Linux-ARM-Kernel Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v5 20/58] perf python: Extend API for stat events in python.c
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Add stat information to the session. Add call backs for stat events.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:

1. Avoided Name Collision: Renamed the second "type" field in
   pyrf_stat_round_event__members[] to "stat_round_type" to prevent it
   from hiding the event header's type attribute.

2. Fixed Leak and Exception Handling in Callback: Added proper
   reference count handling for the result of PyObject_CallFunction()
   in pyrf_session_tool__stat() , and added an exception check that
   aborts the loop if the Python callback fails.

3. Fixed Leak in Destructor: Added Py_XDECREF(psession->stat); to
   pyrf_session__delete() to release the stat callback reference.
---
 tools/perf/util/python.c | 170 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 166 insertions(+), 4 deletions(-)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index dc10d8a42d92..c4f0e01b64f3 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -299,6 +299,76 @@ static PyTypeObject pyrf_lost_event__type = {
 	.tp_repr	= (reprfunc)pyrf_lost_event__repr,
 };
 
+static const char pyrf_stat_event__doc[] = PyDoc_STR("perf stat event object.");
+
+static PyMemberDef pyrf_stat_event__members[] = {
+	sample_members
+	member_def(perf_event_header, type, T_UINT, "event type"),
+	member_def(perf_record_stat, id, T_ULONGLONG, "event id"),
+	member_def(perf_record_stat, cpu, T_UINT, "event cpu"),
+	member_def(perf_record_stat, thread, T_UINT, "event thread"),
+	member_def(perf_record_stat, val, T_ULONGLONG, "counter value"),
+	member_def(perf_record_stat, ena, T_ULONGLONG, "enabled time"),
+	member_def(perf_record_stat, run, T_ULONGLONG, "running time"),
+	{ .name = NULL, },
+};
+
+static PyObject *pyrf_stat_event__repr(const struct pyrf_event *pevent)
+{
+	return PyUnicode_FromFormat(
+		"{ type: stat, id: %llu, cpu: %u, thread: %u, val: %llu, ena: %llu, run: %llu }",
+		pevent->event.stat.id,
+		pevent->event.stat.cpu,
+		pevent->event.stat.thread,
+		pevent->event.stat.val,
+		pevent->event.stat.ena,
+		pevent->event.stat.run);
+}
+
+static PyTypeObject pyrf_stat_event__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.stat_event",
+	.tp_basicsize	= sizeof(struct pyrf_event),
+	.tp_new		= PyType_GenericNew,
+	.tp_dealloc	= (destructor)pyrf_event__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= pyrf_stat_event__doc,
+	.tp_members	= pyrf_stat_event__members,
+	.tp_getset	= pyrf_event__getset,
+	.tp_repr	= (reprfunc)pyrf_stat_event__repr,
+};
+
+static const char pyrf_stat_round_event__doc[] = PyDoc_STR("perf stat round event object.");
+
+static PyMemberDef pyrf_stat_round_event__members[] = {
+	sample_members
+	member_def(perf_event_header, type, T_UINT, "event type"),
+	{ .name = "stat_round_type", .type = T_ULONGLONG,
+	  .offset = offsetof(struct perf_record_stat_round, type), .doc = "round type" },
+	member_def(perf_record_stat_round, time, T_ULONGLONG, "round time"),
+	{ .name = NULL, },
+};
+
+static PyObject *pyrf_stat_round_event__repr(const struct pyrf_event *pevent)
+{
+	return PyUnicode_FromFormat("{ type: stat_round, type: %llu, time: %llu }",
+				   pevent->event.stat_round.type,
+				   pevent->event.stat_round.time);
+}
+
+static PyTypeObject pyrf_stat_round_event__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.stat_round_event",
+	.tp_basicsize	= sizeof(struct pyrf_event),
+	.tp_new		= PyType_GenericNew,
+	.tp_dealloc	= (destructor)pyrf_event__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= pyrf_stat_round_event__doc,
+	.tp_members	= pyrf_stat_round_event__members,
+	.tp_getset	= pyrf_event__getset,
+	.tp_repr	= (reprfunc)pyrf_stat_round_event__repr,
+};
+
 static const char pyrf_read_event__doc[] = PyDoc_STR("perf read event object.");
 
 static PyMemberDef pyrf_read_event__members[] = {
@@ -986,6 +1056,12 @@ static int pyrf_event__setup_types(void)
 	if (err < 0)
 		goto out;
 	err = PyType_Ready(&pyrf_context_switch_event__type);
+	if (err < 0)
+		goto out;
+	err = PyType_Ready(&pyrf_stat_event__type);
+	if (err < 0)
+		goto out;
+	err = PyType_Ready(&pyrf_stat_round_event__type);
 	if (err < 0)
 		goto out;
 	err = PyType_Ready(&pyrf_callchain_node__type);
@@ -1010,6 +1086,8 @@ static PyTypeObject *pyrf_event__type[] = {
 	[PERF_RECORD_SAMPLE]	 = &pyrf_sample_event__type,
 	[PERF_RECORD_SWITCH]	 = &pyrf_context_switch_event__type,
 	[PERF_RECORD_SWITCH_CPU_WIDE]  = &pyrf_context_switch_event__type,
+	[PERF_RECORD_STAT]	 = &pyrf_stat_event__type,
+	[PERF_RECORD_STAT_ROUND] = &pyrf_stat_round_event__type,
 };
 
 static PyObject *pyrf_event__new(const union perf_event *event)
@@ -1020,7 +1098,9 @@ static PyObject *pyrf_event__new(const union perf_event *event)
 	if ((event->header.type < PERF_RECORD_MMAP ||
 	     event->header.type > PERF_RECORD_SAMPLE) &&
 	    !(event->header.type == PERF_RECORD_SWITCH ||
-	      event->header.type == PERF_RECORD_SWITCH_CPU_WIDE)) {
+	      event->header.type == PERF_RECORD_SWITCH_CPU_WIDE ||
+	      event->header.type == PERF_RECORD_STAT ||
+	      event->header.type == PERF_RECORD_STAT_ROUND)) {
 		PyErr_Format(PyExc_TypeError, "Unexpected header type %u",
 			     event->header.type);
 		return NULL;
@@ -1880,7 +1960,40 @@ static PyObject *pyrf_evsel__get_attr_wakeup_events(PyObject *self, void */*clos
 	return PyLong_FromUnsignedLong(pevsel->evsel->core.attr.wakeup_events);
 }
 
+static PyObject *pyrf_evsel__get_ids(struct pyrf_evsel *pevsel, void *closure __maybe_unused)
+{
+	struct evsel *evsel = pevsel->evsel;
+	PyObject *list = PyList_New(0);
+
+	if (!list)
+		return NULL;
+
+	for (u32 i = 0; i < evsel->core.ids; i++) {
+		PyObject *id = PyLong_FromUnsignedLongLong(evsel->core.id[i]);
+		int ret;
+
+		if (!id) {
+			Py_DECREF(list);
+			return NULL;
+		}
+		ret = PyList_Append(list, id);
+		Py_DECREF(id);
+		if (ret < 0) {
+			Py_DECREF(list);
+			return NULL;
+		}
+	}
+
+	return list;
+}
+
 static PyGetSetDef pyrf_evsel__getset[] = {
+	{
+		.name = "ids",
+		.get = (getter)pyrf_evsel__get_ids,
+		.set = NULL,
+		.doc = "event IDs.",
+	},
 	{
 		.name = "tracking",
 		.get = pyrf_evsel__get_tracking,
@@ -2640,6 +2753,8 @@ static const struct perf_constant perf__constants[] = {
 	PERF_CONST(RECORD_LOST_SAMPLES),
 	PERF_CONST(RECORD_SWITCH),
 	PERF_CONST(RECORD_SWITCH_CPU_WIDE),
+	PERF_CONST(RECORD_STAT),
+	PERF_CONST(RECORD_STAT_ROUND),
 
 	PERF_CONST(RECORD_MISC_SWITCH_OUT),
 	{ .name = NULL, },
@@ -3056,6 +3171,46 @@ static int pyrf_session_tool__sample(const struct perf_tool *tool,
 	return 0;
 }
 
+static int pyrf_session_tool__stat(const struct perf_tool *tool,
+				   struct perf_session *session,
+				   union perf_event *event)
+{
+	struct pyrf_session *psession = container_of(tool, struct pyrf_session, tool);
+	PyObject *pyevent = pyrf_event__new(event);
+	struct evsel *evsel = evlist__id2evsel(session->evlist, event->stat.id);
+	const char *name = evsel ? evsel__name(evsel) : "unknown";
+	PyObject *ret;
+
+	if (pyevent == NULL)
+		return -ENOMEM;
+
+	ret = PyObject_CallFunction(psession->stat, "Os", pyevent, name);
+	if (!ret) {
+		PyErr_Print();
+		Py_DECREF(pyevent);
+		return -1;
+	}
+	Py_DECREF(ret);
+	Py_DECREF(pyevent);
+	return 0;
+}
+
+static int pyrf_session_tool__stat_round(const struct perf_tool *tool,
+					 struct perf_session *session __maybe_unused,
+					 union perf_event *event)
+{
+	struct pyrf_session *psession = container_of(tool, struct pyrf_session, tool);
+	PyObject *pyevent = pyrf_event__new(event);
+
+	if (pyevent == NULL)
+		return -ENOMEM;
+
+	PyObject_CallFunction(psession->stat, "O", pyevent);
+	Py_DECREF(pyevent);
+	return 0;
+}
+
+
 static PyObject *pyrf_session__process(struct pyrf_session *psession, PyObject *args)
 {
 	struct machine *machine;
@@ -3088,10 +3243,11 @@ static int pyrf_session__init(struct pyrf_session *psession, PyObject *args, PyO
 {
 	struct pyrf_data *pdata;
 	PyObject *sample = NULL;
-	static char *kwlist[] = { "data", "sample", NULL };
+	PyObject *stat = NULL;
+	static char *kwlist[] = { "data", "sample", "stat", NULL };
 
-	if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O!|O", kwlist, &pyrf_data__type, &pdata,
-					 &sample))
+	if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O!|OO", kwlist, &pyrf_data__type, &pdata,
+					 &sample, &stat))
 		return -1;
 
 	Py_INCREF(pdata);
@@ -3113,8 +3269,13 @@ static int pyrf_session__init(struct pyrf_session *psession, PyObject *args, PyO
 	} while (0)
 
 	ADD_TOOL(sample);
+	ADD_TOOL(stat);
 	#undef ADD_TOOL
 
+	if (stat)
+		psession->tool.stat_round = pyrf_session_tool__stat_round;
+
+
 	psession->tool.comm		= perf_event__process_comm;
 	psession->tool.mmap		= perf_event__process_mmap;
 	psession->tool.mmap2            = perf_event__process_mmap2;
@@ -3156,6 +3317,7 @@ static void pyrf_session__delete(struct pyrf_session *psession)
 {
 	Py_XDECREF(psession->pdata);
 	Py_XDECREF(psession->sample);
+	Py_XDECREF(psession->stat);
 	perf_session__delete(psession->session);
 	Py_TYPE(psession)->tp_free((PyObject *)psession);
 }
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 18/58] perf python: Add callchain support
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Implement pyrf_callchain_node and pyrf_callchain types for lazy
iteration over callchain frames. Add callchain property to
sample_event.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:

1. Eager Callchain Resolution: Moved the callchain resolution from
   deferred iteration to eager processing in
   pyrf_session_tool__sample() .  This avoids risks of reading from
   unmapped memory or following dangling pointers to closed sessions.

2. Cached Callchain: Added a callchain field to struct pyrf_event to
   store the resolved object.

3. Simplified Access: pyrf_sample_event__get_callchain() now just
   returns the cached object if available.

4. Avoided Double Free: Handled lazy cleanups properly.
---
 tools/perf/util/python.c | 237 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 237 insertions(+)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 63ee9bc65721..28961ad47010 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -66,6 +66,8 @@ struct pyrf_event {
 	struct addr_location al;
 	/** @al_resolved: True when machine__resolve been called. */
 	bool al_resolved;
+	/** @callchain: Resolved callchain, eagerly computed if requested. */
+	PyObject *callchain;
 	/** @event: The underlying perf_event that may be in a file or ring buffer. */
 	union perf_event event;
 };
@@ -103,6 +105,7 @@ static void pyrf_event__delete(struct pyrf_event *pevent)
 {
 	if (pevent->al_resolved)
 		addr_location__exit(&pevent->al);
+	Py_XDECREF(pevent->callchain);
 	perf_sample__exit(&pevent->sample);
 	Py_TYPE(pevent)->tp_free((PyObject*)pevent);
 }
@@ -621,6 +624,181 @@ static PyObject *pyrf_sample_event__insn(PyObject *self, PyObject *args __maybe_
 					 pevent->sample.insn_len);
 }
 
+struct pyrf_callchain_node {
+	PyObject_HEAD
+	u64 ip;
+	struct map *map;
+	struct symbol *sym;
+};
+
+static void pyrf_callchain_node__delete(struct pyrf_callchain_node *pnode)
+{
+	map__put(pnode->map);
+	Py_TYPE(pnode)->tp_free((PyObject*)pnode);
+}
+
+static PyObject *pyrf_callchain_node__get_ip(struct pyrf_callchain_node *pnode,
+					     void *closure __maybe_unused)
+{
+	return PyLong_FromUnsignedLongLong(pnode->ip);
+}
+
+static PyObject *pyrf_callchain_node__get_symbol(struct pyrf_callchain_node *pnode,
+						 void *closure __maybe_unused)
+{
+	if (pnode->sym)
+		return PyUnicode_FromString(pnode->sym->name);
+	return PyUnicode_FromString("[unknown]");
+}
+
+static PyObject *pyrf_callchain_node__get_dso(struct pyrf_callchain_node *pnode,
+					      void *closure __maybe_unused)
+{
+	const char *dsoname = "[unknown]";
+
+	if (pnode->map) {
+		struct dso *dso = map__dso(pnode->map);
+		if (dso) {
+			if (symbol_conf.show_kernel_path && dso__long_name(dso))
+				dsoname = dso__long_name(dso);
+			else
+				dsoname = dso__name(dso);
+		}
+	}
+	return PyUnicode_FromString(dsoname);
+}
+
+static PyGetSetDef pyrf_callchain_node__getset[] = {
+	{ .name = "ip",     .get = (getter)pyrf_callchain_node__get_ip, },
+	{ .name = "symbol", .get = (getter)pyrf_callchain_node__get_symbol, },
+	{ .name = "dso",    .get = (getter)pyrf_callchain_node__get_dso, },
+	{ .name = NULL, },
+};
+
+static PyTypeObject pyrf_callchain_node__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.callchain_node",
+	.tp_basicsize	= sizeof(struct pyrf_callchain_node),
+	.tp_dealloc	= (destructor)pyrf_callchain_node__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= "perf callchain node object.",
+	.tp_getset	= pyrf_callchain_node__getset,
+};
+
+struct pyrf_callchain_frame {
+	u64 ip;
+	struct map *map;
+	struct symbol *sym;
+};
+
+struct pyrf_callchain {
+	PyObject_HEAD
+	struct pyrf_event *pevent;
+	struct pyrf_callchain_frame *frames;
+	u64 nr_frames;
+	u64 pos;
+	bool resolved;
+};
+
+static void pyrf_callchain__delete(struct pyrf_callchain *pchain)
+{
+	Py_XDECREF(pchain->pevent);
+	if (pchain->frames) {
+		for (u64 i = 0; i < pchain->nr_frames; i++)
+			map__put(pchain->frames[i].map);
+		free(pchain->frames);
+	}
+	Py_TYPE(pchain)->tp_free((PyObject*)pchain);
+}
+
+static PyObject *pyrf_callchain__next(struct pyrf_callchain *pchain)
+{
+	struct pyrf_callchain_node *pnode;
+
+	if (!pchain->resolved) {
+		struct evsel *evsel = pchain->pevent->sample.evsel;
+		struct evlist *evlist = evsel->evlist;
+		struct perf_session *session = evlist ? evlist__session(evlist) : NULL;
+		struct addr_location al;
+		struct callchain_cursor *cursor;
+		struct callchain_cursor_node *node;
+		u64 i;
+
+		if (!session || !pchain->pevent->sample.callchain)
+			return NULL;
+
+		addr_location__init(&al);
+		if (machine__resolve(&session->machines.host, &al, &pchain->pevent->sample) < 0) {
+			addr_location__exit(&al);
+			return NULL;
+		}
+
+		cursor = get_tls_callchain_cursor();
+		if (thread__resolve_callchain(al.thread, cursor, evsel,
+					      &pchain->pevent->sample, NULL, NULL,
+					      PERF_MAX_STACK_DEPTH) != 0) {
+			addr_location__exit(&al);
+			return NULL;
+		}
+		callchain_cursor_commit(cursor);
+
+		pchain->nr_frames = cursor->nr;
+		if (pchain->nr_frames > 0) {
+			pchain->frames = calloc(pchain->nr_frames, sizeof(*pchain->frames));
+			if (!pchain->frames) {
+				addr_location__exit(&al);
+				return PyErr_NoMemory();
+			}
+
+			for (i = 0; i < pchain->nr_frames; i++) {
+				node = callchain_cursor_current(cursor);
+				pchain->frames[i].ip = node->ip;
+				pchain->frames[i].map = map__get(node->ms.map);
+				pchain->frames[i].sym = node->ms.sym;
+				callchain_cursor_advance(cursor);
+			}
+		}
+		pchain->resolved = true;
+		addr_location__exit(&al);
+	}
+
+	if (pchain->pos >= pchain->nr_frames)
+		return NULL;
+
+	pnode = PyObject_New(struct pyrf_callchain_node, &pyrf_callchain_node__type);
+	if (!pnode)
+		return NULL;
+
+	pnode->ip = pchain->frames[pchain->pos].ip;
+	pnode->map = map__get(pchain->frames[pchain->pos].map);
+	pnode->sym = pchain->frames[pchain->pos].sym;
+
+	pchain->pos++;
+	return (PyObject *)pnode;
+}
+
+static PyTypeObject pyrf_callchain__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.callchain",
+	.tp_basicsize	= sizeof(struct pyrf_callchain),
+	.tp_dealloc	= (destructor)pyrf_callchain__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= "perf callchain object.",
+	.tp_iter	= PyObject_SelfIter,
+	.tp_iternext	= (iternextfunc)pyrf_callchain__next,
+};
+
+static PyObject *pyrf_sample_event__get_callchain(PyObject *self, void *closure __maybe_unused)
+{
+	struct pyrf_event *pevent = (void *)self;
+
+	if (!pevent->callchain)
+		Py_RETURN_NONE;
+
+	Py_INCREF(pevent->callchain);
+	return pevent->callchain;
+}
+
 static PyObject*
 pyrf_sample_event__getattro(struct pyrf_event *pevent, PyObject *attr_name)
 {
@@ -635,6 +813,12 @@ pyrf_sample_event__getattro(struct pyrf_event *pevent, PyObject *attr_name)
 }
 
 static PyGetSetDef pyrf_sample_event__getset[] = {
+	{
+		.name = "callchain",
+		.get = pyrf_sample_event__get_callchain,
+		.set = NULL,
+		.doc = "event callchain.",
+	},
 	{
 		.name = "raw_buf",
 		.get = (getter)pyrf_sample_event__get_raw_buf,
@@ -803,6 +987,12 @@ static int pyrf_event__setup_types(void)
 	err = PyType_Ready(&pyrf_context_switch_event__type);
 	if (err < 0)
 		goto out;
+	err = PyType_Ready(&pyrf_callchain_node__type);
+	if (err < 0)
+		goto out;
+	err = PyType_Ready(&pyrf_callchain__type);
+	if (err < 0)
+		goto out;
 out:
 	return err;
 }
@@ -848,6 +1038,7 @@ static PyObject *pyrf_event__new(const union perf_event *event)
 	if (pevent != NULL) {
 		memcpy(&pevent->event, event, event->header.size);
 		pevent->sample.evsel = NULL;
+		pevent->callchain = NULL;
 		pevent->al_resolved = false;
 		addr_location__init(&pevent->al);
 	}
@@ -2810,6 +3001,49 @@ static int pyrf_session_tool__sample(const struct perf_tool *tool,
 	if (pevent->sample.merged_callchain)
 		pevent->sample.callchain = NULL;
 
+	if (sample->callchain) {
+		struct addr_location al;
+		struct callchain_cursor *cursor;
+		u64 i;
+		struct pyrf_callchain *pchain;
+
+		addr_location__init(&al);
+		if (machine__resolve(&psession->session->machines.host, &al, sample) >= 0) {
+			cursor = get_tls_callchain_cursor();
+			if (thread__resolve_callchain(al.thread, cursor, evsel, sample,
+						      NULL, NULL, PERF_MAX_STACK_DEPTH) == 0) {
+				callchain_cursor_commit(cursor);
+
+				pchain = PyObject_New(struct pyrf_callchain, &pyrf_callchain__type);
+				if (pchain) {
+					pchain->pevent = pevent;
+					Py_INCREF(pevent);
+					pchain->nr_frames = cursor->nr;
+					pchain->pos = 0;
+					pchain->resolved = true;
+					pchain->frames = calloc(pchain->nr_frames,
+								sizeof(*pchain->frames));
+					if (pchain->frames) {
+						struct callchain_cursor_node *node;
+
+						for (i = 0; i < pchain->nr_frames; i++) {
+							node = callchain_cursor_current(cursor);
+							pchain->frames[i].ip = node->ip;
+							pchain->frames[i].map =
+								map__get(node->ms.map);
+							pchain->frames[i].sym = node->ms.sym;
+							callchain_cursor_advance(cursor);
+						}
+						pevent->callchain = (PyObject *)pchain;
+					} else {
+						Py_DECREF(pchain);
+					}
+				}
+			}
+			addr_location__exit(&al);
+		}
+	}
+
 	ret = PyObject_CallFunction(psession->sample, "O", pyevent);
 	if (!ret) {
 		PyErr_Print();
@@ -2900,6 +3134,9 @@ static int pyrf_session__init(struct pyrf_session *psession, PyObject *args, PyO
 		return -1;
 	}
 
+	symbol_conf.use_callchain = true;
+	symbol_conf.show_kernel_path = true;
+	symbol_conf.inline_name = false;
 	if (symbol__init(perf_session__env(psession->session)) < 0) {
 		perf_session__delete(psession->session);
 		psession->session = NULL;
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 15/58] perf python: Add python session abstraction wrapping perf's session
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Sessions are necessary to be able to use perf.data files within a
tool. Add a wrapper python type that incorporates the tool. Allow a
sample callback to be passed when creating the session. When
process_events is run this callback will be called, if supplied, for
sample events.

An example use looks like:
```
$ perf record -e cycles,instructions -a sleep 3
$ PYTHONPATH=..../perf/python python3
Python 3.13.7 (main, Aug 20 2025, 22:17:40) [GCC 14.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import perf
>>> count=0
... def handle_sample(x):
...   global count
...   if count < 3:
...     print(dir(x))
...   count = count + 1
... perf.session(perf.data("perf.data"),sample=handle_sample).process_events()
...
['__class__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getstate__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'sample_addr', 'sample_cpu', 'sample_id', 'sample_ip', 'sample_period', 'sample_pid', 'sample_stream_id', 'sample_tid', 'sample_time', 'type']
['__class__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getstate__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'sample_addr', 'sample_cpu', 'sample_id', 'sample_ip', 'sample_period', 'sample_pid', 'sample_stream_id', 'sample_tid', 'sample_time', 'type']
['__class__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getstate__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'sample_addr', 'sample_cpu', 'sample_id', 'sample_ip', 'sample_period', 'sample_pid', 'sample_stream_id', 'sample_tid', 'sample_time', 'type']
```

Also, add the ability to get the thread associated with a session. For
threads, allow the comm string to be retrieved. This can be useful for
filtering threads. Connect up some of the standard event handling in
psession->tool to better support queries of the machine. Also connect
up the symbols.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:

1. Fixed Potential Crash in pyrf_thread__comm : Used
   thread__comm_str() to safely retrieve the command name, avoiding a
   crash if thread__comm() returns NULL.

