ata: libata-scsi: fix ata_dsm_trim_pages() kernel-docgrafted

make htmldocs fails with:

  Documentation/driver-api/libata:604: ./drivers/ata/libata-scsi.c:2301:
  ERROR: Unexpected indentation. [docutils]

The Return: section of ata_dsm_trim_pages() ends its first sentence with
a colon and continues with an indented bullet list. reStructuredText
requires a blank line before an indented block, so docutils chokes on
the list.

Simply adding the missing blank line does not work either: Return: is a
kernel-doc "special section", which is terminated by the first blank
line, so the bullet list would end up in the Description section,
detached from the sentence introducing it.

Spell the two bounds out as prose instead, so that the Return: section
stays self-contained. Documentation-only change.

Fixes: e64e6b5dc867 ("ata: libata-scsi: scale DSM TRIM payload by MAX PAGES PER DSM COMMAND")
Reported-by: Thomas Huth <thuth@redhat.com>
Closes: https://lore.kernel.org/linux-ide/b0a0b8e8-cc4f-4c7b-8bc5-fee0712d405a@redhat.com/
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Link: https://lore.kernel.org/r/20260916130031.29990-2-cassel@kernel.org
Signed-off-by: Niklas Cassel <cassel@kernel.org>

20 files changed, 2050 insertions, 0 deletions

diff --git a/tools/lib/perf/include/internal/cpumap.h b/tools/lib/perf/include/internal/cpumap.h
new file mode 100644
index 000000000..c19678188
--- /dev/null
+++ b/tools/lib/perf/include/internal/cpumap.h
@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_CPUMAP_H
+#define __LIBPERF_INTERNAL_CPUMAP_H
+
+#include <linux/refcount.h>
+#include <perf/cpumap.h>
+#include <internal/rc_check.h>
+
+/**
+ * A sized, reference counted, sorted array of integers representing CPU
+ * numbers. This is commonly used to capture which CPUs a PMU is associated
+ * with. The indices into the cpumap are frequently used as they avoid having
+ * gaps if CPU numbers were used. For events associated with a pid, rather than
+ * a CPU, a single dummy map with an entry of -1 is used.
+ */
+DECLARE_RC_STRUCT(perf_cpu_map) {
+	refcount_t	refcnt;
+	/** Length of the map array. */
+	unsigned int	nr;
+	/** The CPU values. */
+	struct perf_cpu	map[];
+};
+
+struct perf_cpu_map *perf_cpu_map__alloc(unsigned int nr_cpus);
+int perf_cpu_map__idx(const struct perf_cpu_map *cpus, struct perf_cpu cpu);
+bool perf_cpu_map__is_subset(const struct perf_cpu_map *a, const struct perf_cpu_map *b);
+
+void perf_cpu_map__set_nr(struct perf_cpu_map *map, unsigned int nr_cpus);
+
+static inline refcount_t *perf_cpu_map__refcnt(struct perf_cpu_map *map)
+{
+	return &RC_CHK_ACCESS(map)->refcnt;
+}
+#endif /* __LIBPERF_INTERNAL_CPUMAP_H */
diff --git a/tools/lib/perf/include/internal/evlist.h b/tools/lib/perf/include/internal/evlist.h
new file mode 100644
index 000000000..f43bdb9b6
--- /dev/null
+++ b/tools/lib/perf/include/internal/evlist.h
@@ -0,0 +1,140 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_EVLIST_H
+#define __LIBPERF_INTERNAL_EVLIST_H
+
+#include <linux/list.h>
+#include <api/fd/array.h>
+#include <internal/cpumap.h>
+#include <internal/evsel.h>
+
+#define PERF_EVLIST__HLIST_BITS 8
+#define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS)
+
+struct perf_cpu_map;
+struct perf_thread_map;
+struct perf_mmap_param;
+
+struct perf_evlist {
+	struct list_head	 entries;
+	int			 nr_entries;
+	bool			 has_user_cpus;
+	bool			 needs_map_propagation;
+	/**
+	 * The cpus passed from the command line or all online CPUs by
+	 * default.
+	 */
+	struct perf_cpu_map	*user_requested_cpus;
+	/** The union of all evsel cpu maps. */
+	struct perf_cpu_map	*all_cpus;
+	struct perf_thread_map	*threads;
+	int			 nr_mmaps;
+	size_t			 mmap_len;
+	struct fdarray		 pollfd;
+	struct hlist_head	 heads[PERF_EVLIST__HLIST_SIZE];
+	struct perf_mmap	*mmap;
+	struct perf_mmap	*mmap_ovw;
+	struct perf_mmap	*mmap_first;
+	struct perf_mmap	*mmap_ovw_first;
+};
+
+typedef void
+(*perf_evlist_mmap__cb_idx_t)(struct perf_evlist*, struct perf_evsel*,
+			      struct perf_mmap_param*, int);
+typedef struct perf_mmap*
+(*perf_evlist_mmap__cb_get_t)(struct perf_evlist*, bool, int);
+typedef int
+(*perf_evlist_mmap__cb_mmap_t)(struct perf_mmap*, struct perf_mmap_param*, int, struct perf_cpu);
+
+struct perf_evlist_mmap_ops {
+	perf_evlist_mmap__cb_idx_t	idx;
+	perf_evlist_mmap__cb_get_t	get;
+	perf_evlist_mmap__cb_mmap_t	mmap;
+};
+
+int perf_evlist__alloc_pollfd(struct perf_evlist *evlist);
+int perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd,
+			    void *ptr, short revent, enum fdarray_flags flags);
+
+int perf_evlist__mmap_ops(struct perf_evlist *evlist,
+			  struct perf_evlist_mmap_ops *ops,
+			  struct perf_mmap_param *mp);
+
+void perf_evlist__init(struct perf_evlist *evlist);
+void perf_evlist__exit(struct perf_evlist *evlist);
+
+/**
+ * __perf_evlist__for_each_entry - iterate thru all the evsels
+ * @list: list_head instance to iterate
+ * @evsel: struct perf_evsel iterator
+ */
+#define __perf_evlist__for_each_entry(list, evsel) \
+	list_for_each_entry(evsel, list, node)
+
+/**
+ * evlist__for_each_entry - iterate thru all the evsels
+ * @evlist: perf_evlist instance to iterate
+ * @evsel: struct perf_evsel iterator
+ */
+#define perf_evlist__for_each_entry(evlist, evsel) \
+	__perf_evlist__for_each_entry(&(evlist)->entries, evsel)
+
+/**
+ * __perf_evlist__for_each_entry_reverse - iterate thru all the evsels in reverse order
+ * @list: list_head instance to iterate
+ * @evsel: struct evsel iterator
+ */
+#define __perf_evlist__for_each_entry_reverse(list, evsel) \
+	list_for_each_entry_reverse(evsel, list, node)
+
+/**
+ * perf_evlist__for_each_entry_reverse - iterate thru all the evsels in reverse order
+ * @evlist: evlist instance to iterate
+ * @evsel: struct evsel iterator
+ */
+#define perf_evlist__for_each_entry_reverse(evlist, evsel) \
+	__perf_evlist__for_each_entry_reverse(&(evlist)->entries, evsel)
+
+/**
+ * __perf_evlist__for_each_entry_safe - safely iterate thru all the evsels
+ * @list: list_head instance to iterate
+ * @tmp: struct evsel temp iterator
+ * @evsel: struct evsel iterator
+ */
+#define __perf_evlist__for_each_entry_safe(list, tmp, evsel) \
+	list_for_each_entry_safe(evsel, tmp, list, node)
+
+/**
+ * perf_evlist__for_each_entry_safe - safely iterate thru all the evsels
+ * @evlist: evlist instance to iterate
+ * @evsel: struct evsel iterator
+ * @tmp: struct evsel temp iterator
+ */
+#define perf_evlist__for_each_entry_safe(evlist, tmp, evsel) \
+	__perf_evlist__for_each_entry_safe(&(evlist)->entries, tmp, evsel)
+
+static inline struct perf_evsel *perf_evlist__first(struct perf_evlist *evlist)
+{
+	return list_entry(evlist->entries.next, struct perf_evsel, node);
+}
+
+static inline struct perf_evsel *perf_evlist__last(struct perf_evlist *evlist)
+{
+	return list_entry(evlist->entries.prev, struct perf_evsel, node);
+}
+
+u64 perf_evlist__read_format(struct perf_evlist *evlist);
+
+void perf_evlist__id_add(struct perf_evlist *evlist,
+			 struct perf_evsel *evsel,
+			 int cpu_map_idx, int thread, u64 id);
+
+int perf_evlist__id_add_fd(struct perf_evlist *evlist,
+			   struct perf_evsel *evsel,
+			   int cpu_map_idx, int thread, int fd);
+
+void perf_evlist__reset_id_hash(struct perf_evlist *evlist);
+
+void __perf_evlist__set_leader(struct list_head *list, struct perf_evsel *leader);
+
+void perf_evlist__go_system_wide(struct perf_evlist *evlist, struct perf_evsel *evsel);
+#endif /* __LIBPERF_INTERNAL_EVLIST_H */
diff --git a/tools/lib/perf/include/internal/evsel.h b/tools/lib/perf/include/internal/evsel.h
new file mode 100644
index 000000000..b988034f1
--- /dev/null
+++ b/tools/lib/perf/include/internal/evsel.h
@@ -0,0 +1,153 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_EVSEL_H
+#define __LIBPERF_INTERNAL_EVSEL_H
+
+#include <linux/types.h>
+#include <linux/perf_event.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <internal/cpumap.h>
+
+struct perf_thread_map;
+struct xyarray;
+
+/**
+ * The per-thread accumulated period storage node.
+ */
+struct perf_sample_id_period {
+	struct list_head	node;
+	struct hlist_node	hnode;
+	/* Holds total ID period value for PERF_SAMPLE_READ processing. */
+	u64			period;
+	/* The TID that the values belongs to */
+	u32			tid;
+};
+
+/**
+ * perf_evsel_for_each_per_thread_period_safe - safely iterate thru all the
+ * per_stream_periods
+ * @evlist:perf_evsel instance to iterate
+ * @item: struct perf_sample_id_period iterator
+ * @tmp: struct perf_sample_id_period temp iterator
+ */
+#define perf_evsel_for_each_per_thread_period_safe(evsel, tmp, item) \
+	list_for_each_entry_safe(item, tmp, &(evsel)->per_stream_periods, node)
+
+
+#define PERF_SAMPLE_ID__HLIST_BITS 4
+#define PERF_SAMPLE_ID__HLIST_SIZE (1 << PERF_SAMPLE_ID__HLIST_BITS)
+
+/*
+ * Per fd, to map back from PERF_SAMPLE_ID to evsel, only used when there are
+ * more than one entry in the evlist.
