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, 139 insertions, 0 deletions

diff --git a/tools/perf/util/jitdump.h b/tools/perf/util/jitdump.h
new file mode 100644
index 000000000..f57bfebb2
--- /dev/null
+++ b/tools/perf/util/jitdump.h
@@ -0,0 +1,139 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * jitdump.h: jitted code info encapsulation file format
+ *
+ * Adapted from OProfile GPLv2 support jidump.h:
+ * Copyright 2007 OProfile authors
+ * Jens Wilke
+ * Daniel Hansel
+ * Copyright IBM Corporation 2007
+ */
+#ifndef JITDUMP_H
+#define JITDUMP_H
+
+#include <stdint.h>
+#include <string.h>
+
+/* JiTD */
+#define JITHEADER_MAGIC		0x4A695444
+#define JITHEADER_MAGIC_SW	0x4454694A
+
+#define PADDING_8ALIGNED(x) ((((x) + 7) & 7) ^ 7)
+#define ALIGN_8(x) (((x) + 7) & (~7))
+
+#define JITHEADER_VERSION 1
+
+enum jitdump_flags_bits {
+	JITDUMP_FLAGS_ARCH_TIMESTAMP_BIT,
+	JITDUMP_FLAGS_MAX_BIT,
+};
+
+#define JITDUMP_FLAGS_ARCH_TIMESTAMP	(1ULL << JITDUMP_FLAGS_ARCH_TIMESTAMP_BIT)
+
+#define JITDUMP_FLAGS_RESERVED (JITDUMP_FLAGS_MAX_BIT < 64 ? \
+				(~((1ULL << JITDUMP_FLAGS_MAX_BIT) - 1)) : 0)
+
+struct jitheader {
+	uint32_t magic;		/* characters "jItD" */
+	uint32_t version;	/* header version */
+	uint32_t total_size;	/* total size of header */
+	uint32_t elf_mach;	/* elf mach target */
+	uint32_t pad1;		/* reserved */
+	uint32_t pid;		/* JIT process id */
+	uint64_t timestamp;	/* timestamp */
+	uint64_t flags;		/* flags */
+};
+
+enum jit_record_type {
+	JIT_CODE_LOAD		= 0,
+        JIT_CODE_MOVE           = 1,
+	JIT_CODE_DEBUG_INFO	= 2,
+	JIT_CODE_CLOSE		= 3,
+	JIT_CODE_UNWINDING_INFO	= 4,
+
+	JIT_CODE_MAX,
+};
+
+/* record prefix (mandatory in each record) */
+struct jr_prefix {
+	uint32_t id;
+	uint32_t total_size;
+	uint64_t timestamp;
+};
+
+struct jr_code_load {
+	struct jr_prefix p;
+
+	uint32_t pid;
+	uint32_t tid;
+	uint64_t vma;
+	uint64_t code_addr;
+	uint64_t code_size;
+	uint64_t code_index;
+};
+
+struct jr_code_close {
+	struct jr_prefix p;
+};
+
+struct jr_code_move {
+	struct jr_prefix p;
+
+	uint32_t pid;
+	uint32_t tid;
+	uint64_t vma;
+	uint64_t old_code_addr;
+	uint64_t new_code_addr;
+	uint64_t code_size;
+	uint64_t code_index;
+};
+
+struct debug_entry {
+	uint64_t addr;
+	int lineno;	    /* source line number starting at 1 */
+	int discrim;	    /* column discriminator, 0 is default */
+	const char name[]; /* null terminated filename, \xff\0 if same as previous entry */
+};
+
+struct jr_code_debug_info {
+	struct jr_prefix p;
+
+	uint64_t code_addr;
+	uint64_t nr_entry;
+	struct debug_entry entries[];
+};
+
+struct jr_code_unwinding_info {
+	struct jr_prefix p;
+
+	uint64_t unwinding_size;
+	uint64_t eh_frame_hdr_size;
+	uint64_t mapped_size;
+	const char unwinding_data[];
+};
+
+union jr_entry {
+        struct jr_code_debug_info info;
+        struct jr_code_close close;
+        struct jr_code_load load;
+        struct jr_code_move move;
+        struct jr_prefix prefix;
+        struct jr_code_unwinding_info unwinding;
+};
+
+static inline struct debug_entry *
+debug_entry_next(struct debug_entry *ent)
+{
+	void *a = ent + 1;
+	size_t l = strlen(ent->name) + 1;
+	return a + l;
+}
+
+static inline char *
+debug_entry_file(struct debug_entry *ent)
+{
+	void *a = ent + 1;
+	return a;
+}
+
+#endif /* !JITDUMP_H */