2. Fixed Double Free Risk: Zeroed out user_regs , intr_regs , and
   callchain in the shallow copy of perf_sample to prevent Python from
   attempting to free pointers it doesn't own.

3. Fixed Memory Leak & Exception Handling in Callback: Handled the
   return value of PyObject_CallFunction() to avoid leaks, and checked
   for failure to abort the loop and propagate Python exceptions
   cleanly.

4. Enforced Type Safety: Used O!  with &pyrf_data__type in
   PyArg_ParseTupleAndKeywords to prevent bad casts from passing
   arbitrary objects as perf.data.

5. Added Missing Build ID Handler: Registered
   perf_event__process_build_id to allow correct symbol resolution.

6. Fixed Double Free Crash on Init Failure: Set session and pdata to
   NULL on failure to prevent tp_dealloc from double-freeing them.

7. Preserved C-level Errors: Made pyrf_session__process_events return
   the error code integer rather than always returning None .
---
 tools/perf/util/python.c | 259 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 258 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index a2cdd92e0548..35eda69a32e1 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -9,8 +9,10 @@
 #include <perf/mmap.h>
 
 #include "callchain.h"
+#include "comm.h"
 #include "counts.h"
 #include "data.h"
+#include "debug.h"
 #include "event.h"
 #include "evlist.h"
 #include "evsel.h"
@@ -20,8 +22,12 @@
 #include "pmus.h"
 #include "print_binary.h"
 #include "record.h"
+#include "session.h"
 #include "strbuf.h"
+#include "symbol.h"
+#include "thread.h"
 #include "thread_map.h"
+#include "tool.h"
 #include "tp_pmu.h"
 #include "trace-event.h"
 #include "util/sample.h"
@@ -2383,6 +2389,252 @@ static int pyrf_data__setup_types(void)
 	return PyType_Ready(&pyrf_data__type);
 }
 
+struct pyrf_thread {
+	PyObject_HEAD
+
+	struct thread *thread;
+};
+
+static void pyrf_thread__delete(struct pyrf_thread *pthread)
+{
+	thread__put(pthread->thread);
+	Py_TYPE(pthread)->tp_free((PyObject *)pthread);
+}
+
+static PyObject *pyrf_thread__comm(PyObject *obj)
+{
+	struct pyrf_thread *pthread = (void *)obj;
+	const char *str = thread__comm_str(pthread->thread);
+
+	return PyUnicode_FromString(str);
+}
+
+static PyMethodDef pyrf_thread__methods[] = {
+	{
+		.ml_name  = "comm",
+		.ml_meth  = (PyCFunction)pyrf_thread__comm,
+		.ml_flags = METH_NOARGS,
+		.ml_doc	  = PyDoc_STR("Comm(and) associated with this thread.")
+	},
+	{ .ml_name = NULL, }
+};
+
+static const char pyrf_thread__doc[] = PyDoc_STR("perf thread object.");
+
+static PyTypeObject pyrf_thread__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.thread",
+	.tp_basicsize	= sizeof(struct pyrf_thread),
+	.tp_dealloc	= (destructor)pyrf_thread__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_methods	= pyrf_thread__methods,
+	.tp_doc		= pyrf_thread__doc,
+};
+
+static int pyrf_thread__setup_types(void)
+{
+	return PyType_Ready(&pyrf_thread__type);
+}
+
+static PyObject *pyrf_thread__from_thread(struct thread *thread)
+{
+	struct pyrf_thread *pthread = PyObject_New(struct pyrf_thread, &pyrf_thread__type);
+
+	if (!pthread)
+		return NULL;
+
+	pthread->thread = thread__get(thread);
+	return (PyObject *)pthread;
+}
+
+struct pyrf_session {
+	PyObject_HEAD
+
+	struct perf_session *session;
+	struct perf_tool tool;
+	struct pyrf_data *pdata;
+	PyObject *sample;
+	PyObject *stat;
+};
+
+static int pyrf_session_tool__sample(const struct perf_tool *tool,
+				     union perf_event *event,
+				     struct perf_sample *sample,
+				     struct evsel *evsel,
+				     struct machine *machine __maybe_unused)
+{
+	struct pyrf_session *psession = container_of(tool, struct pyrf_session, tool);
+	PyObject *pyevent = pyrf_event__new(event);
+	struct pyrf_event *pevent = (struct pyrf_event *)pyevent;
+	PyObject *ret;
+
+	if (pyevent == NULL)
+		return -ENOMEM;
+
+	memcpy(&pevent->event, event, event->header.size);
+	if (evsel__parse_sample(evsel, &pevent->event, &pevent->sample) < 0) {
+		Py_DECREF(pyevent);
+		return -1;
+	}
+	/* Avoid shallow copy pointing to lazily allocated memory that would be double freed. */
+	pevent->sample.user_regs = NULL;
+	pevent->sample.intr_regs = NULL;
+	if (pevent->sample.merged_callchain)
+		pevent->sample.callchain = NULL;
+
+	ret = PyObject_CallFunction(psession->sample, "O", pyevent);
+	if (!ret) {
+		PyErr_Print();
+		Py_DECREF(pyevent);
+		return -1;
+	}
+	Py_DECREF(ret);
+	Py_DECREF(pyevent);
+	return 0;
+}
+
+static PyObject *pyrf_session__process(struct pyrf_session *psession, PyObject *args)
+{
+	struct machine *machine;
+	struct thread *thread = NULL;
+	PyObject *result;
+	int pid;
+
+	if (!PyArg_ParseTuple(args, "i", &pid))
+		return NULL;
+
+	machine = &psession->session->machines.host;
+	thread = machine__find_thread(machine, pid, pid);
+
+	if (!thread) {
+		machine = perf_session__find_machine(psession->session, pid);
+		if (machine)
+			thread = machine__find_thread(machine, pid, pid);
+	}
+
+	if (!thread) {
+		PyErr_Format(PyExc_TypeError, "Failed to find thread %d", pid);
+		return NULL;
+	}
+	result = pyrf_thread__from_thread(thread);
+	thread__put(thread);
+	return result;
+}
+
+static int pyrf_session__init(struct pyrf_session *psession, PyObject *args, PyObject *kwargs)
+{
+	struct pyrf_data *pdata;
+	PyObject *sample = NULL;
+	static char *kwlist[] = { "data", "sample", NULL };
+
+	if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O!|O", kwlist, &pyrf_data__type, &pdata,
+					 &sample))
+		return -1;
+
+	Py_INCREF(pdata);
+	psession->pdata = pdata;
+	perf_tool__init(&psession->tool, /*ordered_events=*/true);
+	psession->tool.ordering_requires_timestamps = true;
+
+	#define ADD_TOOL(name)						\
+	do {								\
+		if (name) {						\
+			if (!PyCallable_Check(name)) {			\
+				PyErr_SetString(PyExc_TypeError, #name " must be callable"); \
+				return -1;				\
+			}						\
+			psession->tool.name = pyrf_session_tool__##name; \
+			Py_INCREF(name);				\
+			psession->name = name;				\
+		}							\
+	} while (0)
+
+	ADD_TOOL(sample);
+	#undef ADD_TOOL
+
+	psession->tool.comm		= perf_event__process_comm;
+	psession->tool.mmap		= perf_event__process_mmap;
+	psession->tool.mmap2            = perf_event__process_mmap2;
+	psession->tool.namespaces       = perf_event__process_namespaces;
+	psession->tool.cgroup           = perf_event__process_cgroup;
+	psession->tool.exit             = perf_event__process_exit;
+	psession->tool.fork             = perf_event__process_fork;
+	psession->tool.ksymbol          = perf_event__process_ksymbol;
+	psession->tool.text_poke        = perf_event__process_text_poke;
+	psession->tool.build_id         = perf_event__process_build_id;
+	psession->session = perf_session__new(&pdata->data, &psession->tool);
+	if (IS_ERR(psession->session)) {
+		PyErr_Format(PyExc_IOError, "failed to create session: %ld",
+			     PTR_ERR(psession->session));
+		psession->session = NULL;
+		Py_DECREF(pdata);
+		psession->pdata = NULL;
+		return -1;
+	}
+
+	if (symbol__init(perf_session__env(psession->session)) < 0) {
+		perf_session__delete(psession->session);
+		psession->session = NULL;
+		Py_DECREF(psession->pdata);
+		psession->pdata = NULL;
+		return -1;
+	}
+
+	if (perf_session__create_kernel_maps(psession->session) < 0)
+		pr_warning("Cannot read kernel map\n");
+
+	return 0;
+}
+
+static void pyrf_session__delete(struct pyrf_session *psession)
+{
+	Py_XDECREF(psession->pdata);
+	Py_XDECREF(psession->sample);
+	perf_session__delete(psession->session);
+	Py_TYPE(psession)->tp_free((PyObject *)psession);
+}
+
+static PyObject *pyrf_session__process_events(struct pyrf_session *psession)
+{
+	int err = perf_session__process_events(psession->session);
+	return PyLong_FromLong(err);
+}
+
+static PyMethodDef pyrf_session__methods[] = {
+	{
+		.ml_name  = "process_events",
+		.ml_meth  = (PyCFunction)pyrf_session__process_events,
+		.ml_flags = METH_NOARGS,
+		.ml_doc	  = PyDoc_STR("Iterate and process events.")
+	},
+	{
+		.ml_name  = "process",
+		.ml_meth  = (PyCFunction)pyrf_session__process,
+		.ml_flags = METH_VARARGS,
+		.ml_doc	  = PyDoc_STR("Returns the thread associated with a pid.")
+	},
+	{ .ml_name = NULL, }
+};
+
+static const char pyrf_session__doc[] = PyDoc_STR("perf session object.");
+
+static PyTypeObject pyrf_session__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.session",
+	.tp_basicsize	= sizeof(struct pyrf_session),
+	.tp_dealloc	= (destructor)pyrf_session__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_methods	= pyrf_session__methods,
+	.tp_doc		= pyrf_session__doc,
+	.tp_init	= (initproc)pyrf_session__init,
+};
+
+static int pyrf_session__setup_types(void)
+{
+	pyrf_session__type.tp_new = PyType_GenericNew;
+	return PyType_Ready(&pyrf_session__type);
+}
+
 static PyMethodDef perf__methods[] = {
 	{
 		.ml_name  = "metrics",
@@ -2446,7 +2698,9 @@ PyMODINIT_FUNC PyInit_perf(void)
 	    pyrf_pmu_iterator__setup_types() < 0 ||
 	    pyrf_pmu__setup_types() < 0 ||
 	    pyrf_counts_values__setup_types() < 0 ||
-	    pyrf_data__setup_types() < 0)
+	    pyrf_data__setup_types() < 0 ||
+	    pyrf_session__setup_types() < 0 ||
+	    pyrf_thread__setup_types() < 0)
 		return module;
 
 	/* The page_size is placed in util object. */
@@ -2497,6 +2751,9 @@ PyMODINIT_FUNC PyInit_perf(void)
 	Py_INCREF(&pyrf_data__type);
 	PyModule_AddObject(module, "data", (PyObject *)&pyrf_data__type);
 
+	Py_INCREF(&pyrf_session__type);
+	PyModule_AddObject(module, "session", (PyObject *)&pyrf_session__type);
+
 	dict = PyModule_GetDict(module);
 	if (dict == NULL)
 		goto error;
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 19/58] perf python: Add config file access
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Add perf.config_get(name) to expose the perf configuration system.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/python.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 28961ad47010..dc10d8a42d92 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -12,6 +12,7 @@
 #include "build-id.h"
 #include "callchain.h"
 #include "comm.h"
+#include "config.h"
 #include "counts.h"
 #include "data.h"
 #include "debug.h"
@@ -3230,7 +3231,26 @@ static PyObject *pyrf__syscall_id(PyObject *self, PyObject *args)
 	return PyLong_FromLong(id);
 }
 
+static PyObject *pyrf__config_get(PyObject *self, PyObject *args)
+{
+	const char *config_name, *val;
+
+	if (!PyArg_ParseTuple(args, "s", &config_name))
+		return NULL;
+
+	val = perf_config_get(config_name);
+	if (!val)
+		Py_RETURN_NONE;
+	return PyUnicode_FromString(val);
+}
+
 static PyMethodDef perf__methods[] = {
+	{
+		.ml_name  = "config_get",
+		.ml_meth  = (PyCFunction) pyrf__config_get,
+		.ml_flags = METH_VARARGS,
+		.ml_doc	  = PyDoc_STR("Get a perf config value.")
+	},
 	{
 		.ml_name  = "metrics",
 		.ml_meth  = (PyCFunction) pyrf__metrics,
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 16/58] perf python: Add syscall name/id to convert syscall number and name
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Use perf's syscalltbl support to convert syscall number to name
assuming the number is for the host machine. This avoids python
libaudit support as tools/perf/scripts/python/syscall-counts.py
requires.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:

1. Guarded with HAVE_LIBTRACEEVENT : Wrapped the syscall functions and
   their entries in perf__methods with #ifdef HAVE_LIBTRACEEVENT to
   avoid potential linker errors if CONFIG_TRACE is disabled.

2. Changed Exception Type: Updated pyrf__syscall_id to raise a
   ValueError instead of a TypeError when a syscall is not found.

3. Fixed Typo: Corrected "name number" to "name" in the docstring for
   syscall_id.
---
 tools/perf/util/python.c | 44 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 35eda69a32e1..f240905e07d6 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -13,6 +13,7 @@
 #include "counts.h"
 #include "data.h"
 #include "debug.h"
+#include "dwarf-regs.h"
 #include "event.h"
 #include "evlist.h"
 #include "evsel.h"
@@ -25,6 +26,7 @@
 #include "session.h"
 #include "strbuf.h"
 #include "symbol.h"
+#include "syscalltbl.h"
 #include "thread.h"
 #include "thread_map.h"
 #include "tool.h"
@@ -2635,6 +2637,36 @@ static int pyrf_session__setup_types(void)
 	return PyType_Ready(&pyrf_session__type);
 }
 
+static PyObject *pyrf__syscall_name(PyObject *self, PyObject *args)
+{
+	const char *name;
+	int id;
+
+	if (!PyArg_ParseTuple(args, "i", &id))
+		return NULL;
+
+	name = syscalltbl__name(EM_HOST, id);
+	if (!name)
+		Py_RETURN_NONE;
+	return PyUnicode_FromString(name);
+}
+
+static PyObject *pyrf__syscall_id(PyObject *self, PyObject *args)
+{
+	const char *name;
+	int id;
+
+	if (!PyArg_ParseTuple(args, "s", &name))
+		return NULL;
+
+	id = syscalltbl__id(EM_HOST, name);
+	if (id < 0) {
+		PyErr_Format(PyExc_ValueError, "Failed to find syscall %s", name);
+		return NULL;
+	}
+	return PyLong_FromLong(id);
+}
+
 static PyMethodDef perf__methods[] = {
 	{
 		.ml_name  = "metrics",
@@ -2668,6 +2700,18 @@ static PyMethodDef perf__methods[] = {
 		.ml_flags = METH_NOARGS,
 		.ml_doc	  = PyDoc_STR("Returns a sequence of pmus.")
 	},
+	{
+		.ml_name  = "syscall_name",
+		.ml_meth  = (PyCFunction) pyrf__syscall_name,
+		.ml_flags = METH_VARARGS,
+		.ml_doc	  = PyDoc_STR("Turns a syscall number to a string.")
+	},
+	{
+		.ml_name  = "syscall_id",
+		.ml_meth  = (PyCFunction) pyrf__syscall_id,
+		.ml_flags = METH_VARARGS,
+		.ml_doc	  = PyDoc_STR("Turns a syscall name to a number.")
+	},
 	{ .ml_name = NULL, }
 };
 
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 14/58] perf python: Add wrapper for perf_data file abstraction
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

The perf_data struct is needed for session supported.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:
1. Fixed Memory & FD Leaks in pyrf_data__init : Added cleanup of old
   state (closing file and freeing path) if __init__ is called
   multiple times on the same object.

2. Fixed Invalid Free in pyrf_data__delete : Ensured pdata->data.path
   is always dynamically allocated via strdup() , even for the default
   "perf.data" . This avoids passing a static string literal to free().

3. Fixed NULL Pointer Dereference in pyrf_data__str : Added a check
   for NULL path to prevent segfaults if str() or repr() is called on
   an uninitialized object.

4. Guarded fd Argument Usage: Added a check to ensure that if an fd is
   provided, it corresponds to a pipe, failing gracefully with a
   ValueError otherwise.
---
 tools/perf/util/python.c | 93 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 92 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 0424290f8b77..a2cdd92e0548 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -10,6 +10,7 @@
 
 #include "callchain.h"
 #include "counts.h"
+#include "data.h"
 #include "event.h"
 #include "evlist.h"
 #include "evsel.h"
@@ -2296,6 +2297,92 @@ static PyObject *pyrf__metrics(PyObject *self, PyObject *args)
 	return list;
 }
 
+struct pyrf_data {
+	PyObject_HEAD
+
+	struct perf_data data;
+};
+
+static int pyrf_data__init(struct pyrf_data *pdata, PyObject *args, PyObject *kwargs)
+{
+	static char *kwlist[] = { "path", "fd", NULL };
+	char *path = NULL;
+	int fd = -1;
+
+	if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|si", kwlist, &path, &fd))
+		return -1;
+
+	if (pdata->data.open)
+		perf_data__close(&pdata->data);
+	free((char *)pdata->data.path);
+
+	if (!path)
+		path = "perf.data";
+
+	pdata->data.path = strdup(path);
+	if (!pdata->data.path) {
+		PyErr_NoMemory();
+		return -1;
+	}
+
+	if (fd != -1) {
+		struct stat st;
+
+		if (fstat(fd, &st) < 0 || !S_ISFIFO(st.st_mode)) {
+			PyErr_SetString(PyExc_ValueError,
+					"fd argument is only supported for pipes");
+			free((char *)pdata->data.path);
+			pdata->data.path = NULL;
+			return -1;
+		}
+	}
+
+	pdata->data.mode = PERF_DATA_MODE_READ;
+	pdata->data.file.fd = fd;
+	if (perf_data__open(&pdata->data) < 0) {
+		PyErr_Format(PyExc_IOError, "Failed to open perf data: %s",
+			     pdata->data.path ? pdata->data.path : "perf.data");
+		return -1;
+	}
+	return 0;
+}
+
+static void pyrf_data__delete(struct pyrf_data *pdata)
+{
+	perf_data__close(&pdata->data);
+	free((char *)pdata->data.path);
+	Py_TYPE(pdata)->tp_free((PyObject *)pdata);
+}
+
+static PyObject *pyrf_data__str(PyObject *self)
+{
+	const struct pyrf_data *pdata = (const struct pyrf_data *)self;
+
+	if (!pdata->data.path)
+		return PyUnicode_FromString("[uninitialized]");
+	return PyUnicode_FromString(pdata->data.path);
+}
+
+static const char pyrf_data__doc[] = PyDoc_STR("perf data file object.");
+
+static PyTypeObject pyrf_data__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.data",
+	.tp_basicsize	= sizeof(struct pyrf_data),
+	.tp_dealloc	= (destructor)pyrf_data__delete,
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= pyrf_data__doc,
+	.tp_init	= (initproc)pyrf_data__init,
+	.tp_repr	= pyrf_data__str,
+	.tp_str		= pyrf_data__str,
+};
+
+static int pyrf_data__setup_types(void)
+{
+	pyrf_data__type.tp_new = PyType_GenericNew;
+	return PyType_Ready(&pyrf_data__type);
+}
+
 static PyMethodDef perf__methods[] = {
 	{
 		.ml_name  = "metrics",
@@ -2358,7 +2445,8 @@ PyMODINIT_FUNC PyInit_perf(void)
 	    pyrf_cpu_map__setup_types() < 0 ||
 	    pyrf_pmu_iterator__setup_types() < 0 ||
 	    pyrf_pmu__setup_types() < 0 ||
-	    pyrf_counts_values__setup_types() < 0)
+	    pyrf_counts_values__setup_types() < 0 ||
+	    pyrf_data__setup_types() < 0)
 		return module;
 
 	/* The page_size is placed in util object. */
@@ -2406,6 +2494,9 @@ PyMODINIT_FUNC PyInit_perf(void)
 	Py_INCREF(&pyrf_counts_values__type);
 	PyModule_AddObject(module, "counts_values", (PyObject *)&pyrf_counts_values__type);
 
+	Py_INCREF(&pyrf_data__type);
+	PyModule_AddObject(module, "data", (PyObject *)&pyrf_data__type);
+
 	dict = PyModule_GetDict(module);
 	if (dict == NULL)
 		goto error;
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 10/58] perf evlist: Add reference count
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

This a no-op for most of the perf tool. The reference count is set to
1 at allocation, the put will see the 1, decrement it and perform the
delete. The purpose for adding the reference count is for the python
code. Prior to this change the python code would clone evlists, but
this has issues if events are opened, etc. This change adds a
reference count for the evlists and a later change will add it to
evsels. The combination is needed for the python code to operate
correctly (not hit asserts in the evsel clone), but the changes are
broken apart for the sake of smaller patches.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---

v2: Added evlist__put to pyrf_evlist__init in case init is called more
    than once.