+ */
+struct perf_sample_id {
+	struct hlist_node	 node;
+	u64			 id;
+	struct perf_evsel	*evsel;
+       /*
+	* 'idx' will be used for AUX area sampling. A sample will have AUX area
+	* data that will be queued for decoding, where there are separate
+	* queues for each CPU (per-cpu tracing) or task (per-thread tracing).
+	* The sample ID can be used to lookup 'idx' which is effectively the
+	* queue number.
+	*/
+	int			 idx;
+	struct perf_cpu		 cpu;
+	pid_t			 tid;
+
+	/* Guest machine pid and VCPU, valid only if machine_pid is non-zero */
+	pid_t			 machine_pid;
+	struct perf_cpu		 vcpu;
+
+	/*
+	 * Per-thread, and global event counts are mutually exclusive:
+	 * Whilst it is possible to combine events into a group with differing
+	 * values of PERF_SAMPLE_READ, it is not valid to have inconsistent
+	 * values for `inherit`. Therefore it is not possible to have a
+	 * situation where a per-thread event is sampled as a global event;
+	 * all !inherit groups are global, and all groups where the sampling
+	 * event is inherit + PERF_SAMPLE_READ will be per-thread. Any event
+	 * that is part of such a group that is inherit but not PERF_SAMPLE_READ
+	 * will be read as per-thread. If such an event can also trigger a
+	 * sample (such as with sample_period > 0) then it will not cause
+	 * `read_format` to be included in its PERF_RECORD_SAMPLE, and
+	 * therefore will not expose the per-thread group members as global.
+	 */
+	union {
+		/*
+		 * Holds total ID period value for PERF_SAMPLE_READ processing
+		 * (when period is not per-thread).
+		 */
+		u64			period;
+		/*
+		 * Holds total ID period value for PERF_SAMPLE_READ processing
+		 * (when period is per-thread).
+		 */
+		struct hlist_head	periods[PERF_SAMPLE_ID__HLIST_SIZE];
+	};
+};
+
+struct perf_evsel {
+	struct list_head	 node;
+	struct perf_event_attr	 attr;
+	/** The commonly used cpu map of CPUs the event should be opened upon, etc. */
+	struct perf_cpu_map	*cpus;
+	/**
+	 * The cpu map read from the PMU. For core PMUs this is the list of all
+	 * CPUs the event can be opened upon. For other PMUs this is the default
+	 * cpu map for opening the event on, for example, the first CPU on a
+	 * socket for an uncore event.
+	 */
+	struct perf_cpu_map	*pmu_cpus;
+	struct perf_thread_map	*threads;
+	struct xyarray		*fd;
+	struct xyarray		*mmap;
+	struct xyarray		*sample_id;
+	u64			*id;
+	u32			 ids;
+	struct perf_evsel	*leader;
+
+	/* For events where the read_format value is per-thread rather than
+	 * global, stores the per-thread cumulative period */
+	struct list_head	per_stream_periods;
+
+	/* parse modifier helper */
+	int			 nr_members;
+	/*
+	 * system_wide is for events that need to be on every CPU, irrespective
+	 * of user requested CPUs or threads. Tha main example of this is the
+	 * dummy event. Map propagation will set cpus for this event to all CPUs
+	 * as software PMU events like dummy, have a CPU map that is empty.
+	 */
+	bool			 system_wide;
+	/*
+	 * Some events, for example uncore events, require a CPU.
+	 * i.e. it cannot be the 'any CPU' value of -1.
+	 */
+	bool			 requires_cpu;
+	/** Is the PMU for the event a core one? Effects the handling of own_cpus. */
+	bool			 is_pmu_core;
+	/** Does the evsel on read on the first CPU index such as tool time events? */
+	bool			 reads_only_on_cpu_idx0;
+	int			 idx;
+};
+
+void perf_evsel__init(struct perf_evsel *evsel, struct perf_event_attr *attr,
+		      int idx);
+void perf_evsel__exit(struct perf_evsel *evsel);
+int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads);
+void perf_evsel__close_fd(struct perf_evsel *evsel);
+void perf_evsel__free_fd(struct perf_evsel *evsel);
+int perf_evsel__read_size(struct perf_evsel *evsel);
+int perf_evsel__apply_filter(struct perf_evsel *evsel, const char *filter);
+
+int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads);
+void perf_evsel__free_id(struct perf_evsel *evsel);
+
+bool perf_evsel__attr_has_per_thread_sample_period(struct perf_evsel *evsel);
+
+u64 *perf_sample_id__get_period_storage(struct perf_sample_id *sid, u32 tid,
+					bool per_thread);
+
+#endif /* __LIBPERF_INTERNAL_EVSEL_H */
diff --git a/tools/lib/perf/include/internal/lib.h b/tools/lib/perf/include/internal/lib.h
new file mode 100644
index 000000000..85471a4b9
--- /dev/null
+++ b/tools/lib/perf/include/internal/lib.h
@@ -0,0 +1,14 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_LIB_H
+#define __LIBPERF_INTERNAL_LIB_H
+
+#include <sys/types.h>
+
+extern unsigned int page_size;
+
+ssize_t readn(int fd, void *buf, size_t n);
+ssize_t writen(int fd, const void *buf, size_t n);
+
+ssize_t preadn(int fd, void *buf, size_t n, off_t offs);
+
+#endif /* __LIBPERF_INTERNAL_CPUMAP_H */
diff --git a/tools/lib/perf/include/internal/mmap.h b/tools/lib/perf/include/internal/mmap.h
new file mode 100644
index 000000000..5f08cab61
--- /dev/null
+++ b/tools/lib/perf/include/internal/mmap.h
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_MMAP_H
+#define __LIBPERF_INTERNAL_MMAP_H
+
+#include <linux/compiler.h>
+#include <linux/refcount.h>
+#include <linux/types.h>
+#include <stdbool.h>
+#include <internal/cpumap.h>
+
+/* perf sample has 16 bits size limit */
+#define PERF_SAMPLE_MAX_SIZE (1 << 16)
+
+struct perf_mmap;
+struct perf_counts_values;
+
+typedef void (*libperf_unmap_cb_t)(struct perf_mmap *map);
+
+/**
+ * struct perf_mmap - perf's ring buffer mmap details
+ *
+ * @refcnt - e.g. code using PERF_EVENT_IOC_SET_OUTPUT to share this
+ */
+struct perf_mmap {
+	void			*base;
+	int			 mask;
+	int			 fd;
+	struct perf_cpu		 cpu;
+	refcount_t		 refcnt;
+	u64			 prev;
+	u64			 start;
+	u64			 end;
+	bool			 overwrite;
+	u64			 flush;
+	libperf_unmap_cb_t	 unmap_cb;
+	void			*event_copy;
+	size_t			 event_copy_sz;
+	struct perf_mmap	*next;
+};
+
+struct perf_mmap_param {
+	int	prot;
+	int	mask;
+};
+
+size_t perf_mmap__mmap_len(struct perf_mmap *map);
+
+void perf_mmap__init(struct perf_mmap *map, struct perf_mmap *prev,
+		     bool overwrite, libperf_unmap_cb_t unmap_cb);
+int perf_mmap__mmap(struct perf_mmap *map, struct perf_mmap_param *mp,
+		    int fd, struct perf_cpu cpu);
+void perf_mmap__munmap(struct perf_mmap *map);
+void perf_mmap__get(struct perf_mmap *map);
+void perf_mmap__put(struct perf_mmap *map);
+
+u64 perf_mmap__read_head(struct perf_mmap *map);
+
+int perf_mmap__read_self(struct perf_mmap *map, struct perf_counts_values *count);
+
+#endif /* __LIBPERF_INTERNAL_MMAP_H */
diff --git a/tools/lib/perf/include/internal/rc_check.h b/tools/lib/perf/include/internal/rc_check.h
new file mode 100644
index 000000000..f80ddfc80
--- /dev/null
+++ b/tools/lib/perf/include/internal/rc_check.h
@@ -0,0 +1,113 @@
+/* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */
+#ifndef __LIBPERF_INTERNAL_RC_CHECK_H
+#define __LIBPERF_INTERNAL_RC_CHECK_H
+
+#include <stdlib.h>
+#include <linux/zalloc.h>
+
+/*
+ * Enable reference count checking implicitly with leak checking, which is
+ * integrated into address sanitizer.
+ */
+#if defined(__SANITIZE_ADDRESS__) || defined(LEAK_SANITIZER) || defined(ADDRESS_SANITIZER)
+#define REFCNT_CHECKING 1
+#elif defined(__has_feature)
+#if __has_feature(address_sanitizer) || __has_feature(leak_sanitizer)
+#define REFCNT_CHECKING 1
+#endif
+#endif
+
+/*
+ * Shared reference count checking macros.
+ *
+ * Reference count checking is an approach to sanitizing the use of reference
+ * counted structs. It leverages address and leak sanitizers to make sure gets
+ * are paired with a put. Reference count checking adds a malloc-ed layer of
+ * indirection on a get, and frees it on a put. A missed put will be reported as
+ * a memory leak. A double put will be reported as a double free. Accessing
+ * after a put will cause a use-after-free and/or a segfault.
+ */
+
+#ifndef REFCNT_CHECKING
+/* Replaces "struct foo" so that the pointer may be interposed. */
+#define DECLARE_RC_STRUCT(struct_name)		\
+	struct struct_name
+
+/* Declare a reference counted struct variable. */
+#define RC_STRUCT(struct_name) struct struct_name
+
+/*
+ * Interpose the indirection. Result will hold the indirection and object is the
+ * reference counted struct.
+ */
+#define ADD_RC_CHK(result, object) (result = object, object)
+
+/* Strip the indirection layer. */
+#define RC_CHK_ACCESS(object) object
+
+/* Frees the object and the indirection layer. */
+#define RC_CHK_FREE(object) free(object)
+
+/* A get operation adding the indirection layer. */
+#define RC_CHK_GET(result, object) ADD_RC_CHK(result, object)
+
+/* A put operation removing the indirection layer. */
+#define RC_CHK_PUT(object) {}
+
+/* Pointer equality when the indirection may or may not be there. */
+#define RC_CHK_EQUAL(object1, object2) (object1 == object2)
+
+#else
+
+/* Replaces "struct foo" so that the pointer may be interposed. */
+#define DECLARE_RC_STRUCT(struct_name)			\
+	struct original_##struct_name;			\
+	struct struct_name {				\
+		struct original_##struct_name *orig;	\
+	};						\
+	struct original_##struct_name
+
+/* Declare a reference counted struct variable. */
+#define RC_STRUCT(struct_name) struct original_##struct_name
+
+/*
+ * Interpose the indirection. Result will hold the indirection and object is the
+ * reference counted struct.
