndevs;
+	struct cb_devicenotifyitem	 *devs;
+};
+
+extern __be32 nfs4_callback_devicenotify(void *argp, void *resp,
+		struct cb_process_state *cps);
+
+struct cb_notify_lock_args {
+	struct nfs_fh			cbnl_fh;
+	struct nfs_lowner		cbnl_owner;
+	bool				cbnl_valid;
+};
+
+extern __be32 nfs4_callback_notify_lock(void *argp, void *resp,
+					 struct cb_process_state *cps);
+#ifdef CONFIG_NFS_V4_2
+struct cb_offloadargs {
+	struct nfs_fh		coa_fh;
+	nfs4_stateid		coa_stateid;
+	uint32_t		error;
+	uint64_t		wr_count;
+	struct nfs_writeverf	wr_writeverf;
+};
+
+extern __be32 nfs4_callback_offload(void *args, void *dummy,
+				    struct cb_process_state *cps);
+#endif /* CONFIG_NFS_V4_2 */
+extern int check_gss_callback_principal(struct nfs_client *, struct svc_rqst *);
+extern __be32 nfs4_callback_getattr(void *argp, void *resp,
+				    struct cb_process_state *cps);
+extern __be32 nfs4_callback_recall(void *argp, void *resp,
+				   struct cb_process_state *cps);
+#if IS_ENABLED(CONFIG_NFS_V4)
+extern int nfs_callback_up(u32 minorversion, struct rpc_xprt *xprt);
+extern void nfs_callback_down(int minorversion, struct net *net,
+			      struct rpc_xprt *xprt);
+#endif /* CONFIG_NFS_V4 */
+/*
+ * nfs41: Callbacks are expected to not cause substantial latency,
+ * so we limit their concurrency to 1 by setting up the maximum number
+ * of slots for the backchannel.
+ */
+#define NFS41_BC_MIN_CALLBACKS 1
+#define NFS41_BC_MAX_CALLBACKS 1
+
+#define NFS4_MIN_NR_CALLBACK_THREADS 1
+
+extern unsigned int nfs_callback_set_tcpport;
+extern unsigned short nfs_callback_nr_threads;
+
+#endif /* __LINUX_FS_NFS_CALLBACK_H */