    I double-checked trace__replay() and confirmed that trace->evlist
    is not assigned to session->evlist in that function.
    trace__replay creates a new session and uses its own evlist for
    processing file events, leaving trace->evlist pointing to the
    empty list created at startup. Therefore, the
    evlist__put(trace->evlist) call in trace__exit() is safe and
    correct to avoid leaking that empty list.
---
 tools/perf/arch/x86/tests/hybrid.c          |   2 +-
 tools/perf/arch/x86/tests/topdown.c         |   2 +-
 tools/perf/arch/x86/util/iostat.c           |   2 +-
 tools/perf/bench/evlist-open-close.c        |  18 +-
 tools/perf/builtin-ftrace.c                 |   8 +-
 tools/perf/builtin-kvm.c                    |   4 +-
 tools/perf/builtin-lock.c                   |   2 +-
 tools/perf/builtin-record.c                 |   4 +-
 tools/perf/builtin-sched.c                  |   6 +-
 tools/perf/builtin-script.c                 |   2 +-
 tools/perf/builtin-stat.c                   |  10 +-
 tools/perf/builtin-top.c                    |  52 ++---
 tools/perf/builtin-trace.c                  |  26 +--
 tools/perf/tests/backward-ring-buffer.c     |  18 +-
 tools/perf/tests/code-reading.c             |   4 +-
 tools/perf/tests/event-times.c              |   4 +-
 tools/perf/tests/event_update.c             |   2 +-
 tools/perf/tests/evsel-roundtrip-name.c     |   8 +-
 tools/perf/tests/expand-cgroup.c            |   8 +-
 tools/perf/tests/hists_cumulate.c           |   2 +-
 tools/perf/tests/hists_filter.c             |   2 +-
 tools/perf/tests/hists_link.c               |   2 +-
 tools/perf/tests/hists_output.c             |   2 +-
 tools/perf/tests/hwmon_pmu.c                |   2 +-
 tools/perf/tests/keep-tracking.c            |   2 +-
 tools/perf/tests/mmap-basic.c               |  18 +-
 tools/perf/tests/openat-syscall-tp-fields.c |  18 +-
 tools/perf/tests/parse-events.c             |   4 +-
 tools/perf/tests/parse-metric.c             |   4 +-
 tools/perf/tests/parse-no-sample-id-all.c   |   2 +-
 tools/perf/tests/perf-record.c              |  18 +-
 tools/perf/tests/perf-time-to-tsc.c         |   2 +-
 tools/perf/tests/pfm.c                      |   4 +-
 tools/perf/tests/pmu-events.c               |   6 +-
 tools/perf/tests/pmu.c                      |   4 +-
 tools/perf/tests/sw-clock.c                 |  14 +-
 tools/perf/tests/switch-tracking.c          |   2 +-
 tools/perf/tests/task-exit.c                |  14 +-
 tools/perf/tests/tool_pmu.c                 |   2 +-
 tools/perf/tests/topology.c                 |   2 +-
 tools/perf/util/cgroup.c                    |   4 +-
 tools/perf/util/data-convert-bt.c           |   2 +-
 tools/perf/util/evlist.c                    |  20 +-
 tools/perf/util/evlist.h                    |   7 +-
 tools/perf/util/expr.c                      |   2 +-
 tools/perf/util/header.c                    |  12 +-
 tools/perf/util/metricgroup.c               |   6 +-
 tools/perf/util/parse-events.c              |   4 +-
 tools/perf/util/perf_api_probe.c            |   2 +-
 tools/perf/util/python.c                    | 200 +++++++-------------
 tools/perf/util/record.c                    |   2 +-
 tools/perf/util/session.c                   |   2 +-
 tools/perf/util/sideband_evlist.c           |  16 +-
 53 files changed, 268 insertions(+), 319 deletions(-)

diff --git a/tools/perf/arch/x86/tests/hybrid.c b/tools/perf/arch/x86/tests/hybrid.c
index e221ea104174..dfb0ffc0d030 100644
--- a/tools/perf/arch/x86/tests/hybrid.c
+++ b/tools/perf/arch/x86/tests/hybrid.c
@@ -268,7 +268,7 @@ static int test_event(const struct evlist_test *e)
 		ret = e->check(evlist);
 	}
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 
 	return ret;
 }
diff --git a/tools/perf/arch/x86/tests/topdown.c b/tools/perf/arch/x86/tests/topdown.c
index 3ee4e5e71be3..2d0ae5b0d76c 100644
--- a/tools/perf/arch/x86/tests/topdown.c
+++ b/tools/perf/arch/x86/tests/topdown.c
@@ -56,7 +56,7 @@ static int event_cb(void *state, struct pmu_event_info *info)
 			*ret = TEST_FAIL;
 		}
 	}
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return 0;
 }
 
diff --git a/tools/perf/arch/x86/util/iostat.c b/tools/perf/arch/x86/util/iostat.c
index 7442a2cd87ed..e0417552b0cb 100644
--- a/tools/perf/arch/x86/util/iostat.c
+++ b/tools/perf/arch/x86/util/iostat.c
@@ -337,7 +337,7 @@ int iostat_prepare(struct evlist *evlist, struct perf_stat_config *config)
 	if (evlist->core.nr_entries > 0) {
 		pr_warning("The -e and -M options are not supported."
 			   "All chosen events/metrics will be dropped\n");
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		evlist = evlist__new();
 		if (!evlist)
 			return -ENOMEM;
diff --git a/tools/perf/bench/evlist-open-close.c b/tools/perf/bench/evlist-open-close.c
index faf9c34b4a5d..304929d1f67f 100644
--- a/tools/perf/bench/evlist-open-close.c
+++ b/tools/perf/bench/evlist-open-close.c
@@ -76,7 +76,7 @@ static struct evlist *bench__create_evlist(char *evstr, const char *uid_str)
 		parse_events_error__exit(&err);
 		pr_err("Run 'perf list' for a list of valid events\n");
 		ret = 1;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 	parse_events_error__exit(&err);
 	if (uid_str) {
@@ -85,24 +85,24 @@ static struct evlist *bench__create_evlist(char *evstr, const char *uid_str)
 		if (uid == UINT_MAX) {
 			pr_err("Invalid User: %s", uid_str);
 			ret = -EINVAL;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 		ret = parse_uid_filter(evlist, uid);
 		if (ret)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 	ret = evlist__create_maps(evlist, &opts.target);
 	if (ret < 0) {
 		pr_err("Not enough memory to create thread/cpu maps\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__config(evlist, &opts, NULL);
 
 	return evlist;
 
-out_delete_evlist:
-	evlist__delete(evlist);
+out_put_evlist:
+	evlist__put(evlist);
 	return NULL;
 }
 
@@ -151,7 +151,7 @@ static int bench_evlist_open_close__run(char *evstr, const char *uid_str)
 		evlist->core.nr_entries, evlist__count_evsel_fds(evlist));
 	printf("  Number of iterations:\t%d\n", iterations);
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 
 	for (i = 0; i < iterations; i++) {
 		pr_debug("Started iteration %d\n", i);
@@ -162,7 +162,7 @@ static int bench_evlist_open_close__run(char *evstr, const char *uid_str)
 		gettimeofday(&start, NULL);
 		err = bench__do_evlist_open_close(evlist);
 		if (err) {
-			evlist__delete(evlist);
+			evlist__put(evlist);
 			return err;
 		}
 
@@ -171,7 +171,7 @@ static int bench_evlist_open_close__run(char *evstr, const char *uid_str)
 		runtime_us = timeval2usec(&diff);
 		update_stats(&time_stats, runtime_us);
 
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		pr_debug("Iteration %d took:\t%" PRIu64 "us\n", i, runtime_us);
 	}
 
diff --git a/tools/perf/builtin-ftrace.c b/tools/perf/builtin-ftrace.c
index 8a7dbfb14535..676239148b87 100644
--- a/tools/perf/builtin-ftrace.c
+++ b/tools/perf/builtin-ftrace.c
@@ -1999,20 +1999,20 @@ int cmd_ftrace(int argc, const char **argv)
 
 	ret = evlist__create_maps(ftrace.evlist, &ftrace.target);
 	if (ret < 0)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	if (argc) {
 		ret = evlist__prepare_workload(ftrace.evlist, &ftrace.target,
 					       argv, false,
 					       ftrace__workload_exec_failed_signal);
 		if (ret < 0)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 
 	ret = cmd_func(&ftrace);
 
-out_delete_evlist:
-	evlist__delete(ftrace.evlist);
+out_put_evlist:
+	evlist__put(ftrace.evlist);
 
 out_delete_filters:
 	delete_filter_func(&ftrace.filters);
diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c
index 0c5e6b3aac74..d88855e3c7b4 100644
--- a/tools/perf/builtin-kvm.c
+++ b/tools/perf/builtin-kvm.c
@@ -1811,7 +1811,7 @@ static struct evlist *kvm_live_event_list(void)
 
 out:
 	if (err) {
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		evlist = NULL;
 	}
 
@@ -1942,7 +1942,7 @@ static int kvm_events_live(struct perf_kvm_stat *kvm,
 out:
 	perf_session__delete(kvm->session);
 	kvm->session = NULL;
-	evlist__delete(kvm->evlist);
+	evlist__put(kvm->evlist);
 
 	return err;
 }
diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c
index 5585aeb97684..c40d070f6c36 100644
--- a/tools/perf/builtin-lock.c
+++ b/tools/perf/builtin-lock.c
@@ -2145,7 +2145,7 @@ static int __cmd_contention(int argc, const char **argv)
 
 out_delete:
 	lock_filter_finish();
-	evlist__delete(con.evlist);
+	evlist__put(con.evlist);
 	lock_contention_finish(&con);
 	perf_session__delete(session);
 	perf_env__exit(&host_env);
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 4a5eba498c02..b4fffa936e01 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -4289,7 +4289,7 @@ int cmd_record(int argc, const char **argv)
 			goto out;
 
 		evlist__splice_list_tail(rec->evlist, &def_evlist->core.entries);
-		evlist__delete(def_evlist);
+		evlist__put(def_evlist);
 	}
 
 	if (rec->opts.target.tid && !rec->opts.no_inherit_set)
@@ -4397,7 +4397,7 @@ int cmd_record(int argc, const char **argv)
 	auxtrace_record__free(rec->itr);
 out_opts:
 	evlist__close_control(rec->opts.ctl_fd, rec->opts.ctl_fd_ack, &rec->opts.ctl_fd_close);
-	evlist__delete(rec->evlist);
+	evlist__put(rec->evlist);
 	return err;
 }
 
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 555247568e7a..d683642ab4e0 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -3829,7 +3829,7 @@ static int perf_sched__schedstat_record(struct perf_sched *sched,
 	session = perf_session__new(&data, &sched->tool);
 	if (IS_ERR(session)) {
 		pr_err("Perf session creation failed.\n");
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		return PTR_ERR(session);
 	}
 
@@ -3925,7 +3925,7 @@ static int perf_sched__schedstat_record(struct perf_sched *sched,
 	else
 		fprintf(stderr, "[ perf sched stats: Failed !! ]\n");
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	close(fd);
 	return err;
 }
@@ -4724,7 +4724,7 @@ static int perf_sched__schedstat_live(struct perf_sched *sched,
 	free_cpu_domain_info(cd_map, sv, nr);
 out:
 	free_schedstat(&cpu_head);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 853b141a0d50..0ead134940d5 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -2269,7 +2269,7 @@ static int script_find_metrics(const struct pmu_metric *pm,
 	}
 	pr_debug("Found metric '%s' whose evsels match those of in the perf data\n",
 		 pm->metric_name);
-	evlist__delete(metric_evlist);
+	evlist__put(metric_evlist);
 out:
 	return 0;
 }
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 99d7db372b48..bfa3512e1686 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -2113,7 +2113,7 @@ static int add_default_events(void)
 							stat_config.user_requested_cpu_list,
 							stat_config.system_wide,
 							stat_config.hardware_aware_grouping) < 0) {
-				evlist__delete(metric_evlist);
+				evlist__put(metric_evlist);
 				ret = -1;
 				break;
 			}
@@ -2125,7 +2125,7 @@ static int add_default_events(void)
 			metricgroup__copy_metric_events(evlist, /*cgrp=*/NULL,
 							&evlist->metric_events,
 							&metric_evlist->metric_events);
-			evlist__delete(metric_evlist);
+			evlist__put(metric_evlist);
 		}
 		list_sort(/*priv=*/NULL, &evlist->core.entries, default_evlist_evsel_cmp);
 
@@ -2146,7 +2146,7 @@ static int add_default_events(void)
 	metricgroup__copy_metric_events(evsel_list, /*cgrp=*/NULL,
 					&evsel_list->metric_events,
 					&evlist->metric_events);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
@@ -2381,7 +2381,7 @@ static int __cmd_report(int argc, const char **argv)
 
 	perf_stat.session  = session;
 	stat_config.output = stderr;
-	evlist__delete(evsel_list);
+	evlist__put(evsel_list);
 	evsel_list         = session->evlist;
 
 	ret = perf_session__process_events(session);
@@ -3060,7 +3060,7 @@ int cmd_stat(int argc, const char **argv)
 	if (smi_cost && smi_reset)
 		sysfs__write_int(FREEZE_ON_SMI_PATH, 0);
 
-	evlist__delete(evsel_list);
+	evlist__put(evsel_list);
 
 	evlist__close_control(stat_config.ctl_fd, stat_config.ctl_fd_ack, &stat_config.ctl_fd_close);
 
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index f6eb543de537..c509cfef8285 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -1652,14 +1652,14 @@ int cmd_top(int argc, const char **argv)
 	perf_env__init(&host_env);
 	status = perf_config(perf_top_config, &top);
 	if (status)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	/*
 	 * Since the per arch annotation init routine may need the cpuid, read
 	 * it here, since we are not getting this from the perf.data header.
 	 */
 	status = perf_env__set_cmdline(&host_env, argc, argv);
 	if (status)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	status = perf_env__read_cpuid(&host_env);
 	if (status) {
@@ -1680,30 +1680,30 @@ int cmd_top(int argc, const char **argv)
 		annotate_opts.disassembler_style = strdup(disassembler_style);
 		if (!annotate_opts.disassembler_style) {
 			status = -ENOMEM;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 	if (objdump_path) {
 		annotate_opts.objdump_path = strdup(objdump_path);
 		if (!annotate_opts.objdump_path) {
 			status = -ENOMEM;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 	if (addr2line_path) {
 		symbol_conf.addr2line_path = strdup(addr2line_path);
 		if (!symbol_conf.addr2line_path) {
 			status = -ENOMEM;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 
 	status = symbol__validate_sym_arguments();
 	if (status)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	if (annotate_check_args() < 0)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	status = target__validate(target);
 	if (status) {
@@ -1718,15 +1718,15 @@ int cmd_top(int argc, const char **argv)
 		struct evlist *def_evlist = evlist__new_default(target, callchain_param.enabled);
 
 		if (!def_evlist)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 
 		evlist__splice_list_tail(top.evlist, &def_evlist->core.entries);
-		evlist__delete(def_evlist);
+		evlist__put(def_evlist);
 	}
 
 	status = evswitch__init(&top.evswitch, top.evlist, stderr);
 	if (status)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	if (symbol_conf.report_hierarchy) {
 		/* disable incompatible options */
@@ -1737,18 +1737,18 @@ int cmd_top(int argc, const char **argv)
 			pr_err("Error: --hierarchy and --fields options cannot be used together\n");
 			parse_options_usage(top_usage, options, "fields", 0);
 			parse_options_usage(NULL, options, "hierarchy", 0);
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 
 	if (top.stitch_lbr && !(callchain_param.record_mode == CALLCHAIN_LBR)) {
 		pr_err("Error: --stitch-lbr must be used with --call-graph lbr\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (nr_cgroups > 0 && opts->record_cgroup) {
 		pr_err("--cgroup and --all-cgroups cannot be used together\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (branch_call_mode) {
@@ -1772,7 +1772,7 @@ int cmd_top(int argc, const char **argv)
 		status = perf_env__read_core_pmu_caps(&host_env);
 		if (status) {
 			pr_err("PMU capability data is not available\n");
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 
@@ -1795,7 +1795,7 @@ int cmd_top(int argc, const char **argv)
 	if (IS_ERR(top.session)) {
 		status = PTR_ERR(top.session);
 		top.session = NULL;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 	top.evlist->session = top.session;
 
@@ -1805,7 +1805,7 @@ int cmd_top(int argc, const char **argv)
 		if (field_order)
 			parse_options_usage(sort_order ? NULL : top_usage,
 					    options, "fields", 0);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (top.uid_str) {
@@ -1814,18 +1814,18 @@ int cmd_top(int argc, const char **argv)
 		if (uid == UINT_MAX) {
 			ui__error("Invalid User: %s", top.uid_str);
 			status = -EINVAL;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 		status = parse_uid_filter(top.evlist, uid);
 		if (status)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 
 	if (evlist__create_maps(top.evlist, target) < 0) {
 		ui__error("Couldn't create thread/CPU maps: %s\n",
 			  errno == ENOENT ? "No such process" : str_error_r(errno, errbuf, sizeof(errbuf)));
 		status = -errno;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (top.delay_secs < 1)
@@ -1833,7 +1833,7 @@ int cmd_top(int argc, const char **argv)
 
 	if (record_opts__config(opts)) {
 		status = -EINVAL;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	top.sym_evsel = evlist__first(top.evlist);
@@ -1848,14 +1848,14 @@ int cmd_top(int argc, const char **argv)
 
 	status = symbol__annotation_init();
 	if (status < 0)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	annotation_config__init();
 
 	symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL);
 	status = symbol__init(NULL);
 	if (status < 0)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	sort__setup_elide(stdout);
 
@@ -1875,13 +1875,13 @@ int cmd_top(int argc, const char **argv)
 		if (top.sb_evlist == NULL) {
 			pr_err("Couldn't create side band evlist.\n.");
 			status = -EINVAL;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		if (evlist__add_bpf_sb_event(top.sb_evlist, &host_env)) {
 			pr_err("Couldn't ask for PERF_RECORD_BPF_EVENT side band events.\n.");
 			status = -EINVAL;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 #endif
@@ -1896,8 +1896,8 @@ int cmd_top(int argc, const char **argv)
 	if (!opts->no_bpf_event)
 		evlist__stop_sb_thread(top.sb_evlist);
 
-out_delete_evlist:
-	evlist__delete(top.evlist);
+out_put_evlist:
+	evlist__put(top.evlist);
 	perf_session__delete(top.session);
 	annotation_options__exit();
 	perf_env__exit(&host_env);
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index e58c49d047a2..da703d762433 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -4388,7 +4388,7 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 
 	if (trace->summary_bpf) {
 		if (trace_prepare_bpf_summary(trace->summary_mode) < 0)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 
 		if (trace->summary_only)
 			goto create_maps;
@@ -4456,19 +4456,19 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 	err = evlist__create_maps(evlist, &trace->opts.target);
 	if (err < 0) {
 		fprintf(trace->output, "Problems parsing the target to trace, check your options!\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	err = trace__symbols_init(trace, argc, argv, evlist);
 	if (err < 0) {
 		fprintf(trace->output, "Problems initializing symbol libraries!\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (trace->summary_mode == SUMMARY__BY_TOTAL && !trace->summary_bpf) {
 		trace->syscall_stats = alloc_syscall_stats();
 		if (!trace->syscall_stats)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 
 	evlist__config(evlist, &trace->opts, &callchain_param);
@@ -4477,7 +4477,7 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 		err = evlist__prepare_workload(evlist, &trace->opts.target, argv, false, NULL);
 		if (err < 0) {
 			fprintf(trace->output, "Couldn't run the workload!\n");
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 		workload_pid = evlist->workload.pid;
 	}
@@ -4525,7 +4525,7 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 
 	err = trace__expand_filters(trace, &evsel);
 	if (err)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	err = evlist__apply_filters(evlist, &evsel, &trace->opts.target);
 	if (err < 0)
 		goto out_error_apply_filters;
@@ -4642,12 +4642,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 		}
 	}
 
-out_delete_evlist:
+out_put_evlist:
 	trace_cleanup_bpf_summary();
 	delete_syscall_stats(trace->syscall_stats);
 	trace__symbols__exit(trace);
 	evlist__free_syscall_tp_fields(evlist);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	cgroup__put(trace->cgroup);
 	trace->evlist = NULL;
 	trace->live = false;
@@ -4672,21 +4672,21 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 
 out_error:
 	fprintf(trace->output, "%s\n", errbuf);
-	goto out_delete_evlist;
+	goto out_put_evlist;
 
 out_error_apply_filters:
 	fprintf(trace->output,
 		"Failed to set filter \"%s\" on event %s: %m\n",
 		evsel->filter, evsel__name(evsel));
-	goto out_delete_evlist;
+	goto out_put_evlist;
 }
 out_error_mem:
 	fprintf(trace->output, "Not enough memory to run!\n");
-	goto out_delete_evlist;
+	goto out_put_evlist;
 
 out_errno:
 	fprintf(trace->output, "%m\n");
-	goto out_delete_evlist;
+	goto out_put_evlist;
 }
 
 static int trace__replay(struct trace *trace)
@@ -5364,7 +5364,7 @@ static void trace__exit(struct trace *trace)
 		zfree(&trace->syscalls.table);
 	}
 	zfree(&trace->perfconfig_events);
-	evlist__delete(trace->evlist);
+	evlist__put(trace->evlist);
 	trace->evlist = NULL;
 	ordered_events__free(&trace->oe.data);
 #ifdef HAVE_LIBBPF_SUPPORT
diff --git a/tools/perf/tests/backward-ring-buffer.c b/tools/perf/tests/backward-ring-buffer.c
index c5e7999f2817..2b49b002d749 100644
--- a/tools/perf/tests/backward-ring-buffer.c
+++ b/tools/perf/tests/backward-ring-buffer.c
@@ -111,7 +111,7 @@ static int test__backward_ring_buffer(struct test_suite *test __maybe_unused, in
 	err = evlist__create_maps(evlist, &opts.target);
 	if (err < 0) {
 		pr_debug("Not enough memory to create thread/cpu maps\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	parse_events_error__init(&parse_error);
@@ -124,7 +124,7 @@ static int test__backward_ring_buffer(struct test_suite *test __maybe_unused, in
 	if (err) {
 		pr_debug("Failed to parse tracepoint event, try use root\n");
 		ret = TEST_SKIP;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__config(evlist, &opts, NULL);
@@ -133,19 +133,19 @@ static int test__backward_ring_buffer(struct test_suite *test __maybe_unused, in
 	if (err < 0) {
 		pr_debug("perf_evlist__open: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	ret = TEST_FAIL;
 	err = do_test(evlist, opts.mmap_pages, &sample_count,
 		      &comm_count);
 	if (err != TEST_OK)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	if ((sample_count != NR_ITERS) || (comm_count != NR_ITERS)) {
 		pr_err("Unexpected counter: sample_count=%d, comm_count=%d\n",
 		       sample_count, comm_count);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__close(evlist);
@@ -154,16 +154,16 @@ static int test__backward_ring_buffer(struct test_suite *test __maybe_unused, in
 	if (err < 0) {
 		pr_debug("perf_evlist__open: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	err = do_test(evlist, 1, &sample_count, &comm_count);
 	if (err != TEST_OK)
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	ret = TEST_OK;
-out_delete_evlist:
-	evlist__delete(evlist);
+out_put_evlist:
+	evlist__put(evlist);
 	return ret;
 }
 
diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
index 47043a3a2fb4..fc65a17f67f7 100644
--- a/tools/perf/tests/code-reading.c
+++ b/tools/perf/tests/code-reading.c
@@ -807,7 +807,7 @@ static int do_test_code_reading(bool try_kcore)
 			}
 
 			perf_evlist__set_maps(&evlist->core, NULL, NULL);
-			evlist__delete(evlist);
+			evlist__put(evlist);
 			evlist = NULL;
 			continue;
 		}
@@ -844,7 +844,7 @@ static int do_test_code_reading(bool try_kcore)
 out_put:
 	thread__put(thread);
 out_err:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
 	machine__delete(machine);
diff --git a/tools/perf/tests/event-times.c b/tools/perf/tests/event-times.c
index ae3b98bb42cf..94ab54ecd3f9 100644
--- a/tools/perf/tests/event-times.c
+++ b/tools/perf/tests/event-times.c
@@ -186,7 +186,7 @@ static int test_times(int (attach)(struct evlist *),
 	err = attach(evlist);
 	if (err == TEST_SKIP) {
 		pr_debug("  SKIP  : not enough rights\n");
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		return err;
 	}
 