+ */
+#define ADD_RC_CHK(result, object)					\
+	(								\
+		object ? (result = malloc(sizeof(*result)),		\
+			result ? (result->orig = object, result)	\
+			: (result = NULL, NULL))			\
+		: (result = NULL, NULL)					\
+		)
+
+/* Strip the indirection layer. */
+#define RC_CHK_ACCESS(object) object->orig
+
+/* Frees the object and the indirection layer. */
+#define RC_CHK_FREE(object)			\
+	do {					\
+		zfree(&object->orig);		\
+		free(object);			\
+	} while(0)
+
+/* A get operation adding the indirection layer. */
+#define RC_CHK_GET(result, object) ADD_RC_CHK(result, (object ? object->orig : NULL))
+
+/* A put operation removing the indirection layer. */
+#define RC_CHK_PUT(object)			\
+	do {					\
+		if (object) {			\
+			object->orig = NULL;	\
+			free(object);		\
+		}				\
+	} while(0)
+
+/* Pointer equality when the indirection may or may not be there. */
+#define RC_CHK_EQUAL(object1, object2) (object1 == object2 || \
+		(object1 && object2 && object1->orig == object2->orig))
+
+#endif
+
+#endif /* __LIBPERF_INTERNAL_RC_CHECK_H */
diff --git a/tools/lib/perf/include/internal/tests.h b/tools/lib/perf/include/internal/tests.h
new file mode 100644
index 000000000..b130a6663
--- /dev/null
+++ b/tools/lib/perf/include/internal/tests.h
@@ -0,0 +1,67 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_TESTS_H
+#define __LIBPERF_INTERNAL_TESTS_H
+
+#include <stdio.h>
+#include <unistd.h>
+
+extern int tests_failed;
+extern int tests_verbose;
+
+static inline int get_verbose(char **argv, int argc)
+{
+	int c;
+	int verbose = 0;
+
+	while ((c = getopt(argc, argv, "v")) != -1) {
+		switch (c)
+		{
+		case 'v':
+			verbose = 1;
+			break;
+		default:
+			break;
+		}
+	}
+	optind = 1;
+
+	return verbose;
+}
+
+#define __T_START					\
+do {							\
+	tests_verbose = get_verbose(argv, argc);	\
+	fprintf(stdout, "- running %s...", __FILE__);	\
+	fflush(NULL);					\
+	tests_failed = 0;				\
+} while (0)
+
+#define __T_END								\
+do {									\
+	if (tests_failed)						\
+		fprintf(stdout, "  FAILED (%d)\n", tests_failed);	\
+	else								\
+		fprintf(stdout, "OK\n");				\
+} while (0)
+
+#define __T(text, cond)                                                          \
+do {                                                                             \
+	if (!(cond)) {                                                           \
+		fprintf(stderr, "FAILED %s:%d %s\n", __FILE__, __LINE__, text);  \
+		tests_failed++;                                                  \
+		return -1;                                                       \
+	}                                                                        \
+} while (0)
+
+#define __T_VERBOSE(...)						\
+do {									\
+	if (tests_verbose) {						\
+		if (tests_verbose == 1) {				\
+			fputc('\n', stderr);				\
+			tests_verbose++;				\
+		}							\
+		fprintf(stderr, ##__VA_ARGS__);				\
+	}								\
+} while (0)
+
+#endif /* __LIBPERF_INTERNAL_TESTS_H */
diff --git a/tools/lib/perf/include/internal/threadmap.h b/tools/lib/perf/include/internal/threadmap.h
new file mode 100644
index 000000000..df748baf9
--- /dev/null
+++ b/tools/lib/perf/include/internal/threadmap.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_THREADMAP_H
+#define __LIBPERF_INTERNAL_THREADMAP_H
+
+#include <linux/refcount.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+struct thread_map_data {
+	pid_t	 pid;
+	char	*comm;
+};
+
+struct perf_thread_map {
+	refcount_t	refcnt;
+	int		nr;
+	int		err_thread;
+	struct thread_map_data map[];
+};
+
+struct perf_thread_map *perf_thread_map__realloc(struct perf_thread_map *map, int nr);
+
+#endif /* __LIBPERF_INTERNAL_THREADMAP_H */
diff --git a/tools/lib/perf/include/internal/xyarray.h b/tools/lib/perf/include/internal/xyarray.h
new file mode 100644
index 000000000..f10af3da7
--- /dev/null
+++ b/tools/lib/perf/include/internal/xyarray.h
@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_INTERNAL_XYARRAY_H
+#define __LIBPERF_INTERNAL_XYARRAY_H
+
+#include <linux/compiler.h>
+#include <sys/types.h>
+
+struct xyarray {
+	size_t row_size;
+	size_t entry_size;
+	size_t entries;
+	size_t max_x;
+	size_t max_y;
+	char contents[] __aligned(8);
+};
+
+struct xyarray *xyarray__new(int xlen, int ylen, size_t entry_size);
+void xyarray__delete(struct xyarray *xy);
+void xyarray__reset(struct xyarray *xy);
+
+static inline void *__xyarray__entry(struct xyarray *xy, int x, int y)
+{
+	return &xy->contents[x * xy->row_size + y * xy->entry_size];
+}
+
+static inline void *xyarray__entry(struct xyarray *xy, size_t x, size_t y)
+{
+	if (x >= xy->max_x || y >= xy->max_y)
+		return NULL;
+	return __xyarray__entry(xy, x, y);
+}
+
+static inline int xyarray__max_y(struct xyarray *xy)
+{
+	return xy->max_y;
+}
+
+static inline int xyarray__max_x(struct xyarray *xy)
+{
+	return xy->max_x;
+}
+
+#endif /* __LIBPERF_INTERNAL_XYARRAY_H */
diff --git a/tools/lib/perf/include/perf/bpf_perf.h b/tools/lib/perf/include/perf/bpf_perf.h
new file mode 100644
index 000000000..e7cf6ba7b
--- /dev/null
+++ b/tools/lib/perf/include/perf/bpf_perf.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */
+#ifndef __LIBPERF_BPF_PERF_H
+#define __LIBPERF_BPF_PERF_H
+
+#include <linux/types.h>  /* for __u32 */
+
+/*
+ * bpf_perf uses a hashmap, the attr_map, to track all the leader programs.
+ * The hashmap is pinned in bpffs. flock() on this file is used to ensure
+ * no concurrent access to the attr_map.  The key of attr_map is struct
+ * perf_event_attr, and the value is struct perf_event_attr_map_entry.
+ *
+ * struct perf_event_attr_map_entry contains two __u32 IDs, bpf_link of the
+ * leader prog, and the diff_map. Each perf-stat session holds a reference
+ * to the bpf_link to make sure the leader prog is attached to sched_switch
+ * tracepoint.
+ *
+ * Since the hashmap only contains IDs of the bpf_link and diff_map, it
+ * does not hold any references to the leader program. Once all perf-stat
+ * sessions of these events exit, the leader prog, its maps, and the
+ * perf_events will be freed.
+ */
+struct perf_event_attr_map_entry {
+	__u32 link_id;
+	__u32 diff_map_id;
+};
+
+/* default attr_map name */
+#define BPF_PERF_DEFAULT_ATTR_MAP_PATH "perf_attr_map"
+
+#endif /* __LIBPERF_BPF_PERF_H */
diff --git a/tools/lib/perf/include/perf/core.h b/tools/lib/perf/include/perf/core.h
new file mode 100644
index 000000000..06cc132d8
--- /dev/null
+++ b/tools/lib/perf/include/perf/core.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_CORE_H
+#define __LIBPERF_CORE_H
+
+#include <stdarg.h>
+
+#ifndef LIBPERF_API
+#define LIBPERF_API extern __attribute__((visibility("default")))
+#endif
+
+enum libperf_print_level {
+	LIBPERF_ERR,
+	LIBPERF_WARN,
+	LIBPERF_INFO,
+	LIBPERF_DEBUG,
+	LIBPERF_DEBUG2,
+	LIBPERF_DEBUG3,
+};
+
+typedef int (*libperf_print_fn_t)(enum libperf_print_level level,
+				  const char *, va_list ap);
+
+LIBPERF_API void libperf_init(libperf_print_fn_t fn);
+
+#endif /* __LIBPERF_CORE_H */
diff --git a/tools/lib/perf/include/perf/cpumap.h b/tools/lib/perf/include/perf/cpumap.h
new file mode 100644
index 000000000..e1a0b0d27
--- /dev/null
+++ b/tools/lib/perf/include/perf/cpumap.h
@@ -0,0 +1,110 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_CPUMAP_H
+#define __LIBPERF_CPUMAP_H
+
+#include <perf/core.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+/**
+ * struct perf_cpu - wrapper around a CPU number.
+ * @cpu: CPU number, -1 for the "any CPU"/dummy value.
+ *
+ * int16_t limits this to 32767 CPUs.  Widening to int requires a libperf
+ * ABI bump — see tools/lib/perf/TODO for the full scope.
+ */
+struct perf_cpu {
+	int16_t cpu;
+};
+
+struct perf_cache {
+	int cache_lvl;
+	int cache;
+};
+
+struct perf_cpu_map;
+
+/**
+ * perf_cpu_map__new_any_cpu - a map with a singular "any CPU"/dummy -1 value.
+ */
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__new_any_cpu(void);
+/**
+ * perf_cpu_map__new_online_cpus - a map read from
+ *                                 /sys/devices/system/cpu/online if
+ *                                 available. If reading wasn't possible a map
+ *                                 is created using the online processors
+ *                                 assuming the first 'n' processors are all
+ *                                 online.
+ */
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__new_online_cpus(void);
+/**
+ * perf_cpu_map__new - create a map from the given cpu_list such as "0-7". If no
+ *                     cpu_list argument is provided then
+ *                     perf_cpu_map__new_online_cpus is returned.