@@ -205,7 +205,7 @@ static int test_times(int (attach)(struct evlist *),
 		 count.ena, count.run);
 
 out_err:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return !err ? TEST_OK : TEST_FAIL;
 }
 
diff --git a/tools/perf/tests/event_update.c b/tools/perf/tests/event_update.c
index facc65e29f20..73141b122d2f 100644
--- a/tools/perf/tests/event_update.c
+++ b/tools/perf/tests/event_update.c
@@ -117,7 +117,7 @@ static int test__event_update(struct test_suite *test __maybe_unused, int subtes
 	TEST_ASSERT_VAL("failed to synthesize attr update cpus",
 			!perf_event__synthesize_event_update_cpus(&tmp.tool, evsel, process_event_cpus));
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return 0;
 }
 
diff --git a/tools/perf/tests/evsel-roundtrip-name.c b/tools/perf/tests/evsel-roundtrip-name.c
index 1922cac13a24..6a220634c52f 100644
--- a/tools/perf/tests/evsel-roundtrip-name.c
+++ b/tools/perf/tests/evsel-roundtrip-name.c
@@ -33,7 +33,7 @@ static int perf_evsel__roundtrip_cache_name_test(void)
 				if (err) {
 					pr_debug("Failure to parse cache event '%s' possibly as PMUs don't support it",
 						name);
-					evlist__delete(evlist);
+					evlist__put(evlist);
 					continue;
 				}
 				evlist__for_each_entry(evlist, evsel) {
@@ -42,7 +42,7 @@ static int perf_evsel__roundtrip_cache_name_test(void)
 						ret = TEST_FAIL;
 					}
 				}
-				evlist__delete(evlist);
+				evlist__put(evlist);
 			}
 		}
 	}
@@ -66,7 +66,7 @@ static int perf_evsel__name_array_test(const char *const names[], int nr_names)
 		if (err) {
 			pr_debug("failed to parse event '%s', err %d\n",
 				 names[i], err);
-			evlist__delete(evlist);
+			evlist__put(evlist);
 			ret = TEST_FAIL;
 			continue;
 		}
@@ -76,7 +76,7 @@ static int perf_evsel__name_array_test(const char *const names[], int nr_names)
 				ret = TEST_FAIL;
 			}
 		}
-		evlist__delete(evlist);
+		evlist__put(evlist);
 	}
 	return ret;
 }
diff --git a/tools/perf/tests/expand-cgroup.c b/tools/perf/tests/expand-cgroup.c
index dd547f2f77cc..a7a445f12693 100644
--- a/tools/perf/tests/expand-cgroup.c
+++ b/tools/perf/tests/expand-cgroup.c
@@ -106,7 +106,7 @@ static int expand_default_events(void)
 	TEST_ASSERT_VAL("failed to get evlist", evlist);
 
 	ret = test_expand_events(evlist);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
@@ -133,7 +133,7 @@ static int expand_group_events(void)
 	ret = test_expand_events(evlist);
 out:
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
@@ -164,7 +164,7 @@ static int expand_libpfm_events(void)
 
 	ret = test_expand_events(evlist);
 out:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
@@ -188,7 +188,7 @@ static int expand_metric_events(void)
 	ret = test_expand_events(evlist);
 
 out:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
diff --git a/tools/perf/tests/hists_cumulate.c b/tools/perf/tests/hists_cumulate.c
index 606aa926a8fc..eca4ecb63ca8 100644
--- a/tools/perf/tests/hists_cumulate.c
+++ b/tools/perf/tests/hists_cumulate.c
@@ -744,7 +744,7 @@ static int test__hists_cumulate(struct test_suite *test __maybe_unused, int subt
 
 out:
 	/* tear down everything */
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	machines__exit(&machines);
 	put_fake_samples();
 
diff --git a/tools/perf/tests/hists_filter.c b/tools/perf/tests/hists_filter.c
index cc6b26e373d1..0d09dc306019 100644
--- a/tools/perf/tests/hists_filter.c
+++ b/tools/perf/tests/hists_filter.c
@@ -332,7 +332,7 @@ static int test__hists_filter(struct test_suite *test __maybe_unused, int subtes
 
 out:
 	/* tear down everything */
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	reset_output_field();
 	machines__exit(&machines);
 	put_fake_samples();
diff --git a/tools/perf/tests/hists_link.c b/tools/perf/tests/hists_link.c
index 996f5f0b3bd1..9646c3b7b4de 100644
--- a/tools/perf/tests/hists_link.c
+++ b/tools/perf/tests/hists_link.c
@@ -352,7 +352,7 @@ static int test__hists_link(struct test_suite *test __maybe_unused, int subtest
 
 out:
 	/* tear down everything */
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	reset_output_field();
 	machines__exit(&machines);
 	put_fake_samples();
diff --git a/tools/perf/tests/hists_output.c b/tools/perf/tests/hists_output.c
index 7818950d786e..3f3bc978553e 100644
--- a/tools/perf/tests/hists_output.c
+++ b/tools/perf/tests/hists_output.c
@@ -631,7 +631,7 @@ static int test__hists_output(struct test_suite *test __maybe_unused, int subtes
 
 out:
 	/* tear down everything */
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	machines__exit(&machines);
 	put_fake_samples();
 
diff --git a/tools/perf/tests/hwmon_pmu.c b/tools/perf/tests/hwmon_pmu.c
index ada6e445c4c4..1b60c3a900f1 100644
--- a/tools/perf/tests/hwmon_pmu.c
+++ b/tools/perf/tests/hwmon_pmu.c
@@ -214,7 +214,7 @@ static int do_test(size_t i, bool with_pmu, bool with_alias)
 
 out:
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
diff --git a/tools/perf/tests/keep-tracking.c b/tools/perf/tests/keep-tracking.c
index 729cc9cc1cb7..51cfd6522867 100644
--- a/tools/perf/tests/keep-tracking.c
+++ b/tools/perf/tests/keep-tracking.c
@@ -153,7 +153,7 @@ static int test__keep_tracking(struct test_suite *test __maybe_unused, int subte
 out_err:
 	if (evlist) {
 		evlist__disable(evlist);
-		evlist__delete(evlist);
+		evlist__put(evlist);
 	}
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
diff --git a/tools/perf/tests/mmap-basic.c b/tools/perf/tests/mmap-basic.c
index 8d04f6edb004..e6501791c505 100644
--- a/tools/perf/tests/mmap-basic.c
+++ b/tools/perf/tests/mmap-basic.c
@@ -94,7 +94,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 				/* Permissions failure, flag the failure as a skip. */
 				err = TEST_SKIP;
 			}
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		evsels[i]->core.attr.wakeup_events = 1;
@@ -106,7 +106,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 			pr_debug("failed to open counter: %s, "
 				 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
 				 str_error_r(errno, sbuf, sizeof(sbuf)));
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		nr_events[i] = 0;
@@ -116,7 +116,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 	if (evlist__mmap(evlist, 128) < 0) {
 		pr_debug("failed to mmap events: %d (%s)\n", errno,
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	for (i = 0; i < nsyscalls; ++i)
@@ -134,7 +134,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 		if (event->header.type != PERF_RECORD_SAMPLE) {
 			pr_debug("unexpected %s event\n",
 				 perf_event__name(event->header.type));
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		perf_sample__init(&sample, /*all=*/false);
@@ -142,7 +142,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 		if (err) {
 			pr_err("Can't parse sample, err = %d\n", err);
 			perf_sample__exit(&sample);
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		err = -1;
@@ -151,7 +151,7 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 		if (evsel == NULL) {
 			pr_debug("event with id %" PRIu64
 				 " doesn't map to an evsel\n", sample.id);
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 		nr_events[evsel->core.idx]++;
 		perf_mmap__consume(&md->core);
@@ -166,12 +166,12 @@ static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest
 				 expected_nr_events[evsel->core.idx],
 				 evsel__name(evsel), nr_events[evsel->core.idx]);
 			err = -1;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 
-out_delete_evlist:
-	evlist__delete(evlist);
+out_put_evlist:
+	evlist__put(evlist);
 out_free_cpus:
 	perf_cpu_map__put(cpus);
 out_free_threads:
diff --git a/tools/perf/tests/openat-syscall-tp-fields.c b/tools/perf/tests/openat-syscall-tp-fields.c
index 2a139d2781a8..3ff595c7a86a 100644
--- a/tools/perf/tests/openat-syscall-tp-fields.c
+++ b/tools/perf/tests/openat-syscall-tp-fields.c
@@ -51,7 +51,7 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 	if (IS_ERR(evsel)) {
 		pr_debug("%s: evsel__newtp\n", __func__);
 		ret = PTR_ERR(evsel) == -EACCES ? TEST_SKIP : TEST_FAIL;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__add(evlist, evsel);
@@ -59,7 +59,7 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 	err = evlist__create_maps(evlist, &opts.target);
 	if (err < 0) {
 		pr_debug("%s: evlist__create_maps\n", __func__);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evsel__config(evsel, &opts, NULL);
@@ -70,14 +70,14 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 	if (err < 0) {
 		pr_debug("perf_evlist__open: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	err = evlist__mmap(evlist, UINT_MAX);
 	if (err < 0) {
 		pr_debug("evlist__mmap: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__enable(evlist);
@@ -115,7 +115,7 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 				if (err) {
 					pr_debug("Can't parse sample, err = %d\n", err);
 					perf_sample__exit(&sample);
-					goto out_delete_evlist;
+					goto out_put_evlist;
 				}
 
 				tp_flags = evsel__intval(evsel, &sample, "flags");
@@ -123,7 +123,7 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 				if (flags != tp_flags) {
 					pr_debug("%s: Expected flags=%#x, got %#x\n",
 						 __func__, flags, tp_flags);
-					goto out_delete_evlist;
+					goto out_put_evlist;
 				}
 
 				goto out_ok;
@@ -136,13 +136,13 @@ static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused
 
 		if (++nr_polls > 5) {
 			pr_debug("%s: no events!\n", __func__);
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 	}
 out_ok:
 	ret = TEST_OK;
-out_delete_evlist:
-	evlist__delete(evlist);
+out_put_evlist:
+	evlist__put(evlist);
 out:
 	return ret;
 }
diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
index 05c3e899b425..19dc7b7475d2 100644
--- a/tools/perf/tests/parse-events.c
+++ b/tools/perf/tests/parse-events.c
@@ -2568,7 +2568,7 @@ static int test_event(const struct evlist_test *e)
 		ret = e->check(evlist);
 	}
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 
 	return ret;
 }
@@ -2594,7 +2594,7 @@ static int test_event_fake_pmu(const char *str)
 	}
 
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 
 	return ret;
 }
diff --git a/tools/perf/tests/parse-metric.c b/tools/perf/tests/parse-metric.c
index 7c7f489a5eb0..3f0ec839c056 100644
--- a/tools/perf/tests/parse-metric.c
+++ b/tools/perf/tests/parse-metric.c
@@ -84,7 +84,7 @@ static int __compute_metric(const char *name, struct value *vals,
 
 	cpus = perf_cpu_map__new("0");
 	if (!cpus) {
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		return -ENOMEM;
 	}
 
@@ -113,7 +113,7 @@ static int __compute_metric(const char *name, struct value *vals,
 	/* ... cleanup. */
 	evlist__free_stats(evlist);
 	perf_cpu_map__put(cpus);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/tests/parse-no-sample-id-all.c b/tools/perf/tests/parse-no-sample-id-all.c
index 50e68b7d43aa..d5a8d065809e 100644
--- a/tools/perf/tests/parse-no-sample-id-all.c
+++ b/tools/perf/tests/parse-no-sample-id-all.c
@@ -49,7 +49,7 @@ static int process_events(union perf_event **events, size_t count)
 	for (i = 0; i < count && !err; i++)
 		err = process_event(&evlist, events[i]);
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 
 	return err;
 }
diff --git a/tools/perf/tests/perf-record.c b/tools/perf/tests/perf-record.c
index ad44cc68820b..f95752b2ed1c 100644
--- a/tools/perf/tests/perf-record.c
+++ b/tools/perf/tests/perf-record.c
@@ -105,7 +105,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 	err = evlist__create_maps(evlist, &opts.target);
 	if (err < 0) {
 		pr_debug("Not enough memory to create thread/cpu maps\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	/*
@@ -117,7 +117,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 	err = evlist__prepare_workload(evlist, &opts.target, argv, false, NULL);
 	if (err < 0) {
 		pr_debug("Couldn't run the workload!\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	/*
@@ -134,7 +134,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 		pr_debug("sched__get_first_possible_cpu: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		evlist__cancel_workload(evlist);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	cpu = err;
@@ -146,7 +146,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 		pr_debug("sched_setaffinity: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		evlist__cancel_workload(evlist);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	/*
@@ -158,7 +158,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 		pr_debug("perf_evlist__open: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		evlist__cancel_workload(evlist);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	/*
@@ -171,7 +171,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 		pr_debug("evlist__mmap: %s\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		evlist__cancel_workload(evlist);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	/*
@@ -209,7 +209,7 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 					if (verbose > 0)
 						perf_event__fprintf(event, NULL, stderr);
 					pr_debug("Couldn't parse sample\n");
-					goto out_delete_evlist;
+					goto out_put_evlist;
 				}
 
 				if (verbose > 0) {
@@ -350,9 +350,9 @@ static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest
 		pr_debug("PERF_RECORD_MMAP for %s missing!\n", "[vdso]");
 		++errs;
 	}
-out_delete_evlist:
+out_put_evlist:
 	CPU_FREE(cpu_mask);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 out:
 	perf_sample__exit(&sample);
 	if (err == -EACCES)
diff --git a/tools/perf/tests/perf-time-to-tsc.c b/tools/perf/tests/perf-time-to-tsc.c
index cca41bd37ae3..d3538fa20af3 100644
--- a/tools/perf/tests/perf-time-to-tsc.c
+++ b/tools/perf/tests/perf-time-to-tsc.c
@@ -201,7 +201,7 @@ static int test__perf_time_to_tsc(struct test_suite *test __maybe_unused, int su
 	err = TEST_OK;
 
 out_err:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
 	return err;
diff --git a/tools/perf/tests/pfm.c b/tools/perf/tests/pfm.c
index fca4a86452df..8d19b1bfecbc 100644
--- a/tools/perf/tests/pfm.c
+++ b/tools/perf/tests/pfm.c
@@ -80,7 +80,7 @@ static int test__pfm_events(struct test_suite *test __maybe_unused,
 				evlist__nr_groups(evlist),
 				0);
 
-		evlist__delete(evlist);
+		evlist__put(evlist);
 	}
 	return 0;
 }
@@ -165,7 +165,7 @@ static int test__pfm_group(struct test_suite *test __maybe_unused,
 				evlist__nr_groups(evlist),
 				table[i].nr_groups);
 
-		evlist__delete(evlist);
+		evlist__put(evlist);
 	}
 	return 0;
 }
diff --git a/tools/perf/tests/pmu-events.c b/tools/perf/tests/pmu-events.c
index a99716862168..236bbbad5773 100644
--- a/tools/perf/tests/pmu-events.c
+++ b/tools/perf/tests/pmu-events.c
@@ -797,7 +797,7 @@ static int check_parse_id(const char *id, struct parse_events_error *error)
 			     /*warn_if_reordered=*/true, /*fake_tp=*/false);
 	free(dup);
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
@@ -844,7 +844,7 @@ static int test__parsing_callback(const struct pmu_metric *pm,
 
 	cpus = perf_cpu_map__new("0");
 	if (!cpus) {
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		return -ENOMEM;
 	}
 
@@ -899,7 +899,7 @@ static int test__parsing_callback(const struct pmu_metric *pm,
 	/* ... cleanup. */
 	evlist__free_stats(evlist);
 	perf_cpu_map__put(cpus);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/tests/pmu.c b/tools/perf/tests/pmu.c
index 0ebf2d7b2cb4..3d931c1f99dd 100644
--- a/tools/perf/tests/pmu.c
+++ b/tools/perf/tests/pmu.c
@@ -287,7 +287,7 @@ static int test__pmu_usr_chgs(struct test_suite *test __maybe_unused, int subtes
 	ret = TEST_OK;
 err_out:
 	parse_events_terms__exit(&terms);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	test_pmu_put(dir, pmu);
 	return ret;
 }
@@ -339,7 +339,7 @@ static int test__pmu_events(struct test_suite *test __maybe_unused, int subtest
 	ret = TEST_OK;
 err_out:
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	test_pmu_put(dir, pmu);
 	return ret;
 }
diff --git a/tools/perf/tests/sw-clock.c b/tools/perf/tests/sw-clock.c
index b6e46975379c..bb6b62cf51d1 100644
--- a/tools/perf/tests/sw-clock.c
+++ b/tools/perf/tests/sw-clock.c
@@ -59,7 +59,7 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 	evsel = evsel__new(&attr);
 	if (evsel == NULL) {
 		pr_debug("evsel__new\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 	evlist__add(evlist, evsel);
 
@@ -68,7 +68,7 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 	if (!cpus || !threads) {
 		err = -ENOMEM;
 		pr_debug("Not enough memory to create thread/cpu maps\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	perf_evlist__set_maps(&evlist->core, cpus, threads);
@@ -80,14 +80,14 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 		pr_debug("Couldn't open evlist: %s\nHint: check %s, using %" PRIu64 " in this test.\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)),
 			 knob, (u64)attr.sample_freq);
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	err = evlist__mmap(evlist, 128);
 	if (err < 0) {
 		pr_debug("failed to mmap event: %d (%s)\n", errno,
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__enable(evlist);
@@ -113,7 +113,7 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 		if (err < 0) {
 			pr_debug("Error during parse sample\n");
 			perf_sample__exit(&sample);
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		total_periods += sample.period;
@@ -131,10 +131,10 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 		err = -1;
 	}
 
-out_delete_evlist:
+out_put_evlist:
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/tests/switch-tracking.c b/tools/perf/tests/switch-tracking.c
index 72a8289e846d..306151c83af8 100644
--- a/tools/perf/tests/switch-tracking.c
+++ b/tools/perf/tests/switch-tracking.c
@@ -579,7 +579,7 @@ static int test__switch_tracking(struct test_suite *test __maybe_unused, int sub
 out:
 	if (evlist) {
 		evlist__disable(evlist);
-		evlist__delete(evlist);
+		evlist__put(evlist);
 	}
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
diff --git a/tools/perf/tests/task-exit.c b/tools/perf/tests/task-exit.c
index 4053ff2813bb..a46650b10689 100644
--- a/tools/perf/tests/task-exit.c
+++ b/tools/perf/tests/task-exit.c
@@ -74,7 +74,7 @@ static int test__task_exit(struct test_suite *test __maybe_unused, int subtest _
 	if (!cpus || !threads) {
 		err = -ENOMEM;
 		pr_debug("Not enough memory to create thread/cpu maps\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	perf_evlist__set_maps(&evlist->core, cpus, threads);
@@ -82,7 +82,7 @@ static int test__task_exit(struct test_suite *test __maybe_unused, int subtest _
 	err = evlist__prepare_workload(evlist, &target, argv, false, workload_exec_failed_signal);
 	if (err < 0) {
 		pr_debug("Couldn't run the workload!\n");
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evsel = evlist__first(evlist);
@@ -101,14 +101,14 @@ static int test__task_exit(struct test_suite *test __maybe_unused, int subtest _
 	if (err < 0) {
 		pr_debug("Couldn't open the evlist: %s\n",
 			 str_error_r(-err, sbuf, sizeof(sbuf)));
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	if (evlist__mmap(evlist, 128) < 0) {
 		pr_debug("failed to mmap events: %d (%s)\n", errno,
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		err = -1;
-		goto out_delete_evlist;
+		goto out_put_evlist;
 	}
 
 	evlist__start_workload(evlist);
@@ -133,7 +133,7 @@ static int test__task_exit(struct test_suite *test __maybe_unused, int subtest _
 		if (retry_count++ > 1000) {
 			pr_debug("Failed after retrying 1000 times\n");
 			err = -1;
-			goto out_delete_evlist;
+			goto out_put_evlist;
 		}
 
 		goto retry;
@@ -144,10 +144,10 @@ static int test__task_exit(struct test_suite *test __maybe_unused, int subtest _
 		err = -1;
 	}
 
-out_delete_evlist:
+out_put_evlist:
 	perf_cpu_map__put(cpus);
 	perf_thread_map__put(threads);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/tests/tool_pmu.c b/tools/perf/tests/tool_pmu.c
index 1e900ef92e37..e78ff9dcea97 100644
--- a/tools/perf/tests/tool_pmu.c
+++ b/tools/perf/tests/tool_pmu.c
@@ -67,7 +67,7 @@ static int do_test(enum tool_pmu_event ev, bool with_pmu)
 
 out:
 	parse_events_error__exit(&err);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return ret;
 }
 
diff --git a/tools/perf/tests/topology.c b/tools/perf/tests/topology.c
index f54502ebef4b..4ecf5d750313 100644
--- a/tools/perf/tests/topology.c
+++ b/tools/perf/tests/topology.c
@@ -57,7 +57,7 @@ static int session_write_header(char *path)
 			!perf_session__write_header(session, session->evlist,
 						    perf_data__fd(&data), true));
 
-	evlist__delete(session->evlist);
+	evlist__put(session->evlist);
 	perf_session__delete(session);
 
 	return 0;
diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c
index 1b5664d1481f..652a45aac828 100644
--- a/tools/perf/util/cgroup.c
+++ b/tools/perf/util/cgroup.c
@@ -520,8 +520,8 @@ int evlist__expand_cgroup(struct evlist *evlist, const char *str, bool open_cgro
 	cgrp_event_expanded = true;
 
 out_err:
-	evlist__delete(orig_list);
-	evlist__delete(tmp_list);
+	evlist__put(orig_list);
+	evlist__put(tmp_list);
 	metricgroup__rblist_exit(&orig_metric_events);
 	release_cgroup_list();
 
diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 3b8f2df823a9..a85ae53db7c5 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -1363,7 +1363,7 @@ static void cleanup_events(struct perf_session *session)
 		zfree(&evsel->priv);
 	}
 
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	session->evlist = NULL;
 }
 
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 35d65fe50e06..b5a7895debf5 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -75,7 +75,7 @@ int sigqueue(pid_t pid, int sig, const union sigval value);
 #define FD(e, x, y) (*(int *)xyarray__entry(e->core.fd, x, y))
 #define SID(e, x, y) xyarray__entry(e->core.sample_id, x, y)
 
-void evlist__init(struct evlist *evlist, struct perf_cpu_map *cpus,
+static void evlist__init(struct evlist *evlist, struct perf_cpu_map *cpus,
 		  struct perf_thread_map *threads)
 {
 	perf_evlist__init(&evlist->core);
@@ -88,6 +88,7 @@ void evlist__init(struct evlist *evlist, struct perf_cpu_map *cpus,
 	evlist->nr_br_cntr = -1;
 	metricgroup__rblist_init(&evlist->metric_events);
 	INIT_LIST_HEAD(&evlist->deferred_samples);
+	refcount_set(&evlist->refcnt, 1);
 }
 
 struct evlist *evlist__new(void)
@@ -139,7 +140,7 @@ struct evlist *evlist__new_default(const struct target *target, bool sample_call
 
 	return evlist;
 out_err:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return NULL;
 }
 
@@ -148,13 +149,19 @@ struct evlist *evlist__new_dummy(void)
 	struct evlist *evlist = evlist__new();
 
 	if (evlist && evlist__add_dummy(evlist)) {
-		evlist__delete(evlist);
+		evlist__put(evlist);
 		evlist = NULL;
 	}
 
 	return evlist;
 }
 
+struct evlist *evlist__get(struct evlist *evlist)
+{
+	refcount_inc(&evlist->refcnt);
+	return evlist;
+}
+
 /**
  * evlist__set_id_pos - set the positions of event ids.
  * @evlist: selected event list
@@ -193,7 +200,7 @@ static void evlist__purge(struct evlist *evlist)
 	evlist->core.nr_entries = 0;
 }
 
-void evlist__exit(struct evlist *evlist)
+static void evlist__exit(struct evlist *evlist)
 {
 	metricgroup__rblist_exit(&evlist->metric_events);
 	event_enable_timer__exit(&evlist->eet);
@@ -202,11 +209,14 @@ void evlist__exit(struct evlist *evlist)
 	perf_evlist__exit(&evlist->core);
 }
 
-void evlist__delete(struct evlist *evlist)
+void evlist__put(struct evlist *evlist)
 {
 	if (evlist == NULL)
 		return;
 