+ */
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__new(const char *cpu_list);
+/** perf_cpu_map__new_int - create a map with the one given cpu. */
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__new_int(int cpu);
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__get(struct perf_cpu_map *map);
+LIBPERF_API int perf_cpu_map__merge(struct perf_cpu_map **orig,
+				    struct perf_cpu_map *other);
+LIBPERF_API struct perf_cpu_map *perf_cpu_map__intersect(struct perf_cpu_map *orig,
+							 struct perf_cpu_map *other);
+LIBPERF_API void perf_cpu_map__put(struct perf_cpu_map *map);
+/**
+ * perf_cpu_map__cpu - get the CPU value at the given index. Returns -1 if index
+ *                     is invalid.
+ */
+LIBPERF_API struct perf_cpu perf_cpu_map__cpu(const struct perf_cpu_map *cpus, unsigned int idx);
+/**
+ * perf_cpu_map__nr - for an empty map returns 1, as perf_cpu_map__cpu returns a
+ *                    cpu of -1 for an invalid index, this makes an empty map
+ *                    look like it contains the "any CPU"/dummy value. Otherwise
+ *                    the result is the number CPUs in the map plus one if the
+ *                    "any CPU"/dummy value is present.
+ */
+LIBPERF_API unsigned int perf_cpu_map__nr(const struct perf_cpu_map *cpus);
+/**
+ * perf_cpu_map__has_any_cpu_or_is_empty - is map either empty or has the "any CPU"/dummy value.
+ */
+LIBPERF_API bool perf_cpu_map__has_any_cpu_or_is_empty(const struct perf_cpu_map *map);
+/**
+ * perf_cpu_map__is_any_cpu_or_is_empty - is map either empty or the "any CPU"/dummy value.
+ */
+LIBPERF_API bool perf_cpu_map__is_any_cpu_or_is_empty(const struct perf_cpu_map *map);
+/**
+ * perf_cpu_map__is_empty - does the map contain no values and it doesn't
+ *                          contain the special "any CPU"/dummy value.
+ */
+LIBPERF_API bool perf_cpu_map__is_empty(const struct perf_cpu_map *map);
+/**
+ * perf_cpu_map__min - the minimum CPU value or -1 if empty or just the "any CPU"/dummy value.
+ */
+LIBPERF_API struct perf_cpu perf_cpu_map__min(const struct perf_cpu_map *map);
+/**
+ * perf_cpu_map__max - the maximum CPU value or -1 if empty or just the "any CPU"/dummy value.
+ */
+LIBPERF_API struct perf_cpu perf_cpu_map__max(const struct perf_cpu_map *map);
+LIBPERF_API bool perf_cpu_map__has(const struct perf_cpu_map *map, struct perf_cpu cpu);
+LIBPERF_API bool perf_cpu_map__equal(const struct perf_cpu_map *lhs,
+				     const struct perf_cpu_map *rhs);
+/**
+ * perf_cpu_map__any_cpu - Does the map contain the "any CPU"/dummy -1 value?
+ */
+LIBPERF_API bool perf_cpu_map__has_any_cpu(const struct perf_cpu_map *map);
+
+#define perf_cpu_map__for_each_cpu(cpu, idx, cpus)		\
+	for ((idx) = 0, (cpu) = perf_cpu_map__cpu(cpus, idx);	\
+	     (idx) < perf_cpu_map__nr(cpus);			\
+	     (idx)++, (cpu) = perf_cpu_map__cpu(cpus, idx))
+
+#define perf_cpu_map__for_each_cpu_skip_any(_cpu, idx, cpus)	\
+	for ((idx) = 0, (_cpu) = perf_cpu_map__cpu(cpus, idx);	\
+	     (idx) < perf_cpu_map__nr(cpus);			\
+	     (idx)++, (_cpu) = perf_cpu_map__cpu(cpus, idx))	\
+		if ((_cpu).cpu != -1)
+
+#define perf_cpu_map__for_each_idx(idx, cpus)				\
+	for ((idx) = 0; (idx) < perf_cpu_map__nr(cpus); (idx)++)
+
+#endif /* __LIBPERF_CPUMAP_H */
diff --git a/tools/lib/perf/include/perf/event.h b/tools/lib/perf/include/perf/event.h
new file mode 100644
index 000000000..fdced574c
--- /dev/null
+++ b/tools/lib/perf/include/perf/event.h
@@ -0,0 +1,643 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_EVENT_H
+#define __LIBPERF_EVENT_H
+
+#include <linux/perf_event.h>
+#include <linux/types.h>
+#include <linux/limits.h>
+#include <linux/bpf.h>
+#include <sys/types.h> /* pid_t */
+
+/*
+ * Verify the full field fits within the event, not just its start offset.
+ * Only valid for fixed-size scalar fields — for trailing arrays like
+ * filename[PATH_MAX], sizeof() evaluates to the declared maximum, not
+ * the actual string length, so this would spuriously return false.
+ */
+#define event_contains(obj, mem) \
+	((obj).header.size >= offsetof(typeof(obj), mem) + sizeof((obj).mem))
+
+struct perf_record_mmap {
+	struct perf_event_header header;
+	__u32			 pid, tid;
+	__u64			 start;
+	__u64			 len;
+	__u64			 pgoff;
+	char			 filename[PATH_MAX];
+};
+
+struct perf_record_mmap2 {
+	struct perf_event_header header;
+	__u32			 pid, tid;
+	__u64			 start;
+	__u64			 len;
+	__u64			 pgoff;
+	union {
+		struct {
+			__u32	 maj;
+			__u32	 min;
+			__u64	 ino;
+			__u64	 ino_generation;
+		};
+		struct {
+			__u8	 build_id_size;
+			__u8	 __reserved_1;
+			__u16	 __reserved_2;
+			__u8	 build_id[20];
+		};
+	};
+	__u32			 prot;
+	__u32			 flags;
+	char			 filename[PATH_MAX];
+};
+
+struct perf_record_comm {
+	struct perf_event_header header;
+	__u32			 pid, tid;
+	char			 comm[16];
+};
+
+struct perf_record_namespaces {
+	struct perf_event_header header;
+	__u32			 pid, tid;
+	__u64			 nr_namespaces;
+	struct perf_ns_link_info link_info[];
+};
+
+struct perf_record_fork {
+	struct perf_event_header header;
+	__u32			 pid, ppid;
+	__u32			 tid, ptid;
+	__u64			 time;
+};
+
+struct perf_record_lost {
+	struct perf_event_header header;
+	__u64			 id;
+	__u64			 lost;
+};
+
+#define PERF_RECORD_MISC_LOST_SAMPLES_BPF (1 << 15)
+
+struct perf_record_lost_samples {
+	struct perf_event_header header;
+	__u64			 lost;
+};
+
+#define MAX_ID_HDR_ENTRIES  6
+struct perf_record_lost_samples_and_ids {
+	struct perf_record_lost_samples lost;
+	__u64 sample_ids[MAX_ID_HDR_ENTRIES];
+};
+
+/*
+ * PERF_FORMAT_ENABLED | PERF_FORMAT_RUNNING | PERF_FORMAT_ID | PERF_FORMAT_LOST
+ */
+struct perf_record_read {
+	struct perf_event_header header;
+	__u32			 pid, tid;
+	__u64			 value;
+	__u64			 time_enabled;
+	__u64			 time_running;
+	__u64			 id;
+	__u64			 lost;
+};
+
+struct perf_record_throttle {
+	struct perf_event_header header;
+	__u64			 time;
+	__u64			 id;
+	__u64			 stream_id;
+};
+
+#ifndef KSYM_NAME_LEN
+#define KSYM_NAME_LEN 512
+#endif
+
+struct perf_record_ksymbol {
+	struct perf_event_header header;
+	__u64			 addr;
+	__u32			 len;
+	__u16			 ksym_type;
+	__u16			 flags;
+	char			 name[KSYM_NAME_LEN];
+};
+
+struct perf_record_bpf_event {
+	struct perf_event_header header;
+	__u16			 type;
+	__u16			 flags;
+	__u32			 id;
+
+	/* for bpf_prog types */
+	__u8			 tag[BPF_TAG_SIZE];  // prog tag
+};
+
+struct perf_record_cgroup {
+	struct perf_event_header header;
+	__u64			 id;
+	char			 path[PATH_MAX];
+};
+
+struct perf_record_text_poke_event {
+	struct perf_event_header header;
+	__u64			addr;
+	__u16			old_len;
+	__u16			new_len;
+	__u8			bytes[];
+};
+
+struct perf_record_sample {
+	struct perf_event_header header;
+	__u64			 array[];
+};
+
+struct perf_record_switch {
+	struct perf_event_header header;
+	__u32			 next_prev_pid;
+	__u32			 next_prev_tid;
+};
+
+struct perf_record_callchain_deferred {
+	struct perf_event_header header;
+	/*
+	 * This is to match kernel and (deferred) user stacks together.
+	 * The kernel part will be in the sample callchain array after
+	 * the PERF_CONTEXT_USER_DEFERRED entry.
+	 */
+	__u64			 cookie;
+	__u64			 nr;
+	__u64			 ips[];
+};
+
+struct perf_record_header_attr {
+	struct perf_event_header header;
+	struct perf_event_attr	 attr;
+	/*
+	 * Array of u64 id follows here but we cannot use a flexible array
+	 * because size of attr in the data can be different then current
+	 * version.  Please use perf_record_header_attr_id() below.
+	 *
+	 * __u64		 id[];  // do not use this
+	 */
+};
+
+/* Returns the pointer to id array based on the actual attr size. */
+#define perf_record_header_attr_id(evt)			\
+	((void *)&(evt)->attr.attr + (evt)->attr.attr.size)
+
+enum {
+	PERF_CPU_MAP__CPUS = 0,
+	PERF_CPU_MAP__MASK = 1,
+	PERF_CPU_MAP__RANGE_CPUS = 2,
+};
+
+/*
+ * Array encoding of a perf_cpu_map where nr is the number of entries in cpu[]
+ * and each entry is a value for a CPU in the map.
+ */
+struct cpu_map_entries {
+	__u16			 nr;
+	__u16			 cpu[];
+};
+
+/* Bitmap encoding of a perf_cpu_map where bitmap entries are 32-bit. */
+struct perf_record_mask_cpu_map32 {
+	/* Number of mask values. */
+	__u16			 nr;
+	/* Constant 4. */
+	__u16			 long_size;
+	/* Bitmap data. */
+	__u32			 mask[];
+};
+
+/* Bitmap encoding of a perf_cpu_map where bitmap entries are 64-bit. */
+struct perf_record_mask_cpu_map64 {
+	/* Number of mask values. */
+	__u16			 nr;
+	/* Constant 8. */
+	__u16			 long_size;
+	/* Legacy padding. */
+	char                     __pad[4];
+	/* Bitmap data. */
+	__u64			 mask[];
+};
+
+/*
+ * 'struct perf_record_cpu_map_data' is packed as unfortunately an earlier
+ * version had unaligned data and we wish to retain file format compatibility.