+	if (!refcount_dec_and_test(&evlist->refcnt))
+		return;
+
 	evlist__free_stats(evlist);
 	evlist__munmap(evlist);
 	evlist__close(evlist);
diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index e54761c670b6..a9820a6aad5b 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -61,6 +61,7 @@ struct event_enable_timer;
 
 struct evlist {
 	struct perf_evlist core;
+	refcount_t	 refcnt;
 	bool		 enabled;
 	bool		 no_affinity;
 	int		 id_pos;
@@ -109,10 +110,8 @@ struct evsel_str_handler {
 struct evlist *evlist__new(void);
 struct evlist *evlist__new_default(const struct target *target, bool sample_callchains);
 struct evlist *evlist__new_dummy(void);
-void evlist__init(struct evlist *evlist, struct perf_cpu_map *cpus,
-		  struct perf_thread_map *threads);
-void evlist__exit(struct evlist *evlist);
-void evlist__delete(struct evlist *evlist);
+struct evlist *evlist__get(struct evlist *evlist);
+void evlist__put(struct evlist *evlist);
 
 void evlist__add(struct evlist *evlist, struct evsel *entry);
 void evlist__remove(struct evlist *evlist, struct evsel *evsel);
diff --git a/tools/perf/util/expr.c b/tools/perf/util/expr.c
index 644769e92708..cf54bbbc8ddc 100644
--- a/tools/perf/util/expr.c
+++ b/tools/perf/util/expr.c
@@ -450,7 +450,7 @@ double expr__has_event(const struct expr_parse_ctx *ctx, bool compute_ids, const
 		ret = parse_event(tmp, id) ? 0 : 1;
 	}
 out:
-	evlist__delete(tmp);
+	evlist__put(tmp);
 	return ret;
 }
 
diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index f30e48eb3fc3..f9887d2fc8ed 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -4909,12 +4909,12 @@ int perf_session__read_header(struct perf_session *session)
 		evsel = evsel__new(&f_attr.attr);
 
 		if (evsel == NULL)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 
 		evsel->needs_swap = header->needs_swap;
 		/*
 		 * Do it before so that if perf_evsel__alloc_id fails, this
-		 * entry gets purged too at evlist__delete().
+		 * entry gets purged too at evlist__put().
 		 */
 		evlist__add(session->evlist, evsel);
 
@@ -4925,7 +4925,7 @@ int perf_session__read_header(struct perf_session *session)
 		 * hattr->ids threads.
 		 */
 		if (perf_evsel__alloc_id(&evsel->core, 1, nr_ids))
-			goto out_delete_evlist;
+			goto out_put_evlist;
 
 		lseek(fd, f_attr.ids.offset, SEEK_SET);
 
@@ -4944,7 +4944,7 @@ int perf_session__read_header(struct perf_session *session)
 				      perf_file_section__process);
 
 	if (evlist__prepare_tracepoint_events(session->evlist, session->tevent.pevent))
-		goto out_delete_evlist;
+		goto out_put_evlist;
 #else
 	perf_header__process_sections(header, fd, NULL, perf_file_section__process);
 #endif
@@ -4953,8 +4953,8 @@ int perf_session__read_header(struct perf_session *session)
 out_errno:
 	return -errno;
 
-out_delete_evlist:
-	evlist__delete(session->evlist);
+out_put_evlist:
+	evlist__put(session->evlist);
 	session->evlist = NULL;
 	return -ENOMEM;
 }
diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c
index 4db9578efd81..191ec2d8a250 100644
--- a/tools/perf/util/metricgroup.c
+++ b/tools/perf/util/metricgroup.c
@@ -214,7 +214,7 @@ static void metric__free(struct metric *m)
 	zfree(&m->metric_refs);
 	expr__ctx_free(m->pctx);
 	zfree(&m->modifier);
-	evlist__delete(m->evlist);
+	evlist__put(m->evlist);
 	free(m);
 }
 
@@ -1335,7 +1335,7 @@ static int parse_ids(bool metric_no_merge, bool fake_pmu,
 	parsed_evlist = NULL;
 err_out:
 	parse_events_error__exit(&parse_error);
-	evlist__delete(parsed_evlist);
+	evlist__put(parsed_evlist);
 	strbuf_release(&events);
 	return ret;
 }
@@ -1546,7 +1546,7 @@ static int parse_groups(struct evlist *perf_evlist,
 
 	if (combined_evlist) {
 		evlist__splice_list_tail(perf_evlist, &combined_evlist->core.entries);
-		evlist__delete(combined_evlist);
+		evlist__put(combined_evlist);
 	}
 
 	list_for_each_entry(m, &metric_list, nd) {
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index 1497e1f2a08c..f0809be63ad8 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -2316,7 +2316,7 @@ int __parse_events(struct evlist *evlist, const char *str, const char *pmu_filte
 
 	/*
 	 * There are 2 users - builtin-record and builtin-test objects.
-	 * Both call evlist__delete in case of error, so we dont
+	 * Both call evlist__put in case of error, so we dont
 	 * need to bother.
 	 */
 	return ret;
@@ -2519,7 +2519,7 @@ int parse_events_option_new_evlist(const struct option *opt, const char *str, in
 	}
 	ret = parse_events_option(opt, str, unset);
 	if (ret) {
-		evlist__delete(*args->evlistp);
+		evlist__put(*args->evlistp);
 		*args->evlistp = NULL;
 	}
 
diff --git a/tools/perf/util/perf_api_probe.c b/tools/perf/util/perf_api_probe.c
index e1904a330b28..f61c4ec52827 100644
--- a/tools/perf/util/perf_api_probe.c
+++ b/tools/perf/util/perf_api_probe.c
@@ -57,7 +57,7 @@ static int perf_do_probe_api(setup_probe_fn_t fn, struct perf_cpu cpu, const cha
 	err = 0;
 
 out_delete:
-	evlist__delete(evlist);
+	evlist__put(evlist);
 	return err;
 }
 
diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 1e6c99efff90..3f0758d5bd90 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -1272,7 +1272,7 @@ static int pyrf_evsel__setup_types(void)
 struct pyrf_evlist {
 	PyObject_HEAD
 
-	struct evlist evlist;
+	struct evlist *evlist;
 };
 
 static int pyrf_evlist__init(struct pyrf_evlist *pevlist,
@@ -1285,15 +1285,22 @@ static int pyrf_evlist__init(struct pyrf_evlist *pevlist,
 	if (!PyArg_ParseTuple(args, "OO", &pcpus, &pthreads))
 		return -1;
 
+	evlist__put(pevlist->evlist);
+	pevlist->evlist = evlist__new();
+	if (!pevlist->evlist) {
+		PyErr_NoMemory();
+		return -1;
+	}
 	threads = ((struct pyrf_thread_map *)pthreads)->threads;
 	cpus = ((struct pyrf_cpu_map *)pcpus)->cpus;
-	evlist__init(&pevlist->evlist, cpus, threads);
+	perf_evlist__set_maps(&pevlist->evlist->core, cpus, threads);
+
 	return 0;
 }
 
 static void pyrf_evlist__delete(struct pyrf_evlist *pevlist)
 {
-	evlist__exit(&pevlist->evlist);
+	evlist__put(pevlist->evlist);
 	Py_TYPE(pevlist)->tp_free((PyObject*)pevlist);
 }
 
@@ -1302,7 +1309,7 @@ static PyObject *pyrf_evlist__all_cpus(struct pyrf_evlist *pevlist)
 	struct pyrf_cpu_map *pcpu_map = PyObject_New(struct pyrf_cpu_map, &pyrf_cpu_map__type);
 
 	if (pcpu_map)
-		pcpu_map->cpus = perf_cpu_map__get(pevlist->evlist.core.all_cpus);
+		pcpu_map->cpus = perf_cpu_map__get(pevlist->evlist->core.all_cpus);
 
 	return (PyObject *)pcpu_map;
 }
@@ -1315,7 +1322,7 @@ static PyObject *pyrf_evlist__metrics(struct pyrf_evlist *pevlist)
 	if (!list)
 		return NULL;
 
-	for (node = rb_first_cached(&pevlist->evlist.metric_events.entries); node;
+	for (node = rb_first_cached(&pevlist->evlist->metric_events.entries); node;
 	     node = rb_next(node)) {
 		struct metric_event *me = container_of(node, struct metric_event, nd);
 		struct list_head *pos;
@@ -1421,7 +1428,7 @@ static PyObject *pyrf_evlist__compute_metric(struct pyrf_evlist *pevlist,
 	if (!PyArg_ParseTuple(args, "sii", &metric, &cpu, &thread))
 		return NULL;
 
-	for (node = rb_first_cached(&pevlist->evlist.metric_events.entries);
+	for (node = rb_first_cached(&pevlist->evlist->metric_events.entries);
 	     mexp == NULL && node;
 	     node = rb_next(node)) {
 		struct metric_event *me = container_of(node, struct metric_event, nd);
@@ -1437,7 +1444,7 @@ static PyObject *pyrf_evlist__compute_metric(struct pyrf_evlist *pevlist,
 			if (e->metric_events[0] == NULL)
 				continue;
 
-			evlist__for_each_entry(&pevlist->evlist, pos2) {
+			evlist__for_each_entry(pevlist->evlist, pos2) {
 				if (pos2->metric_leader != e->metric_events[0])
 					continue;
 				cpu_idx = perf_cpu_map__idx(pos2->core.cpus,
@@ -1482,7 +1489,7 @@ static PyObject *pyrf_evlist__compute_metric(struct pyrf_evlist *pevlist,
 static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist,
 				   PyObject *args, PyObject *kwargs)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 	static char *kwlist[] = { "pages", "overwrite", NULL };
 	int pages = 128, overwrite = false;
 
@@ -1502,7 +1509,7 @@ static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist,
 static PyObject *pyrf_evlist__poll(struct pyrf_evlist *pevlist,
 				   PyObject *args, PyObject *kwargs)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 	static char *kwlist[] = { "timeout", NULL };
 	int timeout = -1, n;
 
@@ -1522,7 +1529,7 @@ static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist,
 					 PyObject *args __maybe_unused,
 					 PyObject *kwargs __maybe_unused)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
         PyObject *list = PyList_New(0);
 	int i;
 
@@ -1551,7 +1558,7 @@ static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist,
 				  PyObject *args,
 				  PyObject *kwargs __maybe_unused)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 	PyObject *pevsel;
 	struct evsel *evsel;
 
@@ -1583,7 +1590,7 @@ static struct mmap *get_md(struct evlist *evlist, int cpu)
 static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist,
 					  PyObject *args, PyObject *kwargs)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 	union perf_event *event;
 	int sample_id_all = 1, cpu;
 	static char *kwlist[] = { "cpu", "sample_id_all", NULL };
@@ -1640,7 +1647,7 @@ static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist,
 static PyObject *pyrf_evlist__open(struct pyrf_evlist *pevlist,
 				   PyObject *args, PyObject *kwargs)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 
 	if (evlist__open(evlist) < 0) {
 		PyErr_SetFromErrno(PyExc_OSError);
@@ -1653,7 +1660,7 @@ static PyObject *pyrf_evlist__open(struct pyrf_evlist *pevlist,
 
 static PyObject *pyrf_evlist__close(struct pyrf_evlist *pevlist)
 {
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 
 	evlist__close(evlist);
 
@@ -1679,7 +1686,7 @@ static PyObject *pyrf_evlist__config(struct pyrf_evlist *pevlist)
 		.no_buffering        = true,
 		.no_inherit          = true,
 	};
-	struct evlist *evlist = &pevlist->evlist;
+	struct evlist *evlist = pevlist->evlist;
 
 	evlist__config(evlist, &opts, &callchain_param);
 	Py_INCREF(Py_None);
@@ -1688,14 +1695,14 @@ static PyObject *pyrf_evlist__config(struct pyrf_evlist *pevlist)
 
 static PyObject *pyrf_evlist__disable(struct pyrf_evlist *pevlist)
 {
-	evlist__disable(&pevlist->evlist);
+	evlist__disable(pevlist->evlist);
 	Py_INCREF(Py_None);
 	return Py_None;
 }
 
 static PyObject *pyrf_evlist__enable(struct pyrf_evlist *pevlist)
 {
-	evlist__enable(&pevlist->evlist);
+	evlist__enable(pevlist->evlist);
 	Py_INCREF(Py_None);
 	return Py_None;
 }
@@ -1786,7 +1793,23 @@ static Py_ssize_t pyrf_evlist__length(PyObject *obj)
 {
 	struct pyrf_evlist *pevlist = (void *)obj;
 
-	return pevlist->evlist.core.nr_entries;
+	return pevlist->evlist->core.nr_entries;
+}
+
+static PyObject *pyrf_evsel__from_evsel(struct evsel *evsel)
+{
+	struct pyrf_evsel *pevsel = PyObject_New(struct pyrf_evsel, &pyrf_evsel__type);
+
+	if (!pevsel)
+		return NULL;
+
+	memset(&pevsel->evsel, 0, sizeof(pevsel->evsel));
+	evsel__init(&pevsel->evsel, &evsel->core.attr, evsel->core.idx);
+
+	evsel__clone(&pevsel->evsel, evsel);
+	if (evsel__is_group_leader(evsel))
+		evsel__set_leader(&pevsel->evsel, &pevsel->evsel);
+	return (PyObject *)pevsel;
 }
 
 static PyObject *pyrf_evlist__item(PyObject *obj, Py_ssize_t i)
@@ -1794,17 +1817,16 @@ static PyObject *pyrf_evlist__item(PyObject *obj, Py_ssize_t i)
 	struct pyrf_evlist *pevlist = (void *)obj;
 	struct evsel *pos;
 
-	if (i >= pevlist->evlist.core.nr_entries) {
+	if (i >= pevlist->evlist->core.nr_entries) {
 		PyErr_SetString(PyExc_IndexError, "Index out of range");
 		return NULL;
 	}
 
-	evlist__for_each_entry(&pevlist->evlist, pos) {
+	evlist__for_each_entry(pevlist->evlist, pos) {
 		if (i-- == 0)
 			break;
 	}
-
-	return Py_BuildValue("O", container_of(pos, struct pyrf_evsel, evsel));
+	return pyrf_evsel__from_evsel(pos);
 }
 
 static PyObject *pyrf_evlist__str(PyObject *self)
@@ -1816,7 +1838,7 @@ static PyObject *pyrf_evlist__str(PyObject *self)
 	PyObject *result;
 
 	strbuf_addstr(&sb, "evlist([");
-	evlist__for_each_entry(&pevlist->evlist, pos) {
+	evlist__for_each_entry(pevlist->evlist, pos) {
 		if (!first)
 			strbuf_addch(&sb, ',');
 		if (!pos->pmu)
@@ -1957,157 +1979,74 @@ static PyObject *pyrf__tracepoint(struct pyrf_evsel *pevsel,
 	return PyLong_FromLong(tp_pmu__id(sys, name));
 }
 
-static PyObject *pyrf_evsel__from_evsel(struct evsel *evsel)
-{
-	struct pyrf_evsel *pevsel = PyObject_New(struct pyrf_evsel, &pyrf_evsel__type);
-
-	if (!pevsel)
-		return NULL;
-
-	memset(&pevsel->evsel, 0, sizeof(pevsel->evsel));
-	evsel__init(&pevsel->evsel, &evsel->core.attr, evsel->core.idx);
-
-	evsel__clone(&pevsel->evsel, evsel);
-	if (evsel__is_group_leader(evsel))
-		evsel__set_leader(&pevsel->evsel, &pevsel->evsel);
-	return (PyObject *)pevsel;
-}
-
-static int evlist__pos(struct evlist *evlist, struct evsel *evsel)
-{
-	struct evsel *pos;
-	int idx = 0;
-
-	evlist__for_each_entry(evlist, pos) {
-		if (evsel == pos)
-			return idx;
-		idx++;
-	}
-	return -1;
-}
-
-static struct evsel *evlist__at(struct evlist *evlist, int idx)
-{
-	struct evsel *pos;
-	int idx2 = 0;
-
-	evlist__for_each_entry(evlist, pos) {
-		if (idx == idx2)
-			return pos;
-		idx2++;
-	}
-	return NULL;
-}
-
 static PyObject *pyrf_evlist__from_evlist(struct evlist *evlist)
 {
 	struct pyrf_evlist *pevlist = PyObject_New(struct pyrf_evlist, &pyrf_evlist__type);
-	struct evsel *pos;
-	struct rb_node *node;
 
 	if (!pevlist)
 		return NULL;
 
-	memset(&pevlist->evlist, 0, sizeof(pevlist->evlist));
-	evlist__init(&pevlist->evlist, evlist->core.all_cpus, evlist->core.threads);
-	evlist__for_each_entry(evlist, pos) {
-		struct pyrf_evsel *pevsel = (void *)pyrf_evsel__from_evsel(pos);
-
-		evlist__add(&pevlist->evlist, &pevsel->evsel);
-	}
-	evlist__for_each_entry(&pevlist->evlist, pos) {
-		struct evsel *leader = evsel__leader(pos);
-
-		if (pos != leader) {
-			int idx = evlist__pos(evlist, leader);
-
-			if (idx >= 0)
-				evsel__set_leader(pos, evlist__at(&pevlist->evlist, idx));
-			else if (leader == NULL)
-				evsel__set_leader(pos, pos);
-		}
-
-		leader = pos->metric_leader;
-
-		if (pos != leader) {
-			int idx = evlist__pos(evlist, leader);
-
-			if (idx >= 0)
-				pos->metric_leader = evlist__at(&pevlist->evlist, idx);
-			else if (leader == NULL)
-				pos->metric_leader = pos;
-		}
-	}
-	metricgroup__copy_metric_events(&pevlist->evlist, /*cgrp=*/NULL,
-					&pevlist->evlist.metric_events,
-					&evlist->metric_events);
-	for (node = rb_first_cached(&pevlist->evlist.metric_events.entries); node;
-	     node = rb_next(node)) {
-		struct metric_event *me = container_of(node, struct metric_event, nd);
-		struct list_head *mpos;
-		int idx = evlist__pos(evlist, me->evsel);
-
-		if (idx >= 0)
-			me->evsel = evlist__at(&pevlist->evlist, idx);
-		list_for_each(mpos, &me->head) {
-			struct metric_expr *e = container_of(mpos, struct metric_expr, nd);
-
-			for (int j = 0; e->metric_events[j]; j++) {
-				idx = evlist__pos(evlist, e->metric_events[j]);
-				if (idx >= 0)
-					e->metric_events[j] = evlist__at(&pevlist->evlist, idx);
-			}
-		}
-	}
+	pevlist->evlist = evlist__get(evlist);
 	return (PyObject *)pevlist;
 }
 
 static PyObject *pyrf__parse_events(PyObject *self, PyObject *args)
 {
 	const char *input;
-	struct evlist evlist = {};
+	struct evlist *evlist = evlist__new();
 	struct parse_events_error err;
 	PyObject *result;
 	PyObject *pcpus = NULL, *pthreads = NULL;
 	struct perf_cpu_map *cpus;
 	struct perf_thread_map *threads;
 
-	if (!PyArg_ParseTuple(args, "s|OO", &input, &pcpus, &pthreads))
+	if (!evlist)
+		return PyErr_NoMemory();
+
+	if (!PyArg_ParseTuple(args, "s|OO", &input, &pcpus, &pthreads)) {
+		evlist__put(evlist);
 		return NULL;
+	}
 
 	threads = pthreads ? ((struct pyrf_thread_map *)pthreads)->threads : NULL;
 	cpus = pcpus ? ((struct pyrf_cpu_map *)pcpus)->cpus : NULL;
 
 	parse_events_error__init(&err);
-	evlist__init(&evlist, cpus, threads);
-	if (parse_events(&evlist, input, &err)) {
+	perf_evlist__set_maps(&evlist->core, cpus, threads);
+	if (parse_events(evlist, input, &err)) {
 		parse_events_error__print(&err, input);
 		PyErr_SetFromErrno(PyExc_OSError);
+		evlist__put(evlist);
 		return NULL;
 	}
-	result = pyrf_evlist__from_evlist(&evlist);
-	evlist__exit(&evlist);
+	result = pyrf_evlist__from_evlist(evlist);
+	evlist__put(evlist);
 	return result;
 }
 
 static PyObject *pyrf__parse_metrics(PyObject *self, PyObject *args)
 {
 	const char *input, *pmu = NULL;
-	struct evlist evlist = {};
+	struct evlist *evlist = evlist__new();
 	PyObject *result;
 	PyObject *pcpus = NULL, *pthreads = NULL;
 	struct perf_cpu_map *cpus;
 	struct perf_thread_map *threads;
 	int ret;
 
-	if (!PyArg_ParseTuple(args, "s|sOO", &input, &pmu, &pcpus, &pthreads))
+	if (!evlist)
+		return PyErr_NoMemory();
+
+	if (!PyArg_ParseTuple(args, "s|sOO", &input, &pmu, &pcpus, &pthreads)) {
+		evlist__put(evlist);
 		return NULL;
+	}
 
 	threads = pthreads ? ((struct pyrf_thread_map *)pthreads)->threads : NULL;
 	cpus = pcpus ? ((struct pyrf_cpu_map *)pcpus)->cpus : NULL;
 
-	evlist__init(&evlist, cpus, threads);
-	ret = metricgroup__parse_groups(&evlist, pmu ?: "all", input,
+	perf_evlist__set_maps(&evlist->core, cpus, threads);
+	ret = metricgroup__parse_groups(evlist, pmu ?: "all", input,
 					/*metric_no_group=*/ false,
 					/*metric_no_merge=*/ false,
 					/*metric_no_threshold=*/ true,
@@ -2115,12 +2054,13 @@ static PyObject *pyrf__parse_metrics(PyObject *self, PyObject *args)
 					/*system_wide=*/true,
 					/*hardware_aware_grouping=*/ false);
 	if (ret) {
+		evlist__put(evlist);
 		errno = -ret;
 		PyErr_SetFromErrno(PyExc_OSError);
 		return NULL;
 	}
-	result = pyrf_evlist__from_evlist(&evlist);
-	evlist__exit(&evlist);
+	result = pyrf_evlist__from_evlist(evlist);
+	evlist__put(evlist);
 	return result;
 }
 
diff --git a/tools/perf/util/record.c b/tools/perf/util/record.c
index e867de8ddaaa..8a5fc7d5e43c 100644
--- a/tools/perf/util/record.c
+++ b/tools/perf/util/record.c
@@ -264,7 +264,7 @@ bool evlist__can_select_event(struct evlist *evlist, const char *str)
 	ret = true;
 
 out_delete:
-	evlist__delete(temp_evlist);
+	evlist__put(temp_evlist);
 	return ret;
 }
 
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index fe0de2a0277f..1ac6cd43c38b 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -264,7 +264,7 @@ void perf_session__delete(struct perf_session *session)
 	machines__exit(&session->machines);
 	if (session->data) {
 		if (perf_data__is_read(session->data))
-			evlist__delete(session->evlist);
+			evlist__put(session->evlist);
 		perf_data__close(session->data);
 	}
 #ifdef HAVE_LIBTRACEEVENT
diff --git a/tools/perf/util/sideband_evlist.c b/tools/perf/util/sideband_evlist.c
index 388846f17bc1..b84a5463e039 100644
--- a/tools/perf/util/sideband_evlist.c
+++ b/tools/perf/util/sideband_evlist.c
@@ -102,7 +102,7 @@ int evlist__start_sb_thread(struct evlist *evlist, struct target *target)
 		return 0;
 
 	if (evlist__create_maps(evlist, target))
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	if (evlist->core.nr_entries > 1) {
 		bool can_sample_identifier = perf_can_sample_identifier();
@@ -116,25 +116,25 @@ int evlist__start_sb_thread(struct evlist *evlist, struct target *target)
 	evlist__for_each_entry(evlist, counter) {
 		if (evsel__open(counter, evlist->core.user_requested_cpus,
 				evlist->core.threads) < 0)
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 
 	if (evlist__mmap(evlist, UINT_MAX))
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	evlist__for_each_entry(evlist, counter) {
 		if (evsel__enable(counter))
-			goto out_delete_evlist;
+			goto out_put_evlist;
 	}
 
 	evlist->thread.done = 0;
 	if (pthread_create(&evlist->thread.th, NULL, perf_evlist__poll_thread, evlist))
-		goto out_delete_evlist;
+		goto out_put_evlist;
 
 	return 0;
 
-out_delete_evlist:
-	evlist__delete(evlist);
+out_put_evlist:
+	evlist__put(evlist);
 	evlist = NULL;
 	return -1;
 }
@@ -145,5 +145,5 @@ void evlist__stop_sb_thread(struct evlist *evlist)
 		return;
 	evlist->thread.done = 1;
 	pthread_join(evlist->thread.th, NULL);
-	evlist__delete(evlist);
+	evlist__put(evlist);
 }
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 11/58] perf evsel: Add reference count
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

As with evlist this a no-op for most of the perf tool. The reference
count is set to 1 at allocation, the put will see the 1, decrement it
and perform the delete. The purpose for adding the reference count is
for the python code. Prior to this change the python code would clone
evsels, but this has issues if events are opened, etc. leading to
assertion failures. With a reference count the same evsel can be used
and the reference count incremented for the python usage.  To not
change the python evsel API getset functions are added for the evsel
members, no set function is provided for size as it doesn't make sense
to alter this.