+ * -irogers
+ */
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wpacked"
+#pragma GCC diagnostic ignored "-Wattributes"
+
+/*
+ * An encoding of a CPU map for a range starting at start_cpu through to
+ * end_cpu. If any_cpu is 1, an any CPU (-1) value (aka dummy value) is present.
+ */
+struct perf_record_range_cpu_map {
+	__u8 any_cpu;
+	__u8 __pad;
+	__u16 start_cpu;
+	__u16 end_cpu;
+};
+
+struct perf_record_cpu_map_data {
+	__u16			 type;
+	union {
+		/* Used when type == PERF_CPU_MAP__CPUS. */
+		struct cpu_map_entries cpus_data;
+		/* Used when type == PERF_CPU_MAP__MASK and long_size == 4. */
+		struct perf_record_mask_cpu_map32 mask32_data;
+		/* Used when type == PERF_CPU_MAP__MASK and long_size == 8. */
+		struct perf_record_mask_cpu_map64 mask64_data;
+		/* Used when type == PERF_CPU_MAP__RANGE_CPUS. */
+		struct perf_record_range_cpu_map range_cpu_data;
+	};
+} __attribute__((packed));
+
+#pragma GCC diagnostic pop
+
+struct perf_record_cpu_map {
+	struct perf_event_header	 header;
+	struct perf_record_cpu_map_data	 data;
+};
+
+enum {
+	PERF_EVENT_UPDATE__UNIT  = 0,
+	PERF_EVENT_UPDATE__SCALE = 1,
+	PERF_EVENT_UPDATE__NAME  = 2,
+	PERF_EVENT_UPDATE__CPUS  = 3,
+};
+
+struct perf_record_event_update_cpus {
+	struct perf_record_cpu_map_data	 cpus;
+};
+
+struct perf_record_event_update_scale {
+	double			 scale;
+};
+
+struct perf_record_event_update {
+	struct perf_event_header header;
+	__u64			 type;
+	__u64			 id;
+	union {
+		/* Used when type == PERF_EVENT_UPDATE__SCALE. */
+		struct perf_record_event_update_scale scale;
+		/* Used when type == PERF_EVENT_UPDATE__UNIT. */
+		char unit[0];
+		/* Used when type == PERF_EVENT_UPDATE__NAME. */
+		char name[0];
+		/* Used when type == PERF_EVENT_UPDATE__CPUS. */
+		struct perf_record_event_update_cpus cpus;
+	};
+};
+
+#define MAX_EVENT_NAME 64
+
+struct perf_trace_event_type {
+	__u64			 event_id;
+	char			 name[MAX_EVENT_NAME];
+};
+
+struct perf_record_header_event_type {
+	struct perf_event_header	 header;
+	struct perf_trace_event_type	 event_type;
+};
+
+struct perf_record_header_tracing_data {
+	struct perf_event_header header;
+	__u32			 size;
+	__u32			 pad;
+};
+
+#define PERF_RECORD_MISC_BUILD_ID_SIZE (1 << 15)
+
+struct perf_record_header_build_id {
+	struct perf_event_header header;
+	pid_t			 pid;
+	union {
+		__u8		 build_id[24];
+		struct {
+			__u8	 data[20];
+			__u8	 size;
+			__u8	 reserved1__;
+			__u16	 reserved2__;
+		};
+	};
+	char			 filename[];
+};
+
+struct id_index_entry {
+	__u64			 id;
+	__u64			 idx;
+	__u64			 cpu;
+	__u64			 tid;
+};
+
+struct id_index_entry_2 {
+	__u64			 machine_pid;
+	__u64			 vcpu;
+};
+
+struct perf_record_id_index {
+	struct perf_event_header header;
+	__u64			 nr;
+	struct id_index_entry	 entries[];
+};
+
+struct perf_record_auxtrace_info {
+	struct perf_event_header header;
+	__u32			 type;
+	__u32			 reserved__; /* For alignment */
+	__u64			 priv[];
+};
+
+struct perf_record_auxtrace {
+	struct perf_event_header header;
+	__u64			 size;
+	__u64			 offset;
+	__u64			 reference;
+	__u32			 idx;
+	__u32			 tid;
+	__u32			 cpu;
+	__u32			 reserved__; /* For alignment */
+};
+
+#define MAX_AUXTRACE_ERROR_MSG 64
+
+struct perf_record_auxtrace_error {
+	struct perf_event_header header;
+	__u32			 type;
+	__u32			 code;
+	__u32			 cpu;
+	__u32			 pid;
+	__u32			 tid;
+	__u32			 fmt;
+	__u64			 ip;
+	__u64			 time;
+	char			 msg[MAX_AUXTRACE_ERROR_MSG];
+	__u32			 machine_pid;
+	__u32			 vcpu;
+};
+
+struct perf_record_aux {
+	struct perf_event_header header;
+	__u64			 aux_offset;
+	__u64			 aux_size;
+	__u64			 flags;
+};
+
+struct perf_record_itrace_start {
+	struct perf_event_header header;
+	__u32			 pid;
+	__u32			 tid;
+};
+
+struct perf_record_aux_output_hw_id {
+	struct perf_event_header header;
+	__u64			hw_id;
+};
+
+struct perf_record_thread_map_entry {
+	__u64			 pid;
+	char			 comm[16];
+};
+
+struct perf_record_thread_map {
+	struct perf_event_header		 header;
+	__u64					 nr;
+	struct perf_record_thread_map_entry	 entries[];
+};
+
+enum {
+	PERF_STAT_CONFIG_TERM__AGGR_MODE	= 0,
+	PERF_STAT_CONFIG_TERM__INTERVAL		= 1,
+	PERF_STAT_CONFIG_TERM__SCALE		= 2,
+	PERF_STAT_CONFIG_TERM__AGGR_LEVEL	= 3,
+	PERF_STAT_CONFIG_TERM__MAX		= 4,
+};
+
+struct perf_record_stat_config_entry {
+	__u64			 tag;
+	__u64			 val;
+};
+
+struct perf_record_stat_config {
+	struct perf_event_header		 header;
+	__u64					 nr;
+	struct perf_record_stat_config_entry	 data[];
+};
+
+struct perf_record_stat {
+	struct perf_event_header header;
+
+	__u64			 id;
+	__u32			 cpu;
+	__u32			 thread;
+
+	union {
+		struct {
+			__u64	 val;
+			__u64	 ena;
+			__u64	 run;
+		};
+		__u64		 values[3];
+	};
+};
+
+struct perf_record_stat_round {
+	struct perf_event_header header;
+	__u64			 type;
+	__u64			 time;
+};
+
+struct perf_record_time_conv {
+	struct perf_event_header header;
+	__u64			 time_shift;
+	__u64			 time_mult;
+	__u64			 time_zero;
+	__u64			 time_cycles;
+	__u64			 time_mask;
+	__u8			 cap_user_time_zero;
+	__u8			 cap_user_time_short;
+	__u8			 reserved[6];	/* For alignment */
+};
+
+struct perf_record_header_feature {
+	struct perf_event_header header;
+	__u64			 feat_id;
+	char			 data[];
+};
+
+struct perf_record_compressed {
+	struct perf_event_header header;
+	char			 data[];
+};
+
+/*
+ * `header.size` includes the padding we are going to add while writing the record.
+ * `data_size` only includes the size of `data[]` itself.
+ */
+struct perf_record_compressed2 {
+	struct perf_event_header header;
+	__u64			 data_size;
+	char			 data[];
+};
+
+#define BPF_METADATA_KEY_LEN   64
+#define BPF_METADATA_VALUE_LEN 256
+#define BPF_PROG_NAME_LEN      KSYM_NAME_LEN
+
+struct perf_record_bpf_metadata_entry {
+	char key[BPF_METADATA_KEY_LEN];
+	char value[BPF_METADATA_VALUE_LEN];
+};
+
+struct perf_record_bpf_metadata {
+	struct perf_event_header	      header;
+	char				      prog_name[BPF_PROG_NAME_LEN];
+	__u64				      nr_entries;
+	struct perf_record_bpf_metadata_entry entries[];
+};
+
+struct perf_record_schedstat_cpu_v15 {
+#define CPU_FIELD(_type, _name, _desc, _format, _is_pct, _pct_of, _ver)		_type _name
+#include "schedstat-v15.h"
+#undef CPU_FIELD
+};
+
+struct perf_record_schedstat_cpu_v16 {
+#define CPU_FIELD(_type, _name, _desc, _format, _is_pct, _pct_of, _ver)		_type _name
+#include "schedstat-v16.h"
+#undef CPU_FIELD
+};
+
+struct perf_record_schedstat_cpu_v17 {
+#define CPU_FIELD(_type, _name, _desc, _format, _is_pct, _pct_of, _ver)		_type _name
+#include "schedstat-v17.h"
+#undef CPU_FIELD
+};
+
+struct perf_record_schedstat_cpu {
+	struct perf_event_header header;
+	__u64			 timestamp;
+	__u32			 cpu;
+	__u16			 version;
+	/* Padding */
+	char			 __pad[2];
+	union {
+		struct perf_record_schedstat_cpu_v15 v15;
+		struct perf_record_schedstat_cpu_v16 v16;
+		struct perf_record_schedstat_cpu_v17 v17;
+	};
+};
+
+struct perf_record_schedstat_domain_v15 {
+#define DOMAIN_FIELD(_type, _name, _desc, _format, _is_jiffies, _ver)		_type _name
+#include "schedstat-v15.h"
+#undef DOMAIN_FIELD
+};
+
+struct perf_record_schedstat_domain_v16 {
+#define DOMAIN_FIELD(_type, _name, _desc, _format, _is_jiffies, _ver)		_type _name
+#include "schedstat-v16.h"
+#undef DOMAIN_FIELD
+};
+
+struct perf_record_schedstat_domain_v17 {
+#define DOMAIN_FIELD(_type, _name, _desc, _format, _is_jiffies, _ver)		_type _name
+#include "schedstat-v17.h"
+#undef DOMAIN_FIELD
+};
+
+#define DOMAIN_NAME_LEN		16
+
+struct perf_record_schedstat_domain {
+	struct perf_event_header header;
+	__u64			 timestamp;
+	__u32			 cpu;
+	__u16			 version;
+	__u16			 domain;
+	union {
+		struct perf_record_schedstat_domain_v15 v15;
+		struct perf_record_schedstat_domain_v16 v16;
+		struct perf_record_schedstat_domain_v17 v17;
+	};
+};
+
+enum perf_user_event_type { /* above any possible kernel type */
+	PERF_RECORD_USER_TYPE_START		= 64,
+	PERF_RECORD_HEADER_ATTR			= 64,
+	PERF_RECORD_HEADER_EVENT_TYPE		= 65, /* deprecated */
+	PERF_RECORD_HEADER_TRACING_DATA		= 66,
+	PERF_RECORD_HEADER_BUILD_ID		= 67,
+	PERF_RECORD_FINISHED_ROUND		= 68,
+	PERF_RECORD_ID_INDEX			= 69,
+	PERF_RECORD_AUXTRACE_INFO		= 70,
+	PERF_RECORD_AUXTRACE			= 71,
+	PERF_RECORD_AUXTRACE_ERROR		= 72,
+	PERF_RECORD_THREAD_MAP			= 73,
+	PERF_RECORD_CPU_MAP			= 74,
+	PERF_RECORD_STAT_CONFIG			= 75,
+	PERF_RECORD_STAT			= 76,
+	PERF_RECORD_STAT_ROUND			= 77,
+	PERF_RECORD_EVENT_UPDATE		= 78,
+	PERF_RECORD_TIME_CONV			= 79,
+	PERF_RECORD_HEADER_FEATURE		= 80,
+	PERF_RECORD_COMPRESSED			= 81,
+	PERF_RECORD_FINISHED_INIT		= 82,
+	PERF_RECORD_COMPRESSED2			= 83,
+	PERF_RECORD_BPF_METADATA		= 84,
+	PERF_RECORD_SCHEDSTAT_CPU		= 85,
+	PERF_RECORD_SCHEDSTAT_DOMAIN		= 86,
+	PERF_RECORD_HEADER_MAX
+};
+
+union perf_event {
+	struct perf_event_header		header;
+	struct perf_record_mmap			mmap;
+	struct perf_record_mmap2		mmap2;
+	struct perf_record_comm			comm;
+	struct perf_record_namespaces		namespaces;
+	struct perf_record_cgroup		cgroup;
+	struct perf_record_fork			fork;
+	struct perf_record_lost			lost;
+	struct perf_record_lost_samples		lost_samples;
+	struct perf_record_read			read;
+	struct perf_record_throttle		throttle;
+	struct perf_record_sample		sample;
+	struct perf_record_callchain_deferred	callchain_deferred;
+	struct perf_record_bpf_event		bpf;
+	struct perf_record_ksymbol		ksymbol;
+	struct perf_record_text_poke_event	text_poke;
+	struct perf_record_header_attr		attr;
+	struct perf_record_event_update		event_update;
+	struct perf_record_header_event_type	event_type;
+	struct perf_record_header_tracing_data	tracing_data;
+	struct perf_record_header_build_id	build_id;
+	struct perf_record_id_index		id_index;
+	struct perf_record_auxtrace_info	auxtrace_info;
+	struct perf_record_auxtrace		auxtrace;
+	struct perf_record_auxtrace_error	auxtrace_error;
+	struct perf_record_aux			aux;
+	struct perf_record_itrace_start		itrace_start;
+	struct perf_record_aux_output_hw_id	aux_output_hw_id;
+	struct perf_record_switch		context_switch;
+	struct perf_record_thread_map		thread_map;
+	struct perf_record_cpu_map		cpu_map;
+	struct perf_record_stat_config		stat_config;
+	struct perf_record_stat			stat;
+	struct perf_record_stat_round		stat_round;
+	struct perf_record_time_conv		time_conv;
+	struct perf_record_header_feature	feat;
+	struct perf_record_compressed		pack;
+	struct perf_record_compressed2		pack2;
+	struct perf_record_bpf_metadata		bpf_metadata;
+	struct perf_record_schedstat_cpu	schedstat_cpu;
+	struct perf_record_schedstat_domain	schedstat_domain;
+};
+
+#endif /* __LIBPERF_EVENT_H */
diff --git a/tools/lib/perf/include/perf/evlist.h b/tools/lib/perf/include/perf/evlist.h
new file mode 100644
index 000000000..e894b7707
--- /dev/null
+++ b/tools/lib/perf/include/perf/evlist.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_EVLIST_H
+#define __LIBPERF_EVLIST_H
+
+#include <perf/core.h>
+#include <stdbool.h>
+
+struct perf_evlist;
+struct perf_evsel;
+struct perf_cpu_map;
+struct perf_thread_map;
+
+LIBPERF_API void perf_evlist__add(struct perf_evlist *evlist,
+				  struct perf_evsel *evsel);
+LIBPERF_API void perf_evlist__remove(struct perf_evlist *evlist,
+				     struct perf_evsel *evsel);
+LIBPERF_API struct perf_evlist *perf_evlist__new(void);
+LIBPERF_API void perf_evlist__delete(struct perf_evlist *evlist);
+LIBPERF_API struct perf_evsel* perf_evlist__next(struct perf_evlist *evlist,
+						 struct perf_evsel *evsel);
+LIBPERF_API int perf_evlist__open(struct perf_evlist *evlist);
+LIBPERF_API void perf_evlist__close(struct perf_evlist *evlist);
+LIBPERF_API void perf_evlist__enable(struct perf_evlist *evlist);
+LIBPERF_API void perf_evlist__disable(struct perf_evlist *evlist);
+
+#define perf_evlist__for_each_evsel(evlist, pos)	\
+	for ((pos) = perf_evlist__next((evlist), NULL);	\
+	     (pos) != NULL;				\
+	     (pos) = perf_evlist__next((evlist), (pos)))
+
+LIBPERF_API void perf_evlist__set_maps(struct perf_evlist *evlist,
+				       struct perf_cpu_map *cpus,
+				       struct perf_thread_map *threads);
+LIBPERF_API int perf_evlist__poll(struct perf_evlist *evlist, int timeout);
+LIBPERF_API int perf_evlist__filter_pollfd(struct perf_evlist *evlist,
+					   short revents_and_mask);
+
+LIBPERF_API int perf_evlist__mmap(struct perf_evlist *evlist, int pages);
+LIBPERF_API void perf_evlist__munmap(struct perf_evlist *evlist);
+
+LIBPERF_API struct perf_mmap *perf_evlist__next_mmap(struct perf_evlist *evlist,
+						     struct perf_mmap *map,
+						     bool overwrite);
+#define perf_evlist__for_each_mmap(evlist, pos, overwrite)		\
+	for ((pos) = perf_evlist__next_mmap((evlist), NULL, overwrite);	\
+	     (pos) != NULL;						\
+	     (pos) = perf_evlist__next_mmap((evlist), (pos), overwrite))
+
+LIBPERF_API void perf_evlist__set_leader(struct perf_evlist *evlist);
+LIBPERF_API int perf_evlist__nr_groups(struct perf_evlist *evlist);
+#endif /* __LIBPERF_EVLIST_H */
diff --git a/tools/lib/perf/include/perf/evsel.h b/tools/lib/perf/include/perf/evsel.h
new file mode 100644
index 000000000..6f9220407
--- /dev/null
+++ b/tools/lib/perf/include/perf/evsel.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_EVSEL_H
+#define __LIBPERF_EVSEL_H
+
+#include <stdint.h>
+#include <perf/core.h>
+#include <stdbool.h>
+#include <linux/types.h>
+
+struct perf_evsel;
+struct perf_event_attr;
+struct perf_cpu_map;
+struct perf_thread_map;
+
+struct perf_counts_values {
+	union {
+		struct {
+			uint64_t val;
+			uint64_t ena;
+			uint64_t run;
+			uint64_t id;
+			uint64_t lost;
+		};
+		uint64_t values[5];
+	};
+};
+
+LIBPERF_API struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr);
+LIBPERF_API void perf_evsel__delete(struct perf_evsel *evsel);
+LIBPERF_API int perf_evsel__open(struct perf_evsel *evsel, struct perf_cpu_map *cpus,
+				 struct perf_thread_map *threads);
+LIBPERF_API void perf_evsel__close(struct perf_evsel *evsel);
+LIBPERF_API void perf_evsel__close_cpu(struct perf_evsel *evsel, int cpu_map_idx);
+LIBPERF_API int perf_evsel__mmap(struct perf_evsel *evsel, int pages);
+LIBPERF_API void perf_evsel__munmap(struct perf_evsel *evsel);
+LIBPERF_API void *perf_evsel__mmap_base(struct perf_evsel *evsel, int cpu_map_idx, int thread);
+LIBPERF_API int perf_evsel__read(struct perf_evsel *evsel, int cpu_map_idx, int thread,
+				 struct perf_counts_values *count);
+LIBPERF_API int perf_evsel__enable(struct perf_evsel *evsel);
+LIBPERF_API int perf_evsel__enable_cpu(struct perf_evsel *evsel, int cpu_map_idx);
+LIBPERF_API int perf_evsel__enable_thread(struct perf_evsel *evsel, int thread);
+LIBPERF_API int perf_evsel__disable(struct perf_evsel *evsel);
+LIBPERF_API int perf_evsel__disable_cpu(struct perf_evsel *evsel, int cpu_map_idx);
+LIBPERF_API struct perf_cpu_map *perf_evsel__cpus(struct perf_evsel *evsel);
+LIBPERF_API struct perf_thread_map *perf_evsel__threads(struct perf_evsel *evsel);
+LIBPERF_API struct perf_event_attr *perf_evsel__attr(struct perf_evsel *evsel);
+LIBPERF_API void perf_counts_values__scale(struct perf_counts_values *count,
+					   bool scale, __s8 *pscaled);
+
+#endif /* __LIBPERF_EVSEL_H */
diff --git a/tools/lib/perf/include/perf/mmap.h b/tools/lib/perf/include/perf/mmap.h
new file mode 100644
index 000000000..9508ad90d
--- /dev/null
+++ b/tools/lib/perf/include/perf/mmap.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_MMAP_H
+#define __LIBPERF_MMAP_H
+
+#include <perf/core.h>
+
+struct perf_mmap;
+union perf_event;
+
+LIBPERF_API void perf_mmap__consume(struct perf_mmap *map);
+LIBPERF_API int perf_mmap__read_init(struct perf_mmap *map);
+LIBPERF_API void perf_mmap__read_done(struct perf_mmap *map);
+LIBPERF_API union perf_event *perf_mmap__read_event(struct perf_mmap *map);
+
+#endif /* __LIBPERF_MMAP_H */
diff --git a/tools/lib/perf/include/perf/schedstat-v15.h b/tools/lib/perf/include/perf/schedstat-v15.h
new file mode 100644
index 000000000..639458df0
--- /dev/null
+++ b/tools/lib/perf/include/perf/schedstat-v15.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifdef CPU_FIELD
+CPU_FIELD(__u32, yld_count, "sched_yield() count",
+	  "%11u", false, yld_count, v15);
+CPU_FIELD(__u32, array_exp, "Legacy counter can be ignored",
+	  "%11u", false, array_exp, v15);
+CPU_FIELD(__u32, sched_count, "schedule() called",
+	  "%11u", false, sched_count, v15);
+CPU_FIELD(__u32, sched_goidle, "schedule() left the processor idle",
+	  "%11u", true, sched_count, v15);