Signed-off-by: Ian Rogers <irogers@google.com>
---
v2:
1. Fixed Potential Crash in pyrf_event__new : Initialized
   pevent->evsel = NULL; to avoid garbage pointer dereference if
   evlist__event2evsel() fails in read_on_cpu .

2. Fixed Memory Leak in pyrf_evsel__init : Added
   evsel__put(pevsel->evsel) before overwriting it to handle repeated
   __init__ calls.

3. Fixed Exception Contract: Added PyErr_NoMemory() when evsel__new()
   fails in pyrf_evsel__init .

4. Fixed NULL Pointer Dereference on Property Access: Added a custom
   tp_getattro ( pyrf_evsel__getattro ) to pyrf_evsel__type to check
   if pevsel->evsel is NULL and raise a ValueError if so, covering all
   property accesses.

5. Fixed Reference Count in pyrf_evlist__add : Added evsel__get(evsel)
   when adding to the evlist .

6. Fixed Reference Count in pyrf_evlist__read_on_cpu : Added
   evsel__get(evsel) when assigning to pevent->evsel .
---
 tools/perf/builtin-trace.c                 |  12 +-
 tools/perf/tests/evsel-tp-sched.c          |   4 +-
 tools/perf/tests/openat-syscall-all-cpus.c |   6 +-
 tools/perf/tests/openat-syscall.c          |   6 +-
 tools/perf/util/bpf_counter_cgroup.c       |   2 +-
 tools/perf/util/cgroup.c                   |   2 +-
 tools/perf/util/evlist.c                   |   2 +-
 tools/perf/util/evsel.c                    |  26 ++-
 tools/perf/util/evsel.h                    |  11 +-
 tools/perf/util/parse-events.y             |   2 +-
 tools/perf/util/pfm.c                      |   2 +-
 tools/perf/util/print-events.c             |   2 +-
 tools/perf/util/python.c                   | 231 +++++++++++++++++----
 tools/perf/util/session.c                  |   1 +
 14 files changed, 237 insertions(+), 72 deletions(-)

diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index da703d762433..6ea935c13538 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -448,10 +448,10 @@ static int evsel__init_tp_ptr_field(struct evsel *evsel, struct tp_field *field,
 	({ struct syscall_tp *sc = __evsel__syscall_tp(evsel);\
 	   evsel__init_tp_ptr_field(evsel, &sc->name, #name); })
 
-static void evsel__delete_priv(struct evsel *evsel)
+static void evsel__put_and_free_priv(struct evsel *evsel)
 {
 	zfree(&evsel->priv);
-	evsel__delete(evsel);
+	evsel__put(evsel);
 }
 
 static int evsel__init_syscall_tp(struct evsel *evsel)
@@ -531,7 +531,7 @@ static struct evsel *perf_evsel__raw_syscall_newtp(const char *direction, void *
 	return evsel;
 
 out_delete:
-	evsel__delete_priv(evsel);
+	evsel__put_and_free_priv(evsel);
 	return NULL;
 }
 
@@ -3584,7 +3584,7 @@ static bool evlist__add_vfs_getname(struct evlist *evlist)
 
 		list_del_init(&evsel->core.node);
 		evsel->evlist = NULL;
-		evsel__delete(evsel);
+		evsel__put(evsel);
 	}
 
 	return found;
@@ -3698,9 +3698,9 @@ static int trace__add_syscall_newtp(struct trace *trace)
 	return ret;
 
 out_delete_sys_exit:
-	evsel__delete_priv(sys_exit);
+	evsel__put_and_free_priv(sys_exit);
 out_delete_sys_enter:
-	evsel__delete_priv(sys_enter);
+	evsel__put_and_free_priv(sys_enter);
 	goto out;
 }
 
diff --git a/tools/perf/tests/evsel-tp-sched.c b/tools/perf/tests/evsel-tp-sched.c
index 226196fb9677..9e456f88a13a 100644
--- a/tools/perf/tests/evsel-tp-sched.c
+++ b/tools/perf/tests/evsel-tp-sched.c
@@ -64,7 +64,7 @@ static int test__perf_evsel__tp_sched_test(struct test_suite *test __maybe_unuse
 	if (evsel__test_field(evsel, "next_prio", 4, true))
 		ret = TEST_FAIL;
 
-	evsel__delete(evsel);
+	evsel__put(evsel);
 
 	evsel = evsel__newtp("sched", "sched_wakeup");
 
@@ -85,7 +85,7 @@ static int test__perf_evsel__tp_sched_test(struct test_suite *test __maybe_unuse
 	if (evsel__test_field(evsel, "target_cpu", 4, true))
 		ret = TEST_FAIL;
 
-	evsel__delete(evsel);
+	evsel__put(evsel);
 	return ret;
 }
 
diff --git a/tools/perf/tests/openat-syscall-all-cpus.c b/tools/perf/tests/openat-syscall-all-cpus.c
index 0be43f8db3bd..cc63df2b3bc5 100644
--- a/tools/perf/tests/openat-syscall-all-cpus.c
+++ b/tools/perf/tests/openat-syscall-all-cpus.c
@@ -59,7 +59,7 @@ static int test__openat_syscall_event_on_all_cpus(struct test_suite *test __mayb
 			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		err = TEST_SKIP;
-		goto out_evsel_delete;
+		goto out_evsel_put;
 	}
 
 	perf_cpu_map__for_each_cpu(cpu, idx, cpus) {
@@ -116,8 +116,8 @@ static int test__openat_syscall_event_on_all_cpus(struct test_suite *test __mayb
 	evsel__free_counts(evsel);
 out_close_fd:
 	perf_evsel__close_fd(&evsel->core);
-out_evsel_delete:
-	evsel__delete(evsel);
+out_evsel_put:
+	evsel__put(evsel);
 out_cpu_map_delete:
 	perf_cpu_map__put(cpus);
 out_thread_map_delete:
diff --git a/tools/perf/tests/openat-syscall.c b/tools/perf/tests/openat-syscall.c
index b54cbe5f1808..9f16f0dd3a29 100644
--- a/tools/perf/tests/openat-syscall.c
+++ b/tools/perf/tests/openat-syscall.c
@@ -42,7 +42,7 @@ static int test__openat_syscall_event(struct test_suite *test __maybe_unused,
 			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
 			 str_error_r(errno, sbuf, sizeof(sbuf)));
 		err = TEST_SKIP;
-		goto out_evsel_delete;
+		goto out_evsel_put;
 	}
 
 	for (i = 0; i < nr_openat_calls; ++i) {
@@ -64,8 +64,8 @@ static int test__openat_syscall_event(struct test_suite *test __maybe_unused,
 	err = TEST_OK;
 out_close_fd:
 	perf_evsel__close_fd(&evsel->core);
-out_evsel_delete:
-	evsel__delete(evsel);
+out_evsel_put:
+	evsel__put(evsel);
 out_thread_map_delete:
 	perf_thread_map__put(threads);
 	return err;
diff --git a/tools/perf/util/bpf_counter_cgroup.c b/tools/perf/util/bpf_counter_cgroup.c
index 519fee3dc3d0..339df94ef438 100644
--- a/tools/perf/util/bpf_counter_cgroup.c
+++ b/tools/perf/util/bpf_counter_cgroup.c
@@ -316,7 +316,7 @@ static int bperf_cgrp__destroy(struct evsel *evsel)
 		return 0;
 
 	bperf_cgroup_bpf__destroy(skel);
-	evsel__delete(cgrp_switch);  // it'll destroy on_switch progs too
+	evsel__put(cgrp_switch);  // it'll destroy on_switch progs too
 
 	return 0;
 }
diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c
index 652a45aac828..914744724467 100644
--- a/tools/perf/util/cgroup.c
+++ b/tools/perf/util/cgroup.c
@@ -469,7 +469,7 @@ int evlist__expand_cgroup(struct evlist *evlist, const char *str, bool open_cgro
 
 		/* copy the list and set to the new cgroup. */
 		evlist__for_each_entry(orig_list, pos) {
-			struct evsel *evsel = evsel__clone(/*dest=*/NULL, pos);
+			struct evsel *evsel = evsel__clone(pos);
 
 			if (evsel == NULL)
 				goto out_err;
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index b5a7895debf5..a362f338f104 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -194,7 +194,7 @@ static void evlist__purge(struct evlist *evlist)
 	evlist__for_each_entry_safe(evlist, n, pos) {
 		list_del_init(&pos->core.node);
 		pos->evlist = NULL;
-		evsel__delete(pos);
+		evsel__put(pos);
 	}
 
 	evlist->core.nr_entries = 0;
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index e03727d395e9..a54aae079c22 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -388,10 +388,11 @@ bool evsel__is_function_event(struct evsel *evsel)
 #undef FUNCTION_EVENT
 }
 
-void evsel__init(struct evsel *evsel,
+static void evsel__init(struct evsel *evsel,
 		 struct perf_event_attr *attr, int idx)
 {
 	perf_evsel__init(&evsel->core, attr, idx);
+	refcount_set(&evsel->refcnt, 1);
 	evsel->tracking	   = !idx;
 	evsel->unit	   = strdup("");
 	evsel->scale	   = 1.0;
@@ -472,7 +473,7 @@ static int evsel__copy_config_terms(struct evsel *dst, struct evsel *src)
  * The assumption is that @orig is not configured nor opened yet.
  * So we only care about the attributes that can be set while it's parsed.
  */
-struct evsel *evsel__clone(struct evsel *dest, struct evsel *orig)
+struct evsel *evsel__clone(struct evsel *orig)
 {
 	struct evsel *evsel;
 
@@ -485,11 +486,7 @@ struct evsel *evsel__clone(struct evsel *dest, struct evsel *orig)
 	if (orig->bpf_obj)
 		return NULL;
 
-	if (dest)
-		evsel = dest;
-	else
-		evsel = evsel__new(&orig->core.attr);
-
+	evsel = evsel__new(&orig->core.attr);
 	if (evsel == NULL)
 		return NULL;
 
@@ -574,7 +571,7 @@ struct evsel *evsel__clone(struct evsel *dest, struct evsel *orig)
 	return evsel;
 
 out_err:
-	evsel__delete(evsel);
+	evsel__put(evsel);
 	return NULL;
 }
 
@@ -633,6 +630,12 @@ struct evsel *evsel__newtp_idx(const char *sys, const char *name, int idx, bool
 	return ERR_PTR(err);
 }
 
+struct evsel *evsel__get(struct evsel *evsel)
+{
+	refcount_inc(&evsel->refcnt);
+	return evsel;
+}
+
 #ifdef HAVE_LIBTRACEEVENT
 struct tep_event *evsel__tp_format(struct evsel *evsel)
 {
@@ -1857,7 +1860,7 @@ void evsel__set_priv_destructor(void (*destructor)(void *priv))
 	evsel__priv_destructor = destructor;
 }
 
-void evsel__exit(struct evsel *evsel)
+static void evsel__exit(struct evsel *evsel)
 {
 	assert(list_empty(&evsel->core.node));
 	assert(evsel->evlist == NULL);
@@ -1892,11 +1895,14 @@ void evsel__exit(struct evsel *evsel)
 		xyarray__delete(evsel->start_times);
 }
 
-void evsel__delete(struct evsel *evsel)
+void evsel__put(struct evsel *evsel)
 {
 	if (!evsel)
 		return;
 
+	if (!refcount_dec_and_test(&evsel->refcnt))
+		return;
+
 	evsel__exit(evsel);
 	free(evsel);
 }
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index b099c8e5dd86..35b1bbca9036 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -6,6 +6,7 @@
 
 #include <linux/list.h>
 #include <linux/perf_event.h>
+#include <linux/refcount.h>
 #include <linux/types.h>
 #include <sys/types.h>
 
@@ -47,6 +48,7 @@ typedef int (evsel__sb_cb_t)(union perf_event *event, void *data);
 struct evsel {
 	struct perf_evsel	core;
 	struct evlist		*evlist;
+	refcount_t		refcnt;
 	off_t			id_offset;
 	int			id_pos;
 	int			is_pos;
@@ -262,7 +264,7 @@ static inline struct evsel *evsel__new(struct perf_event_attr *attr)
 	return evsel__new_idx(attr, 0);
 }
 
-struct evsel *evsel__clone(struct evsel *dest, struct evsel *orig);
+struct evsel *evsel__clone(struct evsel *orig);
 
 int copy_config_terms(struct list_head *dst, struct list_head *src);
 void free_config_terms(struct list_head *config_terms);
@@ -277,14 +279,13 @@ static inline struct evsel *evsel__newtp(const char *sys, const char *name)
 	return evsel__newtp_idx(sys, name, 0, true);
 }
 
+struct evsel *evsel__get(struct evsel *evsel);
+void evsel__put(struct evsel *evsel);
+
 #ifdef HAVE_LIBTRACEEVENT
 struct tep_event *evsel__tp_format(struct evsel *evsel);
 #endif
 
-void evsel__init(struct evsel *evsel, struct perf_event_attr *attr, int idx);
-void evsel__exit(struct evsel *evsel);
-void evsel__delete(struct evsel *evsel);
-
 void evsel__set_priv_destructor(void (*destructor)(void *priv));
 
 struct callchain_param;
diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y
index c194de5ec1ec..b531b1f0ceb3 100644
--- a/tools/perf/util/parse-events.y
+++ b/tools/perf/util/parse-events.y
@@ -47,7 +47,7 @@ static void free_list_evsel(struct list_head* list_evsel)
 
 	list_for_each_entry_safe(evsel, tmp, list_evsel, core.node) {
 		list_del_init(&evsel->core.node);
-		evsel__delete(evsel);
+		evsel__put(evsel);
 	}
 	free(list_evsel);
 }
diff --git a/tools/perf/util/pfm.c b/tools/perf/util/pfm.c
index d9043f4afbe7..5f53c2f68a96 100644
--- a/tools/perf/util/pfm.c
+++ b/tools/perf/util/pfm.c
@@ -159,7 +159,7 @@ static bool is_libpfm_event_supported(const char *name, struct perf_cpu_map *cpu
 		result = false;
 
 	evsel__close(evsel);
-	evsel__delete(evsel);
+	evsel__put(evsel);
 
 	return result;
 }
diff --git a/tools/perf/util/print-events.c b/tools/perf/util/print-events.c
index cb27e2898aa0..0242243681b6 100644
--- a/tools/perf/util/print-events.c
+++ b/tools/perf/util/print-events.c
@@ -174,7 +174,7 @@ bool is_event_supported(u8 type, u64 config)
 		}
 
 		evsel__close(evsel);
-		evsel__delete(evsel);
+		evsel__put(evsel);
 	}
 
 	perf_thread_map__put(tmap);
diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 3f0758d5bd90..8585ae992e6b 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -274,6 +274,7 @@ static PyMemberDef pyrf_sample_event__members[] = {
 
 static void pyrf_sample_event__delete(struct pyrf_event *pevent)
 {
+	evsel__put(pevent->evsel);
 	perf_sample__exit(&pevent->sample);
 	Py_TYPE(pevent)->tp_free((PyObject*)pevent);
 }
@@ -506,8 +507,10 @@ static PyObject *pyrf_event__new(const union perf_event *event)
 
 	ptype = pyrf_event__type[event->header.type];
 	pevent = PyObject_New(struct pyrf_event, ptype);
-	if (pevent != NULL)
+	if (pevent != NULL) {
 		memcpy(&pevent->event, event, event->header.size);
+		pevent->evsel = NULL;
+	}
 	return (PyObject *)pevent;
 }
 
@@ -945,7 +948,7 @@ static int pyrf_counts_values__setup_types(void)
 struct pyrf_evsel {
 	PyObject_HEAD
 
-	struct evsel evsel;
+	struct evsel *evsel;
 };
 
 static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
@@ -1053,20 +1056,25 @@ static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
 	attr.sample_id_all  = sample_id_all;
 	attr.size	    = sizeof(attr);
 
-	evsel__init(&pevsel->evsel, &attr, idx);
+	evsel__put(pevsel->evsel);
+	pevsel->evsel = evsel__new(&attr);
+	if (!pevsel->evsel) {
+		PyErr_NoMemory();
+		return -1;
+	}
 	return 0;
 }
 
 static void pyrf_evsel__delete(struct pyrf_evsel *pevsel)
 {
-	evsel__exit(&pevsel->evsel);
+	evsel__put(pevsel->evsel);
 	Py_TYPE(pevsel)->tp_free((PyObject*)pevsel);
 }
 
 static PyObject *pyrf_evsel__open(struct pyrf_evsel *pevsel,
 				  PyObject *args, PyObject *kwargs)
 {
-	struct evsel *evsel = &pevsel->evsel;
+	struct evsel *evsel = pevsel->evsel;
 	struct perf_cpu_map *cpus = NULL;
 	struct perf_thread_map *threads = NULL;
 	PyObject *pcpus = NULL, *pthreads = NULL;
@@ -1102,7 +1110,7 @@ static PyObject *pyrf_evsel__cpus(struct pyrf_evsel *pevsel)
 	struct pyrf_cpu_map *pcpu_map = PyObject_New(struct pyrf_cpu_map, &pyrf_cpu_map__type);
 
 	if (pcpu_map)
-		pcpu_map->cpus = perf_cpu_map__get(pevsel->evsel.core.cpus);
+		pcpu_map->cpus = perf_cpu_map__get(pevsel->evsel->core.cpus);
 
 	return (PyObject *)pcpu_map;
 }
@@ -1113,7 +1121,7 @@ static PyObject *pyrf_evsel__threads(struct pyrf_evsel *pevsel)
 		PyObject_New(struct pyrf_thread_map, &pyrf_thread_map__type);
 
 	if (pthread_map)
-		pthread_map->threads = perf_thread_map__get(pevsel->evsel.core.threads);
+		pthread_map->threads = perf_thread_map__get(pevsel->evsel->core.threads);
 
 	return (PyObject *)pthread_map;
 }
@@ -1147,7 +1155,7 @@ static int evsel__ensure_counts(struct evsel *evsel)
 static PyObject *pyrf_evsel__read(struct pyrf_evsel *pevsel,
 				  PyObject *args, PyObject *kwargs)
 {
-	struct evsel *evsel = &pevsel->evsel;
+	struct evsel *evsel = pevsel->evsel;
 	int cpu = 0, cpu_idx, thread = 0, thread_idx;
 	struct perf_counts_values *old_count, *new_count;
 	struct pyrf_counts_values *count_values = PyObject_New(struct pyrf_counts_values,
@@ -1192,7 +1200,7 @@ static PyObject *pyrf_evsel__read(struct pyrf_evsel *pevsel,
 static PyObject *pyrf_evsel__str(PyObject *self)
 {
 	struct pyrf_evsel *pevsel = (void *)self;
-	struct evsel *evsel = &pevsel->evsel;
+	struct evsel *evsel = pevsel->evsel;
 
 	return PyUnicode_FromFormat("evsel(%s/%s/)", evsel__pmu_name(evsel), evsel__name(evsel));
 }
@@ -1225,30 +1233,183 @@ static PyMethodDef pyrf_evsel__methods[] = {
 	{ .ml_name = NULL, }
 };
 
-#define evsel_member_def(member, ptype, help) \
-	{ #member, ptype, \
-	  offsetof(struct pyrf_evsel, evsel.member), \
-	  0, help }
+static PyObject *pyrf_evsel__get_tracking(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
 
-#define evsel_attr_member_def(member, ptype, help) \
-	{ #member, ptype, \
-	  offsetof(struct pyrf_evsel, evsel.core.attr.member), \
-	  0, help }
+	if (pevsel->evsel->tracking)
+		Py_RETURN_TRUE;
+	else
+		Py_RETURN_FALSE;
+}
 
-static PyMemberDef pyrf_evsel__members[] = {
-	evsel_member_def(tracking, T_BOOL, "tracking event."),
-	evsel_attr_member_def(type, T_UINT, "attribute type."),
-	evsel_attr_member_def(size, T_UINT, "attribute size."),
-	evsel_attr_member_def(config, T_ULONGLONG, "attribute config."),
-	evsel_attr_member_def(sample_period, T_ULONGLONG, "attribute sample_period."),
-	evsel_attr_member_def(sample_type, T_ULONGLONG, "attribute sample_type."),
-	evsel_attr_member_def(read_format, T_ULONGLONG, "attribute read_format."),
-	evsel_attr_member_def(wakeup_events, T_UINT, "attribute wakeup_events."),
-	{ .name = NULL, },
+static int pyrf_evsel__set_tracking(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->tracking = Py_IsTrue(val) ? true : false;
+	return 0;
+}
+
+static int pyrf_evsel__set_attr_config(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.config = PyLong_AsUnsignedLongLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_config(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLongLong(pevsel->evsel->core.attr.config);
+}
+
+static int pyrf_evsel__set_attr_read_format(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.read_format = PyLong_AsUnsignedLongLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_read_format(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLongLong(pevsel->evsel->core.attr.read_format);
+}
+
+static int pyrf_evsel__set_attr_sample_period(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.sample_period = PyLong_AsUnsignedLongLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_sample_period(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLongLong(pevsel->evsel->core.attr.sample_period);
+}
+
+static int pyrf_evsel__set_attr_sample_type(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.sample_type = PyLong_AsUnsignedLongLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_sample_type(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLongLong(pevsel->evsel->core.attr.sample_type);
+}
+
+static PyObject *pyrf_evsel__get_attr_size(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLong(pevsel->evsel->core.attr.size);
+}
+
+static int pyrf_evsel__set_attr_type(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.type = PyLong_AsUnsignedLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_type(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLong(pevsel->evsel->core.attr.type);
+}
+
+static int pyrf_evsel__set_attr_wakeup_events(PyObject *self, PyObject *val, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	pevsel->evsel->core.attr.wakeup_events = PyLong_AsUnsignedLong(val);
+	return PyErr_Occurred() ? -1 : 0;
+}
+
+static PyObject *pyrf_evsel__get_attr_wakeup_events(PyObject *self, void */*closure*/)
+{
+	struct pyrf_evsel *pevsel = (void *)self;
+
+	return PyLong_FromUnsignedLong(pevsel->evsel->core.attr.wakeup_events);
+}
+
+static PyGetSetDef pyrf_evsel__getset[] = {
+	{
+		.name = "tracking",
+		.get = pyrf_evsel__get_tracking,
+		.set = pyrf_evsel__set_tracking,
+		.doc = "tracking event.",
+	},
+	{
+		.name = "config",
+		.get = pyrf_evsel__get_attr_config,
+		.set = pyrf_evsel__set_attr_config,
+		.doc = "attribute config.",
+	},
+	{
+		.name = "read_format",
+		.get = pyrf_evsel__get_attr_read_format,
+		.set = pyrf_evsel__set_attr_read_format,
+		.doc = "attribute read_format.",
+	},
+	{
+		.name = "sample_period",
+		.get = pyrf_evsel__get_attr_sample_period,
+		.set = pyrf_evsel__set_attr_sample_period,
+		.doc = "attribute sample_period.",
+	},
+	{
+		.name = "sample_type",
+		.get = pyrf_evsel__get_attr_sample_type,
+		.set = pyrf_evsel__set_attr_sample_type,
+		.doc = "attribute sample_type.",
+	},
+	{
+		.name = "size",
+		.get = pyrf_evsel__get_attr_size,
+		.doc = "attribute size.",
+	},
+	{
+		.name = "type",
+		.get = pyrf_evsel__get_attr_type,
+		.set = pyrf_evsel__set_attr_type,
+		.doc = "attribute type.",
+	},
+	{
+		.name = "wakeup_events",
+		.get = pyrf_evsel__get_attr_wakeup_events,
+		.set = pyrf_evsel__set_attr_wakeup_events,
+		.doc = "attribute wakeup_events.",
+	},
+	{ .name = NULL},
 };
 
 static const char pyrf_evsel__doc[] = PyDoc_STR("perf event selector list object.");
 