+CPU_FIELD(__u32, ttwu_count, "try_to_wake_up() was called",
+	  "%11u", false, ttwu_count, v15);
+CPU_FIELD(__u32, ttwu_local, "try_to_wake_up() was called to wake up the local cpu",
+	  "%11u", true, ttwu_count, v15);
+CPU_FIELD(__u64, rq_cpu_time, "total runtime by tasks on this processor (in jiffies)",
+	  "%11llu", false, rq_cpu_time, v15);
+CPU_FIELD(__u64, run_delay, "total waittime by tasks on this processor (in jiffies)",
+	  "%11llu", true, rq_cpu_time, v15);
+CPU_FIELD(__u64, pcount, "total timeslices run on this cpu",
+	  "%11llu", false, pcount, v15);
+#endif
+
+#ifdef DOMAIN_FIELD
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category idle> ");
+#endif
+DOMAIN_FIELD(__u32, idle_lb_count,
+	     "load_balance() count on cpu idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, idle_lb_balanced,
+	     "load_balance() found balanced on cpu idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, idle_lb_failed,
+	     "load_balance() move task failed on cpu idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, idle_lb_imbalance,
+	     "imbalance sum on cpu idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, idle_lb_gained,
+	     "pull_task() count on cpu idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, idle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, idle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, idle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu idle", "%11u", true, v15);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(idle_lb_success_count, "load_balance() success count on cpu idle", "%11u",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, v15);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(idle_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu idle)", "%11.2Lf",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, idle_lb_gained, v15);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category busy> ");
+#endif
+DOMAIN_FIELD(__u32, busy_lb_count,
+	     "load_balance() count on cpu busy", "%11u", true, v15);
+DOMAIN_FIELD(__u32, busy_lb_balanced,
+	     "load_balance() found balanced on cpu busy", "%11u", true, v15);
+DOMAIN_FIELD(__u32, busy_lb_failed,
+	     "load_balance() move task failed on cpu busy", "%11u", true, v15);
+DOMAIN_FIELD(__u32, busy_lb_imbalance,
+	     "imbalance sum on cpu busy", "%11u", false, v15);
+DOMAIN_FIELD(__u32, busy_lb_gained,
+	     "pull_task() count on cpu busy", "%11u", false, v15);
+DOMAIN_FIELD(__u32, busy_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu busy", "%11u", false, v15);
+DOMAIN_FIELD(__u32, busy_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu busy", "%11u", true, v15);
+DOMAIN_FIELD(__u32, busy_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu busy", "%11u", true, v15);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(busy_lb_success_count, "load_balance() success count on cpu busy", "%11u",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, v15);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(busy_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu busy)", "%11.2Lf",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, busy_lb_gained, v15);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category newidle> ");
+#endif
+DOMAIN_FIELD(__u32, newidle_lb_count,
+	     "load_balance() count on cpu newly idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, newidle_lb_balanced,
+	     "load_balance() found balanced on cpu newly idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, newidle_lb_failed,
+	     "load_balance() move task failed on cpu newly idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance,
+	     "imbalance sum on cpu newly idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, newidle_lb_gained,
+	     "pull_task() count on cpu newly idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, newidle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu newly idle", "%11u", false, v15);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu newly idle", "%11u", true, v15);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu newly idle", "%11u", true, v15);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(newidle_lb_success_count,
+		  "load_balance() success count on cpu newly idle", "%11u",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, v15);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(newidle_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu newly idle)", "%11.2Lf",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, newidle_lb_gained, v15);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category active_load_balance()> ");
+#endif
+DOMAIN_FIELD(__u32, alb_count,
+	     "active_load_balance() count", "%11u", false, v15);
+DOMAIN_FIELD(__u32, alb_failed,
+	     "active_load_balance() move task failed", "%11u", false, v15);
+DOMAIN_FIELD(__u32, alb_pushed,
+	     "active_load_balance() successfully moved a task", "%11u", false, v15);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_exec()> ");
+#endif
+DOMAIN_FIELD(__u32, sbe_count,
+	     "sbe_count is not used", "%11u", false, v15);
+DOMAIN_FIELD(__u32, sbe_balanced,
+	     "sbe_balanced is not used", "%11u", false, v15);
+DOMAIN_FIELD(__u32, sbe_pushed,
+	     "sbe_pushed is not used", "%11u", false, v15);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_fork()> ");
+#endif
+DOMAIN_FIELD(__u32, sbf_count,
+	     "sbf_count is not used", "%11u", false, v15);
+DOMAIN_FIELD(__u32, sbf_balanced,
+	     "sbf_balanced is not used", "%11u", false, v15);
+DOMAIN_FIELD(__u32, sbf_pushed,
+	     "sbf_pushed is not used", "%11u", false, v15);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Wakeup Info> ");
+#endif
+DOMAIN_FIELD(__u32, ttwu_wake_remote,
+	     "try_to_wake_up() awoke a task that last ran on a diff cpu", "%11u", false, v15);
+DOMAIN_FIELD(__u32, ttwu_move_affine,
+	     "try_to_wake_up() moved task because cache-cold on own cpu", "%11u", false, v15);
+DOMAIN_FIELD(__u32, ttwu_move_balance,
+	     "try_to_wake_up() started passive balancing", "%11u", false, v15);
+#endif /* DOMAIN_FIELD */
diff --git a/tools/lib/perf/include/perf/schedstat-v16.h b/tools/lib/perf/include/perf/schedstat-v16.h
new file mode 100644
index 000000000..3462b79c2
--- /dev/null
+++ b/tools/lib/perf/include/perf/schedstat-v16.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifdef CPU_FIELD
+CPU_FIELD(__u32, yld_count, "sched_yield() count",
+	  "%11u", false, yld_count, v16);
+CPU_FIELD(__u32, array_exp, "Legacy counter can be ignored",
+	  "%11u", false, array_exp, v16);
+CPU_FIELD(__u32, sched_count, "schedule() called",
+	  "%11u", false, sched_count, v16);
+CPU_FIELD(__u32, sched_goidle, "schedule() left the processor idle",
+	  "%11u", true, sched_count, v16);
+CPU_FIELD(__u32, ttwu_count, "try_to_wake_up() was called",
+	  "%11u", false, ttwu_count, v16);
+CPU_FIELD(__u32, ttwu_local, "try_to_wake_up() was called to wake up the local cpu",
+	  "%11u", true, ttwu_count, v16);
+CPU_FIELD(__u64, rq_cpu_time, "total runtime by tasks on this processor (in jiffies)",
+	  "%11llu", false, rq_cpu_time, v16);
+CPU_FIELD(__u64, run_delay, "total waittime by tasks on this processor (in jiffies)",
+	  "%11llu", true, rq_cpu_time, v16);
+CPU_FIELD(__u64, pcount, "total timeslices run on this cpu",
+	  "%11llu", false, pcount, v16);
+#endif /* CPU_FIELD */
+
+#ifdef DOMAIN_FIELD
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category busy> ");
+#endif
+DOMAIN_FIELD(__u32, busy_lb_count,
+	     "load_balance() count on cpu busy", "%11u", true, v16);
+DOMAIN_FIELD(__u32, busy_lb_balanced,
+	     "load_balance() found balanced on cpu busy", "%11u", true, v16);
+DOMAIN_FIELD(__u32, busy_lb_failed,
+	     "load_balance() move task failed on cpu busy", "%11u", true, v16);
+DOMAIN_FIELD(__u32, busy_lb_imbalance,
+	     "imbalance sum on cpu busy", "%11u", false, v16);
+DOMAIN_FIELD(__u32, busy_lb_gained,
+	     "pull_task() count on cpu busy", "%11u", false, v16);
+DOMAIN_FIELD(__u32, busy_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu busy", "%11u", false, v16);
+DOMAIN_FIELD(__u32, busy_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu busy", "%11u", true, v16);
+DOMAIN_FIELD(__u32, busy_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu busy", "%11u", true, v16);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(busy_lb_success_count, "load_balance() success count on cpu busy", "%11u",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, v16);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(busy_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu busy)", "%11.2Lf",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, busy_lb_gained, v16);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category idle> ");
+#endif
+DOMAIN_FIELD(__u32, idle_lb_count,
+	     "load_balance() count on cpu idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, idle_lb_balanced,
+	     "load_balance() found balanced on cpu idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, idle_lb_failed,
+	     "load_balance() move task failed on cpu idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, idle_lb_imbalance,
+	     "imbalance sum on cpu idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, idle_lb_gained,
+	     "pull_task() count on cpu idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, idle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, idle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, idle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu idle", "%11u", true, v16);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(idle_lb_success_count, "load_balance() success count on cpu idle", "%11u",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, v16);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(idle_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu idle)", "%11.2Lf",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, idle_lb_gained, v16);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category newidle> ");
+#endif
+DOMAIN_FIELD(__u32, newidle_lb_count,
+	     "load_balance() count on cpu newly idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, newidle_lb_balanced,
+	     "load_balance() found balanced on cpu newly idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, newidle_lb_failed,
+	     "load_balance() move task failed on cpu newly idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance,
+	     "imbalance sum on cpu newly idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, newidle_lb_gained,
+	     "pull_task() count on cpu newly idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, newidle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu newly idle", "%11u", false, v16);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu newly idle", "%11u", true, v16);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu newly idle", "%11u", true, v16);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(newidle_lb_success_count,