+static PyObject *pyrf_evsel__getattro(struct pyrf_evsel *pevsel, PyObject *attr_name)
+{
+	if (!pevsel->evsel) {
+		PyErr_SetString(PyExc_ValueError, "evsel not initialized");
+		return NULL;
+	}
+	return PyObject_GenericGetAttr((PyObject *) pevsel, attr_name);
+}
+
 static PyTypeObject pyrf_evsel__type = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	.tp_name	= "perf.evsel",
@@ -1256,11 +1417,12 @@ static PyTypeObject pyrf_evsel__type = {
 	.tp_dealloc	= (destructor)pyrf_evsel__delete,
 	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
 	.tp_doc		= pyrf_evsel__doc,
-	.tp_members	= pyrf_evsel__members,
+	.tp_getset	= pyrf_evsel__getset,
 	.tp_methods	= pyrf_evsel__methods,
 	.tp_init	= (initproc)pyrf_evsel__init,
 	.tp_str         = pyrf_evsel__str,
 	.tp_repr        = pyrf_evsel__str,
+	.tp_getattro	= (getattrofunc) pyrf_evsel__getattro,
 };
 
 static int pyrf_evsel__setup_types(void)
@@ -1566,9 +1728,9 @@ static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist,
 		return NULL;
 
 	Py_INCREF(pevsel);
-	evsel = &((struct pyrf_evsel *)pevsel)->evsel;
+	evsel = ((struct pyrf_evsel *)pevsel)->evsel;
 	evsel->core.idx = evlist->core.nr_entries;
-	evlist__add(evlist, evsel);
+	evlist__add(evlist, evsel__get(evsel));
 
 	return Py_BuildValue("i", evlist->core.nr_entries);
 }
@@ -1626,7 +1788,7 @@ static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist,
 			return Py_None;
 		}
 
-		pevent->evsel = evsel;
+		pevent->evsel = evsel__get(evsel);
 
 		perf_mmap__consume(&md->core);
 
@@ -1803,12 +1965,7 @@ static PyObject *pyrf_evsel__from_evsel(struct evsel *evsel)
 	if (!pevsel)
 		return NULL;
 
-	memset(&pevsel->evsel, 0, sizeof(pevsel->evsel));
-	evsel__init(&pevsel->evsel, &evsel->core.attr, evsel->core.idx);
-
-	evsel__clone(&pevsel->evsel, evsel);
-	if (evsel__is_group_leader(evsel))
-		evsel__set_leader(&pevsel->evsel, &pevsel->evsel);
+	pevsel->evsel = evsel__get(evsel);
 	return (PyObject *)pevsel;
 }
 
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 1ac6cd43c38b..deb5b9dfe44c 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -1373,6 +1373,7 @@ static int evlist__deliver_deferred_callchain(struct evlist *evlist,
 		sample->evsel = evlist__id2evsel(evlist, sample->id);
 		ret = tool->callchain_deferred(tool, event, sample,
 					       sample->evsel, machine);
+		evsel__put(sample->evsel);
 		sample->evsel = saved_evsel;
 		return ret;
 	}
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 13/58] perf python: Use evsel in sample in pyrf_event
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Avoid a duplicated evsel by using the one in sample. Add
evsel__get/put to the evsel in perf_sample.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/evsel.c  |  2 +-
 tools/perf/util/python.c | 10 +++-------
 tools/perf/util/sample.c | 17 ++++++++++++-----
 3 files changed, 16 insertions(+), 13 deletions(-)

diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 3015b9b4b4da..9b16d832810f 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -3235,7 +3235,7 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	union u64_swap u;
 
 	perf_sample__init(data, /*all=*/true);
-	data->evsel = evsel;
+	data->evsel = evsel__get(evsel);
 	data->cpu = data->pid = data->tid = -1;
 	data->stream_id = data->id = data->time = -1ULL;
 	data->period = evsel->core.attr.sample_period;
diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 0162d8a625de..0424290f8b77 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -43,7 +43,6 @@ PyMODINIT_FUNC PyInit_perf(void);
 
 struct pyrf_event {
 	PyObject_HEAD
-	struct evsel *evsel;
 	struct perf_sample sample;
 	union perf_event   event;
 };
@@ -274,7 +273,6 @@ static PyMemberDef pyrf_sample_event__members[] = {
 
 static void pyrf_sample_event__delete(struct pyrf_event *pevent)
 {
-	evsel__put(pevent->evsel);
 	perf_sample__exit(&pevent->sample);
 	Py_TYPE(pevent)->tp_free((PyObject*)pevent);
 }
@@ -296,7 +294,7 @@ static PyObject *pyrf_sample_event__repr(const struct pyrf_event *pevent)
 #ifdef HAVE_LIBTRACEEVENT
 static bool is_tracepoint(const struct pyrf_event *pevent)
 {
-	return pevent->evsel->core.attr.type == PERF_TYPE_TRACEPOINT;
+	return pevent->sample.evsel->core.attr.type == PERF_TYPE_TRACEPOINT;
 }
 
 static PyObject*
@@ -343,7 +341,7 @@ tracepoint_field(const struct pyrf_event *pe, struct tep_format_field *field)
 static PyObject*
 get_tracepoint_field(struct pyrf_event *pevent, PyObject *attr_name)
 {
-	struct evsel *evsel = pevent->evsel;
+	struct evsel *evsel = pevent->sample.evsel;
 	struct tep_event *tp_format = evsel__tp_format(evsel);
 	struct tep_format_field *field;
 
@@ -509,7 +507,7 @@ static PyObject *pyrf_event__new(const union perf_event *event)
 	pevent = PyObject_New(struct pyrf_event, ptype);
 	if (pevent != NULL) {
 		memcpy(&pevent->event, event, event->header.size);
-		pevent->evsel = NULL;
+		perf_sample__init(&pevent->sample, /*all=*/false);
 	}
 	return (PyObject *)pevent;
 }
@@ -1788,8 +1786,6 @@ static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist,
 			return Py_None;
 		}
 
-		pevent->evsel = evsel__get(evsel);
-
 		perf_mmap__consume(&md->core);
 
 		err = evsel__parse_sample(evsel, &pevent->event, &pevent->sample);
diff --git a/tools/perf/util/sample.c b/tools/perf/util/sample.c
index cf73329326d7..7ec534dedc5c 100644
--- a/tools/perf/util/sample.c
+++ b/tools/perf/util/sample.c
@@ -1,18 +1,23 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 #include "sample.h"
+
+#include <stdlib.h>
+#include <string.h>
+
+#include <elf.h>
+#include <linux/zalloc.h>
+
 #include "debug.h"
+#include "evsel.h"
 #include "thread.h"
-#include <elf.h>
+#include "../../arch/x86/include/asm/insn.h"
+
 #ifndef EM_CSKY
 #define EM_CSKY		252
 #endif
 #ifndef EM_LOONGARCH
 #define EM_LOONGARCH	258
 #endif
-#include <linux/zalloc.h>
-#include <stdlib.h>
-#include <string.h>
-#include "../../arch/x86/include/asm/insn.h"
 
 void perf_sample__init(struct perf_sample *sample, bool all)
 {
@@ -29,6 +34,8 @@ void perf_sample__init(struct perf_sample *sample, bool all)
 
 void perf_sample__exit(struct perf_sample *sample)
 {
+	evsel__put(sample->evsel);
+	sample->evsel = NULL;
 	zfree(&sample->user_regs);
 	zfree(&sample->intr_regs);
 	if (sample->merged_callchain) {
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 09/58] perf data: Add open flag
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Avoid double opens and ensure only open files are closed. This
addresses some issues with python integration where the data file
wants to be opened before being given to a session.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
Changes in v2:

1. Fixed File Rotation: In perf_data__switch() , I added data->open =
   false; after the file is closed. This ensures that the subsequent
   perf_data__open() call will not exit early and will successfully
   open the new file.

2. Fixed Memory Leak: In open_dir() , I added a call to
   zfree(&data->file.path) if mkdir() fails, preventing the leak of
   the path string.
---
 tools/perf/util/data.c | 26 ++++++++++++++++++++++----
 tools/perf/util/data.h |  4 +++-
 2 files changed, 25 insertions(+), 5 deletions(-)

diff --git a/tools/perf/util/data.c b/tools/perf/util/data.c
index 94dc534a7386..17baf71897d1 100644
--- a/tools/perf/util/data.c
+++ b/tools/perf/util/data.c
@@ -346,8 +346,10 @@ static int open_dir(struct perf_data *data)
 		return -1;
 
 	if (perf_data__is_write(data) &&
-	    mkdir(data->path, S_IRWXU) < 0)
+	    mkdir(data->path, S_IRWXU) < 0) {
+		zfree(&data->file.path);
 		return -1;
+	}
 
 	ret = open_file(data);
 
@@ -360,9 +362,16 @@ static int open_dir(struct perf_data *data)
 
 int perf_data__open(struct perf_data *data)
 {
-	if (check_pipe(data))
+	int ret;
+
+	if (data->open)
 		return 0;
 
+	if (check_pipe(data)) {
+		data->open = true;
+		return 0;
+	}
+
 	/* currently it allows stdio for pipe only */
 	data->file.use_stdio = false;
 
@@ -375,16 +384,24 @@ int perf_data__open(struct perf_data *data)
 	if (perf_data__is_read(data))
 		data->is_dir = is_dir(data);
 
-	return perf_data__is_dir(data) ?
-	       open_dir(data) : open_file_dup(data);
+	ret = perf_data__is_dir(data) ? open_dir(data) : open_file_dup(data);
+
+	if (!ret)
+		data->open = true;
+
+	return ret;
 }
 
 void perf_data__close(struct perf_data *data)
 {
+	if (!data->open)
+		return;
+
 	if (perf_data__is_dir(data))
 		perf_data__close_dir(data);
 
 	perf_data_file__close(&data->file);
+	data->open = false;
 }
 
 static ssize_t perf_data_file__read(struct perf_data_file *file, void *buf, size_t size)
@@ -457,6 +474,7 @@ int perf_data__switch(struct perf_data *data,
 
 	if (!at_exit) {
 		perf_data_file__close(&data->file);
+		data->open = false;
 		ret = perf_data__open(data);
 		if (ret < 0)
 			goto out;
diff --git a/tools/perf/util/data.h b/tools/perf/util/data.h
index 8299fb5fa7da..76f57f60361f 100644
--- a/tools/perf/util/data.h
+++ b/tools/perf/util/data.h
@@ -50,6 +50,8 @@ struct perf_data {
 	const char		*path;
 	/** @file: Underlying file to be used. */
 	struct perf_data_file	 file;
+	/** @open: Has the file or directory been opened. */
+	bool			 open;
 	/** @is_pipe: Underlying file is a pipe. */
 	bool			 is_pipe;
 	/** @is_dir: Underlying file is a directory. */
@@ -59,7 +61,7 @@ struct perf_data {
 	/** @in_place_update: A file opened for reading but will be written to. */
 	bool			 in_place_update;
 	/** @mode: Read or write mode. */
-	enum perf_data_mode	 mode;
+	enum perf_data_mode	 mode:8;
 
 	struct {
 		/** @version: perf_dir_version. */
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 06/58] perf util: Sort includes and add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing includes found while cleaning the evsel/evlist header
files. Sort the remaining header files for consistency with the rest
of the code.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/bpf_off_cpu.c       | 30 +++++-----
 tools/perf/util/bpf_trace_augment.c |  8 +--
 tools/perf/util/evlist.c            | 91 +++++++++++++++--------------
 tools/perf/util/evsel.c             | 75 ++++++++++++------------
 tools/perf/util/map.h               |  9 ++-
 tools/perf/util/perf_api_probe.c    | 18 +++---
 tools/perf/util/s390-sample-raw.c   | 19 +++---
 tools/perf/util/stat-shadow.c       | 20 ++++---
 tools/perf/util/stat.c              | 16 +++--
 9 files changed, 152 insertions(+), 134 deletions(-)

diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
index a3b699a5322f..48cb930cdd2e 100644
--- a/tools/perf/util/bpf_off_cpu.c
+++ b/tools/perf/util/bpf_off_cpu.c
@@ -1,23 +1,25 @@
 // SPDX-License-Identifier: GPL-2.0
-#include "util/bpf_counter.h"
-#include "util/debug.h"
-#include "util/evsel.h"
-#include "util/evlist.h"
-#include "util/off_cpu.h"
-#include "util/perf-hooks.h"
-#include "util/record.h"
-#include "util/session.h"
-#include "util/target.h"
-#include "util/cpumap.h"
-#include "util/thread_map.h"
-#include "util/cgroup.h"
-#include "util/strlist.h"
+#include <linux/time64.h>
+
 #include <bpf/bpf.h>
 #include <bpf/btf.h>
 #include <internal/xyarray.h>
-#include <linux/time64.h>
 
+#include "bpf_counter.h"
 #include "bpf_skel/off_cpu.skel.h"
+#include "cgroup.h"
+#include "cpumap.h"
+#include "debug.h"
+#include "evlist.h"
+#include "evsel.h"
+#include "off_cpu.h"
+#include "parse-events.h"
+#include "perf-hooks.h"
+#include "record.h"
+#include "session.h"
+#include "strlist.h"
+#include "target.h"
+#include "thread_map.h"
 
 #define MAX_STACKS  32
 #define MAX_PROC  4096
diff --git a/tools/perf/util/bpf_trace_augment.c b/tools/perf/util/bpf_trace_augment.c
index 9e706f0fa53d..a9cf2a77ded1 100644
--- a/tools/perf/util/bpf_trace_augment.c
+++ b/tools/perf/util/bpf_trace_augment.c
@@ -1,11 +1,11 @@
 #include <bpf/libbpf.h>
 #include <internal/xyarray.h>
 
-#include "util/debug.h"
-#include "util/evlist.h"
-#include "util/trace_augment.h"
-
 #include "bpf_skel/augmented_raw_syscalls.skel.h"
+#include "debug.h"
+#include "evlist.h"
+#include "parse-events.h"
+#include "trace_augment.h"
 
 static struct augmented_raw_syscalls_bpf *skel;
 static struct evsel *bpf_output;
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index ee971d15b3c6..35d65fe50e06 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -5,67 +5,68 @@
  * Parts came from builtin-{top,stat,record}.c, see those files for further
  * copyright notes.
  */
-#include <api/fs/fs.h>
+#include "evlist.h"
+
 #include <errno.h>
 #include <inttypes.h>
-#include <poll.h>
-#include "cpumap.h"
-#include "util/mmap.h"
-#include "thread_map.h"
-#include "target.h"
-#include "dwarf-regs.h"
-#include "evlist.h"
-#include "evsel.h"
-#include "record.h"
-#include "debug.h"
-#include "units.h"
-#include "bpf_counter.h"
-#include <internal/lib.h> // page_size
-#include "affinity.h"
-#include "../perf.h"
-#include "asm/bug.h"
-#include "bpf-event.h"
-#include "util/event.h"
-#include "util/string2.h"
-#include "util/perf_api_probe.h"
-#include "util/evsel_fprintf.h"
-#include "util/pmu.h"
-#include "util/sample.h"
-#include "util/bpf-filter.h"
-#include "util/stat.h"
-#include "util/util.h"
-#include "util/env.h"
-#include "util/intel-tpebs.h"
-#include "util/metricgroup.h"
-#include "util/strbuf.h"
 #include <signal.h>
-#include <unistd.h>
-#include <sched.h>
 #include <stdlib.h>
 
-#include "parse-events.h"
-#include <subcmd/parse-options.h>
-
 #include <fcntl.h>
-#include <sys/ioctl.h>
-#include <sys/mman.h>
-#include <sys/prctl.h>
-#include <sys/timerfd.h>
-#include <sys/wait.h>
-
 #include <linux/bitops.h>
+#include <linux/err.h>
 #include <linux/hash.h>
 #include <linux/log2.h>
-#include <linux/err.h>
 #include <linux/string.h>
 #include <linux/time64.h>
 #include <linux/zalloc.h>
+#include <poll.h>
+#include <sched.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <sys/prctl.h>
+#include <sys/timerfd.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include <api/fs/fs.h>
+#include <internal/lib.h> // page_size
+#include <internal/xyarray.h>
+#include <perf/cpumap.h>
 #include <perf/evlist.h>
 #include <perf/evsel.h>
-#include <perf/cpumap.h>
 #include <perf/mmap.h>
+#include <subcmd/parse-options.h>
 
-#include <internal/xyarray.h>
+#include "../perf.h"
+#include "affinity.h"
+#include "asm/bug.h"
+#include "bpf-event.h"
+#include "bpf-filter.h"
+#include "bpf_counter.h"
+#include "cpumap.h"
+#include "debug.h"
+#include "dwarf-regs.h"
+#include "env.h"
+#include "event.h"
+#include "evsel.h"
+#include "evsel_fprintf.h"
+#include "intel-tpebs.h"
+#include "metricgroup.h"
+#include "mmap.h"
+#include "parse-events.h"
+#include "perf_api_probe.h"
+#include "pmu.h"
+#include "pmus.h"
+#include "record.h"
+#include "sample.h"
+#include "stat.h"
+#include "strbuf.h"
+#include "string2.h"
+#include "target.h"
+#include "thread_map.h"
+#include "units.h"
+#include "util.h"
 
 #ifdef LACKS_SIGQUEUE_PROTOTYPE
 int sigqueue(pid_t pid, int sig, const union sigval value);
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 2ee87fd84d3e..e03727d395e9 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -11,68 +11,71 @@
  */
 #define __SANE_USERSPACE_TYPES__
 
-#include <byteswap.h>
+#include "evsel.h"
+
 #include <errno.h>
 #include <inttypes.h>
+#include <stdlib.h>
+
+#include <dirent.h>
 #include <linux/bitops.h>
-#include <api/fs/fs.h>
-#include <api/fs/tracing_path.h>
-#include <linux/hw_breakpoint.h>
-#include <linux/perf_event.h>
 #include <linux/compiler.h>
+#include <linux/ctype.h>
 #include <linux/err.h>
+#include <linux/hw_breakpoint.h>
+#include <linux/perf_event.h>
 #include <linux/zalloc.h>
 #include <sys/ioctl.h>
 #include <sys/resource.h>
 #include <sys/syscall.h>
 #include <sys/types.h>
-#include <dirent.h>
-#include <stdlib.h>
+
+#include <api/fs/fs.h>
+#include <api/fs/tracing_path.h>
+#include <byteswap.h>
+#include <internal/lib.h>
+#include <internal/threadmap.h>
+#include <internal/xyarray.h>
+#include <perf/cpumap.h>
 #include <perf/evsel.h>
+
+#include "../perf-sys.h"
 #include "asm/bug.h"
+#include "bpf-filter.h"
 #include "bpf_counter.h"
 #include "callchain.h"
 #include "cgroup.h"
 #include "counts.h"
+#include "debug.h"
+#include "drm_pmu.h"
 #include "dwarf-regs.h"
+#include "env.h"
 #include "event.h"
-#include "evsel.h"
-#include "time-utils.h"
-#include "util/env.h"
-#include "util/evsel_config.h"
-#include "util/evsel_fprintf.h"
 #include "evlist.h"
-#include <perf/cpumap.h>
-#include "thread_map.h"
-#include "target.h"
+#include "evsel_config.h"
+#include "evsel_fprintf.h"
+#include "hashmap.h"
+#include "hist.h"
+#include "hwmon_pmu.h"
+#include "intel-tpebs.h"
+#include "memswap.h"
+#include "off_cpu.h"
+#include "parse-branch-options.h"
 #include "perf_regs.h"
+#include "pmu.h"
+#include "pmus.h"
 #include "record.h"
-#include "debug.h"
-#include "trace-event.h"
+#include "rlimit.h"
 #include "session.h"
 #include "stat.h"
 #include "string2.h"
-#include "memswap.h"
-#include "util.h"
-#include "util/hashmap.h"
-#include "off_cpu.h"
-#include "pmu.h"
-#include "pmus.h"
-#include "drm_pmu.h"
-#include "hwmon_pmu.h"
+#include "target.h"
+#include "thread_map.h"
+#include "time-utils.h"
 #include "tool_pmu.h"
 #include "tp_pmu.h"
-#include "rlimit.h"
-#include "../perf-sys.h"
-#include "util/parse-branch-options.h"
-#include "util/bpf-filter.h"
-#include "util/hist.h"
-#include <internal/xyarray.h>
-#include <internal/lib.h>
-#include <internal/threadmap.h>
-#include "util/intel-tpebs.h"
-
-#include <linux/ctype.h>
+#include "trace-event.h"
+#include "util.h"
 
 #ifdef HAVE_LIBTRACEEVENT
 #include <event-parse.h>
diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
index 979b3e11b9bc..fb0279810ae9 100644
--- a/tools/perf/util/map.h
+++ b/tools/perf/util/map.h
@@ -2,14 +2,13 @@
 #ifndef __PERF_MAP_H
 #define __PERF_MAP_H
 
-#include <linux/refcount.h>
-#include <linux/compiler.h>
-#include <linux/list.h>
-#include <linux/rbtree.h>
+#include <stdbool.h>
 #include <stdio.h>
 #include <string.h>
-#include <stdbool.h>
+
+#include <linux/refcount.h>
 #include <linux/types.h>
+
 #include <internal/rc_check.h>
 
 struct dso;
diff --git a/tools/perf/util/perf_api_probe.c b/tools/perf/util/perf_api_probe.c
index 6ecf38314f01..e1904a330b28 100644
--- a/tools/perf/util/perf_api_probe.c
+++ b/tools/perf/util/perf_api_probe.c
@@ -1,14 +1,18 @@
 /* SPDX-License-Identifier: GPL-2.0 */
+#include "perf_api_probe.h"
 
-#include "perf-sys.h"
-#include "util/cloexec.h"
-#include "util/evlist.h"
-#include "util/evsel.h"
-#include "util/parse-events.h"
-#include "util/perf_api_probe.h"
-#include <perf/cpumap.h>
 #include <errno.h>
 