+		  "load_balance() success count on cpu newly idle", "%11u",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, v16);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(newidle_lb_avg_count,
+		  "avg task pulled per successful lb attempt (cpu newly idle)", "%11.2Lf",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, newidle_lb_gained, v16);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category active_load_balance()> ");
+#endif
+DOMAIN_FIELD(__u32, alb_count,
+	     "active_load_balance() count", "%11u", false, v16);
+DOMAIN_FIELD(__u32, alb_failed,
+	     "active_load_balance() move task failed", "%11u", false, v16);
+DOMAIN_FIELD(__u32, alb_pushed,
+	     "active_load_balance() successfully moved a task", "%11u", false, v16);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_exec()> ");
+#endif
+DOMAIN_FIELD(__u32, sbe_count,
+	     "sbe_count is not used", "%11u", false, v16);
+DOMAIN_FIELD(__u32, sbe_balanced,
+	     "sbe_balanced is not used", "%11u", false, v16);
+DOMAIN_FIELD(__u32, sbe_pushed,
+	     "sbe_pushed is not used", "%11u", false, v16);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_fork()> ");
+#endif
+DOMAIN_FIELD(__u32, sbf_count,
+	     "sbf_count is not used", "%11u", false, v16);
+DOMAIN_FIELD(__u32, sbf_balanced,
+	     "sbf_balanced is not used", "%11u", false, v16);
+DOMAIN_FIELD(__u32, sbf_pushed,
+	     "sbf_pushed is not used", "%11u", false, v16);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Wakeup Info> ");
+#endif
+DOMAIN_FIELD(__u32, ttwu_wake_remote,
+	     "try_to_wake_up() awoke a task that last ran on a diff cpu", "%11u", false, v16);
+DOMAIN_FIELD(__u32, ttwu_move_affine,
+	     "try_to_wake_up() moved task because cache-cold on own cpu", "%11u", false, v16);
+DOMAIN_FIELD(__u32, ttwu_move_balance,
+	     "try_to_wake_up() started passive balancing", "%11u", false, v16);
+#endif /* DOMAIN_FIELD */
diff --git a/tools/lib/perf/include/perf/schedstat-v17.h b/tools/lib/perf/include/perf/schedstat-v17.h
new file mode 100644
index 000000000..865dc7c10
--- /dev/null
+++ b/tools/lib/perf/include/perf/schedstat-v17.h
@@ -0,0 +1,164 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifdef CPU_FIELD
+CPU_FIELD(__u32, yld_count, "sched_yield() count",
+	  "%11u", false, yld_count, v17);
+CPU_FIELD(__u32, array_exp, "Legacy counter can be ignored",
+	  "%11u", false, array_exp, v17);
+CPU_FIELD(__u32, sched_count, "schedule() called",
+	  "%11u", false, sched_count, v17);
+CPU_FIELD(__u32, sched_goidle, "schedule() left the processor idle",
+	  "%11u", true, sched_count, v17);
+CPU_FIELD(__u32, ttwu_count, "try_to_wake_up() was called",
+	  "%11u", false, ttwu_count, v17);
+CPU_FIELD(__u32, ttwu_local, "try_to_wake_up() was called to wake up the local cpu",
+	  "%11u", true, ttwu_count, v17);
+CPU_FIELD(__u64, rq_cpu_time, "total runtime by tasks on this processor (in jiffies)",
+	  "%11llu", false, rq_cpu_time, v17);
+CPU_FIELD(__u64, run_delay, "total waittime by tasks on this processor (in jiffies)",
+	  "%11llu", true, rq_cpu_time, v17);
+CPU_FIELD(__u64, pcount, "total timeslices run on this cpu",
+	  "%11llu", false, pcount, v17);
+#endif /* CPU_FIELD */
+
+#ifdef DOMAIN_FIELD
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category busy> ");
+#endif
+DOMAIN_FIELD(__u32, busy_lb_count,
+	     "load_balance() count on cpu busy", "%11u", true, v17);
+DOMAIN_FIELD(__u32, busy_lb_balanced,
+	     "load_balance() found balanced on cpu busy", "%11u", true, v17);
+DOMAIN_FIELD(__u32, busy_lb_failed,
+	     "load_balance() move task failed on cpu busy", "%11u", true, v17);
+DOMAIN_FIELD(__u32, busy_lb_imbalance_load,
+	     "imbalance in load on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_imbalance_util,
+	     "imbalance in utilization on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_imbalance_task,
+	     "imbalance in number of tasks on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_imbalance_misfit,
+	     "imbalance in misfit tasks on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_gained,
+	     "pull_task() count on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu busy", "%11u", false, v17);
+DOMAIN_FIELD(__u32, busy_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu busy", "%11u", true, v17);
+DOMAIN_FIELD(__u32, busy_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu busy", "%11u", true, v17);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(busy_lb_success_count, "load_balance() success count on cpu busy", "%11u",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, v17);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(busy_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu busy)", "%11.2Lf",
+		  busy_lb_count, busy_lb_balanced, busy_lb_failed, busy_lb_gained, v17);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category idle> ");
+#endif
+DOMAIN_FIELD(__u32, idle_lb_count,
+	     "load_balance() count on cpu idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, idle_lb_balanced,
+	     "load_balance() found balanced on cpu idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, idle_lb_failed,
+	     "load_balance() move task failed on cpu idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, idle_lb_imbalance_load,
+	     "imbalance in load on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_imbalance_util,
+	     "imbalance in utilization on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_imbalance_task,
+	     "imbalance in number of tasks on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_imbalance_misfit,
+	     "imbalance in misfit tasks on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_gained,
+	     "pull_task() count on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, idle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, idle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu idle", "%11u", true, v17);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(idle_lb_success_count, "load_balance() success count on cpu idle", "%11u",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, v17);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(idle_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu idle)", "%11.2Lf",
+		  idle_lb_count, idle_lb_balanced, idle_lb_failed, idle_lb_gained, v17);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category newidle> ");
+#endif
+DOMAIN_FIELD(__u32, newidle_lb_count,
+	     "load_balance() count on cpu newly idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, newidle_lb_balanced,
+	     "load_balance() found balanced on cpu newly idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, newidle_lb_failed,
+	     "load_balance() move task failed on cpu newly idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance_load,
+	     "imbalance in load on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance_util,
+	     "imbalance in utilization on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance_task,
+	     "imbalance in number of tasks on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_imbalance_misfit,
+	     "imbalance in misfit tasks on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_gained,
+	     "pull_task() count on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_hot_gained,
+	     "pull_task() when target task was cache-hot on cpu newly idle", "%11u", false, v17);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyq,
+	     "load_balance() failed to find busier queue on cpu newly idle", "%11u", true, v17);
+DOMAIN_FIELD(__u32, newidle_lb_nobusyg,
+	     "load_balance() failed to find busier group on cpu newly idle", "%11u", true, v17);
+#ifdef DERIVED_CNT_FIELD
+DERIVED_CNT_FIELD(newidle_lb_success_count,
+		  "load_balance() success count on cpu newly idle", "%11u",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, v17);
+#endif
+#ifdef DERIVED_AVG_FIELD
+DERIVED_AVG_FIELD(newidle_lb_avg_pulled,
+		  "avg task pulled per successful lb attempt (cpu newly idle)", "%11.2Lf",
+		  newidle_lb_count, newidle_lb_balanced, newidle_lb_failed, newidle_lb_gained, v17);
+#endif
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category active_load_balance()> ");
+#endif
+DOMAIN_FIELD(__u32, alb_count,
+	     "active_load_balance() count", "%11u", false, v17);
+DOMAIN_FIELD(__u32, alb_failed,
+	     "active_load_balance() move task failed", "%11u", false, v17);
+DOMAIN_FIELD(__u32, alb_pushed,
+	     "active_load_balance() successfully moved a task", "%11u", false, v17);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_exec()> ");
+#endif
+DOMAIN_FIELD(__u32, sbe_count,
+	     "sbe_count is not used", "%11u", false, v17);
+DOMAIN_FIELD(__u32, sbe_balanced,
+	     "sbe_balanced is not used", "%11u", false, v17);
+DOMAIN_FIELD(__u32, sbe_pushed,
+	     "sbe_pushed is not used", "%11u", false, v17);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Category sched_balance_fork()> ");
+#endif
+DOMAIN_FIELD(__u32, sbf_count,
+	     "sbf_count is not used", "%11u", false, v17);
+DOMAIN_FIELD(__u32, sbf_balanced,
+	     "sbf_balanced is not used", "%11u", false, v17);
+DOMAIN_FIELD(__u32, sbf_pushed,
+	     "sbf_pushed is not used", "%11u", false, v17);
+#ifdef DOMAIN_CATEGORY
+DOMAIN_CATEGORY(" <Wakeup Info> ");
+#endif
+DOMAIN_FIELD(__u32, ttwu_wake_remote,
+	     "try_to_wake_up() awoke a task that last ran on a diff cpu", "%11u", false, v17);
+DOMAIN_FIELD(__u32, ttwu_move_affine,
+	     "try_to_wake_up() moved task because cache-cold on own cpu", "%11u", false, v17);
+DOMAIN_FIELD(__u32, ttwu_move_balance,
+	     "try_to_wake_up() started passive balancing", "%11u", false, v17);
+#endif /* DOMAIN_FIELD */
diff --git a/tools/lib/perf/include/perf/threadmap.h b/tools/lib/perf/include/perf/threadmap.h
new file mode 100644
index 000000000..44deb815b
--- /dev/null
+++ b/tools/lib/perf/include/perf/threadmap.h
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBPERF_THREADMAP_H
+#define __LIBPERF_THREADMAP_H
+
+#include <perf/core.h>
+#include <sys/types.h>
+
+struct perf_thread_map;
+
+LIBPERF_API struct perf_thread_map *perf_thread_map__new_dummy(void);
+LIBPERF_API struct perf_thread_map *perf_thread_map__new_array(int nr_threads, pid_t *array);
+
+LIBPERF_API void perf_thread_map__set_pid(struct perf_thread_map *map, int idx, pid_t pid);
+LIBPERF_API char *perf_thread_map__comm(struct perf_thread_map *map, int idx);
+LIBPERF_API int perf_thread_map__nr(struct perf_thread_map *threads);
+LIBPERF_API pid_t perf_thread_map__pid(struct perf_thread_map *map, int idx);
+LIBPERF_API int perf_thread_map__idx(struct perf_thread_map *map, pid_t pid);
+
+LIBPERF_API struct perf_thread_map *perf_thread_map__get(struct perf_thread_map *map);
+LIBPERF_API void perf_thread_map__put(struct perf_thread_map *map);
+
+#endif /* __LIBPERF_THREADMAP_H */