+#include <perf/cpumap.h>
+
+#include "cloexec.h"
+#include "evlist.h"
+#include "evsel.h"
+#include "parse-events.h"
+#include "perf-sys.h"
+#include "pmu.h"
+#include "pmus.h"
+
 typedef void (*setup_probe_fn_t)(struct evsel *evsel);
 
 static int perf_do_probe_api(setup_probe_fn_t fn, struct perf_cpu cpu, const char *str)
diff --git a/tools/perf/util/s390-sample-raw.c b/tools/perf/util/s390-sample-raw.c
index c6ae0ae8d86a..6bf0edf80d4e 100644
--- a/tools/perf/util/s390-sample-raw.c
+++ b/tools/perf/util/s390-sample-raw.c
@@ -12,25 +12,26 @@
  * sample was taken from.
  */
 
-#include <unistd.h>
+#include <inttypes.h>
 #include <stdio.h>
 #include <string.h>
-#include <inttypes.h>
 
-#include <sys/stat.h>
+#include <asm/byteorder.h>
 #include <linux/compiler.h>
 #include <linux/err.h>
-#include <asm/byteorder.h>
+#include <sys/stat.h>
+#include <unistd.h>
 
+#include "color.h"
 #include "debug.h"
-#include "session.h"
 #include "evlist.h"
-#include "color.h"
 #include "hashmap.h"
-#include "sample-raw.h"
+#include "pmu.h"
+#include "pmus.h"
 #include "s390-cpumcf-kernel.h"
-#include "util/pmu.h"
-#include "util/sample.h"
+#include "sample-raw.h"
+#include "sample.h"
+#include "session.h"
 
 static size_t ctrset_size(struct cf_ctrset_entry *set)
 {
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index bc2d44df7baf..48524450326d 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -2,20 +2,24 @@
 #include <errno.h>
 #include <math.h>
 #include <stdio.h>
-#include "evsel.h"
-#include "stat.h"
+
+#include <linux/zalloc.h>
+
+#include "cgroup.h"
 #include "color.h"
 #include "debug.h"
-#include "pmu.h"
-#include "rblist.h"
 #include "evlist.h"
+#include "evsel.h"
 #include "expr.h"
-#include "metricgroup.h"
-#include "cgroup.h"
-#include "units.h"
+#include "hashmap.h"
 #include "iostat.h"
-#include "util/hashmap.h"
+#include "metricgroup.h"
+#include "pmu.h"
+#include "pmus.h"
+#include "rblist.h"
+#include "stat.h"
 #include "tool_pmu.h"
+#include "units.h"
 
 static bool tool_pmu__is_time_event(const struct perf_stat_config *config,
 				   const struct evsel *evsel, int *tool_aggr_idx)
diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
index 14d169e22e8f..66eb9a66a4f7 100644
--- a/tools/perf/util/stat.c
+++ b/tools/perf/util/stat.c
@@ -1,21 +1,25 @@
 // SPDX-License-Identifier: GPL-2.0
+#include "stat.h"
+
 #include <errno.h>
-#include <linux/err.h>
 #include <inttypes.h>
 #include <math.h>
 #include <string.h>
+
+#include <linux/err.h>
+#include <linux/zalloc.h>
+
 #include "counts.h"
 #include "cpumap.h"
 #include "debug.h"
+#include "evlist.h"
+#include "evsel.h"
+#include "hashmap.h"
 #include "header.h"
-#include "stat.h"
+#include "pmu.h"
 #include "session.h"
 #include "target.h"
-#include "evlist.h"
-#include "evsel.h"
 #include "thread_map.h"
-#include "util/hashmap.h"
-#include <linux/zalloc.h>
 
 void update_stats(struct stats *stats, u64 val)
 {
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 08/58] perf evsel/evlist: Avoid unnecessary #includes
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Use forward declarations and remove unnecessary #includes in
evsel.h. Sort the forward declarations in evsel.h and evlist.h.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/evlist.h | 15 +++++++++------
 tools/perf/util/evsel.h  | 20 +++++++++++---------
 2 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index e507f5f20ef6..e54761c670b6 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -2,29 +2,32 @@
 #ifndef __PERF_EVLIST_H
 #define __PERF_EVLIST_H 1
 
+#include <signal.h>
+
 #include <linux/compiler.h>
 #include <linux/kernel.h>
-#include <linux/refcount.h>
 #include <linux/list.h>
+#include <linux/refcount.h>
+#include <pthread.h>
+#include <unistd.h>
+
 #include <api/fd/array.h>
 #include <internal/evlist.h>
 #include <internal/evsel.h>
 #include <perf/evlist.h>
+
 #include "affinity.h"
 #include "events_stats.h"
 #include "evsel.h"
 #include "rblist.h"
-#include <pthread.h>
-#include <signal.h>
-#include <unistd.h>
 
-struct pollfd;
-struct thread_map;
 struct perf_cpu_map;
 struct perf_stat_config;
+struct pollfd;
 struct record_opts;
 struct strbuf;
 struct target;
+struct thread_map;
 
 /*
  * State machine of bkw_mmap_state:
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 339b5c08a33d..b099c8e5dd86 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -2,28 +2,30 @@
 #ifndef __PERF_EVSEL_H
 #define __PERF_EVSEL_H 1
 
-#include <linux/list.h>
 #include <stdbool.h>
-#include <sys/types.h>
+
+#include <linux/list.h>
 #include <linux/perf_event.h>
 #include <linux/types.h>
+#include <sys/types.h>
+
 #include <internal/evsel.h>
 #include <perf/evsel.h>
+
 #include "symbol_conf.h"
-#include "pmus.h"
-#include "pmu.h"
 
+struct bperf_follower_bpf;
+struct bperf_leader_bpf;
+struct bpf_counter_ops;
 struct bpf_object;
 struct cgroup;
+struct hashmap;
 struct perf_counts;
+struct perf_pmu;
 struct perf_stat_config;
 struct perf_stat_evsel;
-union perf_event;
-struct bpf_counter_ops;
 struct target;
-struct hashmap;
-struct bperf_leader_bpf;
-struct bperf_follower_bpf;
+union perf_event;
 
 typedef int (evsel__sb_cb_t)(union perf_event *event, void *data);
 
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 07/58] perf python: Add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing #include of pmus.h found while cleaning the evsel/evlist
header files.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/python.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index cc1019d29a5d..1e6c99efff90 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -5,26 +5,29 @@
 #include <poll.h>
 #include <linux/err.h>
 #include <perf/cpumap.h>
-#ifdef HAVE_LIBTRACEEVENT
-#include <event-parse.h>
-#endif
+#include <internal/lib.h>
 #include <perf/mmap.h>
+
 #include "callchain.h"
 #include "counts.h"
+#include "event.h"
 #include "evlist.h"
 #include "evsel.h"
-#include "event.h"
 #include "expr.h"
+#include "metricgroup.h"
+#include "mmap.h"
+#include "pmus.h"
 #include "print_binary.h"
 #include "record.h"
 #include "strbuf.h"
 #include "thread_map.h"
 #include "tp_pmu.h"
 #include "trace-event.h"
-#include "metricgroup.h"
-#include "mmap.h"
 #include "util/sample.h"
-#include <internal/lib.h>
+
+#ifdef HAVE_LIBTRACEEVENT
+#include <event-parse.h>
+#endif
 
 PyMODINIT_FUNC PyInit_perf(void);
 
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 05/58] perf script: Sort includes and add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing #include of pmu.h found while cleaning the evsel/evlist
header files. Sort the remaining header files for consistency with the
rest of the code. Doing this exposed a missing forward declaration of
addr_location in print_insn.h, add this and sort the forward
declarations.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/builtin-script.c  | 111 ++++++++++++++++++-----------------
 tools/perf/util/print_insn.h |   5 +-
 2 files changed, 60 insertions(+), 56 deletions(-)

diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index c8ac9f01a36b..853b141a0d50 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -1,74 +1,77 @@
 // SPDX-License-Identifier: GPL-2.0
-#include "builtin.h"
+#include <errno.h>
+#include <inttypes.h>
+#include <signal.h>
+#include <stdio.h>
+
+#include <dirent.h>
+#include <fcntl.h>
+#include <linux/bitmap.h>
+#include <linux/compiler.h>
+#include <linux/ctype.h>
+#include <linux/err.h>
+#include <linux/kernel.h>
+#include <linux/stringify.h>
+#include <linux/time64.h>
+#include <linux/unaligned.h>
+#include <linux/zalloc.h>
+#include <sys/param.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <unistd.h>
 
+#include <perf/evlist.h>
+#include <subcmd/exec-cmd.h>
+#include <subcmd/pager.h>
+#include <subcmd/parse-options.h>
+
+#include "asm/bug.h"
+#include "builtin.h"
+#include "perf.h"
+#include "print_binary.h"
+#include "print_insn.h"
+#include "ui/ui.h"
+#include "util/annotate.h"
+#include "util/auxtrace.h"
+#include "util/cgroup.h"
+#include "util/color.h"
 #include "util/counts.h"
+#include "util/cpumap.h"
+#include "util/data.h"
 #include "util/debug.h"
+#include "util/dlfilter.h"
 #include "util/dso.h"
-#include <subcmd/exec-cmd.h>
-#include "util/header.h"
-#include <subcmd/parse-options.h>
-#include "util/perf_regs.h"
-#include "util/session.h"
-#include "util/tool.h"
-#include "util/map.h"
-#include "util/srcline.h"
-#include "util/symbol.h"
-#include "util/thread.h"
-#include "util/trace-event.h"
+#include "util/dump-insn.h"
 #include "util/env.h"
+#include "util/event.h"
 #include "util/evlist.h"
 #include "util/evsel.h"
 #include "util/evsel_fprintf.h"
 #include "util/evswitch.h"
+#include "util/header.h"
+#include "util/map.h"
+#include "util/mem-events.h"
+#include "util/mem-info.h"
+#include "util/metricgroup.h"
+#include "util/path.h"
+#include "util/perf_regs.h"
+#include "util/pmu.h"
+#include "util/record.h"
+#include "util/session.h"
 #include "util/sort.h"
-#include "util/data.h"
-#include "util/auxtrace.h"
-#include "util/cpumap.h"
-#include "util/thread_map.h"
+#include "util/srcline.h"
 #include "util/stat.h"
-#include "util/color.h"
 #include "util/string2.h"
+#include "util/symbol.h"
 #include "util/thread-stack.h"
+#include "util/thread.h"
+#include "util/thread_map.h"
 #include "util/time-utils.h"
-#include "util/path.h"
-#include "util/event.h"
-#include "util/mem-info.h"
-#include "util/metricgroup.h"
-#include "ui/ui.h"
-#include "print_binary.h"
-#include "print_insn.h"
-#include <linux/bitmap.h>
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/stringify.h>
-#include <linux/time64.h>
-#include <linux/zalloc.h>
-#include <linux/unaligned.h>
-#include <sys/utsname.h>
-#include "asm/bug.h"
-#include "util/mem-events.h"
-#include "util/dump-insn.h"
-#include <dirent.h>
-#include <errno.h>
-#include <inttypes.h>
-#include <signal.h>
-#include <stdio.h>
-#include <sys/param.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <fcntl.h>
-#include <unistd.h>
-#include <subcmd/pager.h>
-#include <perf/evlist.h>
-#include <linux/err.h>
-#include "util/dlfilter.h"
-#include "util/record.h"
+#include "util/tool.h"
+#include "util/trace-event.h"
 #include "util/util.h"
-#include "util/cgroup.h"
-#include "util/annotate.h"
-#include "perf.h"
 
-#include <linux/ctype.h>
 #ifdef HAVE_LIBTRACEEVENT
 #include <event-parse.h>
 #endif
diff --git a/tools/perf/util/print_insn.h b/tools/perf/util/print_insn.h
index 07d11af3fc1c..a54f7e858e49 100644
--- a/tools/perf/util/print_insn.h
+++ b/tools/perf/util/print_insn.h
@@ -5,10 +5,11 @@
 #include <stddef.h>
 #include <stdio.h>
 
-struct perf_sample;
-struct thread;
+struct addr_location;
 struct machine;
 struct perf_insn;
+struct perf_sample;
+struct thread;
 
 #define PRINT_INSN_IMM_HEX		(1<<0)
 
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 04/58] perf tests: Sort includes and add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing #includes found while cleaning the evsel/evlist header
files. Sort the remaining header files for consistency with the rest
of the code.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/tests/hwmon_pmu.c  | 14 +++++++++-----
 tools/perf/tests/mmap-basic.c | 18 +++++++++++-------
 2 files changed, 20 insertions(+), 12 deletions(-)

diff --git a/tools/perf/tests/hwmon_pmu.c b/tools/perf/tests/hwmon_pmu.c
index 4aa4aac94f09..ada6e445c4c4 100644
--- a/tools/perf/tests/hwmon_pmu.c
+++ b/tools/perf/tests/hwmon_pmu.c
@@ -1,15 +1,19 @@
 // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
-#include "debug.h"
-#include "evlist.h"
 #include "hwmon_pmu.h"
-#include "parse-events.h"
-#include "tests.h"
+
 #include <errno.h>
+
 #include <fcntl.h>
-#include <sys/stat.h>
 #include <linux/compiler.h>
 #include <linux/kernel.h>
 #include <linux/string.h>
+#include <sys/stat.h>
+
+#include "debug.h"
+#include "evlist.h"
+#include "parse-events.h"
+#include "pmus.h"
+#include "tests.h"
 
 static const struct test_event {
 	const char *name;
diff --git a/tools/perf/tests/mmap-basic.c b/tools/perf/tests/mmap-basic.c
index 3313c236104e..8d04f6edb004 100644
--- a/tools/perf/tests/mmap-basic.c
+++ b/tools/perf/tests/mmap-basic.c
@@ -1,25 +1,29 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <errno.h>
-#include <fcntl.h>
 #include <inttypes.h>
 #include <stdlib.h>
+
+#include <fcntl.h>
+#include <linux/err.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+
 #include <perf/cpumap.h>
+#include <perf/evlist.h>
+#include <perf/mmap.h>
 
 #include "cpumap.h"
 #include "debug.h"
 #include "event.h"
 #include "evlist.h"
 #include "evsel.h"
-#include "thread_map.h"
+#include "pmu.h"
+#include "pmus.h"
 #include "tests.h"
+#include "thread_map.h"
 #include "util/affinity.h"
 #include "util/mmap.h"
 #include "util/sample.h"
-#include <linux/err.h>
-#include <linux/kernel.h>
-#include <linux/string.h>
-#include <perf/evlist.h>
-#include <perf/mmap.h>
 
 /*
  * This test will generate random numbers of calls to some getpid syscalls,
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 03/58] perf arch x86: Sort includes and add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing #includes found while cleaning the evsel/evlist header
files. Sort the remaining header files for consistency with the rest
of the code.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/arch/x86/util/intel-bts.c | 20 +++++++++++--------
 tools/perf/arch/x86/util/intel-pt.c  | 29 +++++++++++++++-------------
 2 files changed, 28 insertions(+), 21 deletions(-)

diff --git a/tools/perf/arch/x86/util/intel-bts.c b/tools/perf/arch/x86/util/intel-bts.c
index 85c8186300c8..100a23d27998 100644
--- a/tools/perf/arch/x86/util/intel-bts.c
+++ b/tools/perf/arch/x86/util/intel-bts.c
@@ -4,26 +4,30 @@
  * Copyright (c) 2013-2015, Intel Corporation.
  */
 
+#include "../../../util/intel-bts.h"
+
 #include <errno.h>
-#include <linux/kernel.h>
-#include <linux/types.h>
+
 #include <linux/bitops.h>
+#include <linux/kernel.h>
 #include <linux/log2.h>
+#include <linux/types.h>
 #include <linux/zalloc.h>
 
+#include <internal/lib.h> // page_size
+
+#include "../../../util/auxtrace.h"
 #include "../../../util/cpumap.h"
+#include "../../../util/debug.h"
 #include "../../../util/event.h"
-#include "../../../util/evsel.h"
 #include "../../../util/evlist.h"
+#include "../../../util/evsel.h"
 #include "../../../util/mmap.h"
-#include "../../../util/session.h"
+#include "../../../util/pmu.h"
 #include "../../../util/pmus.h"
-#include "../../../util/debug.h"
 #include "../../../util/record.h"
+#include "../../../util/session.h"
 #include "../../../util/tsc.h"
-#include "../../../util/auxtrace.h"
-#include "../../../util/intel-bts.h"
-#include <internal/lib.h> // page_size
 
 #define KiB(x) ((x) * 1024)
 #define MiB(x) ((x) * 1024 * 1024)
diff --git a/tools/perf/arch/x86/util/intel-pt.c b/tools/perf/arch/x86/util/intel-pt.c
index c131a727774f..0307ff15d9fc 100644
--- a/tools/perf/arch/x86/util/intel-pt.c
+++ b/tools/perf/arch/x86/util/intel-pt.c
@@ -3,36 +3,39 @@
  * intel_pt.c: Intel Processor Trace support
  * Copyright (c) 2013-2015, Intel Corporation.
  */
+#include "../../../util/intel-pt.h"
 
 #include <errno.h>
 #include <stdbool.h>
-#include <linux/kernel.h>
-#include <linux/types.h>
+
 #include <linux/bitops.h>
+#include <linux/err.h>
+#include <linux/kernel.h>
 #include <linux/log2.h>
+#include <linux/types.h>
 #include <linux/zalloc.h>
-#include <linux/err.h>
 
-#include "../../../util/session.h"
+#include <api/fs/fs.h>
+#include <internal/lib.h> // page_size
+#include <subcmd/parse-options.h>
+
+#include "../../../util/auxtrace.h"
+#include "../../../util/config.h"
+#include "../../../util/cpumap.h"
+#include "../../../util/debug.h"
 #include "../../../util/event.h"
 #include "../../../util/evlist.h"
 #include "../../../util/evsel.h"
 #include "../../../util/evsel_config.h"
-#include "../../../util/config.h"
-#include "../../../util/cpumap.h"
 #include "../../../util/mmap.h"
-#include <subcmd/parse-options.h>
 #include "../../../util/parse-events.h"
-#include "../../../util/pmus.h"
-#include "../../../util/debug.h"
-#include "../../../util/auxtrace.h"
 #include "../../../util/perf_api_probe.h"
+#include "../../../util/pmu.h"
+#include "../../../util/pmus.h"
 #include "../../../util/record.h"
+#include "../../../util/session.h"
 #include "../../../util/target.h"
 #include "../../../util/tsc.h"
-#include <internal/lib.h> // page_size
-#include "../../../util/intel-pt.h"
-#include <api/fs/fs.h>
 #include "cpuid.h"
 
 #define KiB(x) ((x) * 1024)
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 02/58] perf arch arm: Sort includes and add missed explicit dependencies
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

Fix missing #includes found while cleaning the evsel/evlist header
files. Sort the remaining header files for consistency with the rest
of the code.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/arch/arm/util/cs-etm.c | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/tools/perf/arch/arm/util/cs-etm.c b/tools/perf/arch/arm/util/cs-etm.c
index b7a839de8707..cdf8e3e60606 100644
--- a/tools/perf/arch/arm/util/cs-etm.c
+++ b/tools/perf/arch/arm/util/cs-etm.c
@@ -3,10 +3,13 @@
  * Copyright(C) 2015 Linaro Limited. All rights reserved.
  * Author: Mathieu Poirier <mathieu.poirier@linaro.org>
  */
+#include "../../../util/cs-etm.h"
+
+#include <errno.h>
+#include <stdlib.h>
 
-#include <api/fs/fs.h>
-#include <linux/bits.h>
 #include <linux/bitops.h>
+#include <linux/bits.h>
 #include <linux/compiler.h>
 #include <linux/coresight-pmu.h>
 #include <linux/kernel.h>
@@ -14,25 +17,24 @@
 #include <linux/string.h>
 #include <linux/types.h>
 #include <linux/zalloc.h>
+#include <sys/stat.h>
+
+#include <api/fs/fs.h>
+#include <internal/lib.h> // page_size
 
-#include "cs-etm.h"
-#include "../../../util/debug.h"
-#include "../../../util/record.h"
 #include "../../../util/auxtrace.h"
 #include "../../../util/cpumap.h"
+#include "../../../util/debug.h"
 #include "../../../util/event.h"
 #include "../../../util/evlist.h"
 #include "../../../util/evsel.h"
-#include "../../../util/perf_api_probe.h"
 #include "../../../util/evsel_config.h"
+#include "../../../util/perf_api_probe.h"
+#include "../../../util/pmu.h"
 #include "../../../util/pmus.h"
-#include "../../../util/cs-etm.h"
-#include <internal/lib.h> // page_size
+#include "../../../util/record.h"
 #include "../../../util/session.h"
-
-#include <errno.h>
-#include <stdlib.h>
-#include <sys/stat.h>
+#include "cs-etm.h"
 
 struct cs_etm_recording {
 	struct auxtrace_record	itr;
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH v5 00/58] perf: Reorganize scripting support
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260423163406.1779809-1-irogers@google.com>

The perf script command has long supported running Python and Perl scripts by
embedding libpython and libperl. This approach has several drawbacks:
 - overhead by creating Python dictionaries for every event (whether used or
   not),
 - complex build dependencies on specific Python/Perl versions,
 - complications with threading due to perf being the interpreter,
 - no clear way to run standalone scripts like ilist.py.

This series takes a different approach with some initial implementation posted
as an RFC last October:
https://lore.kernel.org/linux-perf-users/20251029053413.355154-1-irogers@google.com/
with the motivation coming up on the mailing list earlier:
https://lore.kernel.org/lkml/CAP-5=fWDqE8SYfOLZkg_0=4Ayx6E7O+h7uUp4NDeCFkiN4b7-w@mail.gmail.com/

The changes remove the embedded libpython and libperl support from perf
entirely. Instead, they expand the existing perf Python module to provide full
access to perf data files and events, allowing scripts to be run as standalone
Python applications.

To demonstrate the benefits, we ported all existing Python and Perl scripts to
use the new Python session API. The performance improvement is dramatic. For
example, porting mem-phys-addr.py:

Before (using embedded libpython in perf script):
```
$ perf mem record a sleep 1
$ time perf script tools/perf/scripts/python/mem-phys-addr.py
Event: cpu_core/mem-loads-aux/
Memory type                                    count  percentage
 ---------------------------------------  ----------  ----------
0-fff : Reserved                                3217       100.0

real    0m3.754s
user    0m0.023s
sys     0m0.018s
```

After (using standalone Python script with perf module):
```
$ PYTHONPATH=/tmp/perf/python time python3 tools/perf/python/mem-phys-addr.py
Event: evsel(cpu_core/mem-loads-aux/)
Memory type                                    count  percentage
 ---------------------------------------  ----------  ----------
0-fff : Reserved                                3217       100.0

real    0m0.106s
user    0m0.021s
sys     0m0.020s
```

This is a roughly 35x speedup!

The change is large (11291 insertions, 15964 deletions, net 4673
deletions) due to porting all existing perl and python code to the new
API. Gemini was used to achieve this and to improve the code
quality. Removing support may be controversial, however, the first 52
patches are additive and merging those would allow us to focus on the
remaining 6 patches that finalize the new perf script behavior.

---
v5 Changes
----------

Resending due to partial send of v4 due to a quota limit.

---
v4 Changes
----------

1. Git Fixup Cleanups
- Squashed the lingering `fixup!` commit remaining from the previous session back
  into `perf check-perf-trace: Port check-perf-trace to use python module`.

---
v3 Changes
----------

1. Memory Safety & Reference Counting Fixes
- Stored transient mmap event data inside the Python object's permanent
  `pevent->event` and invoked `evsel__parse_sample()` to safely point
  attributes into it, resolving Use-After-Free vulnerabilities.
- Nullified `sample->evsel` after calling `evsel__put()` in
  `perf_sample__exit()` to protect against potential refcount double-put
  crashes in error paths.
- Reordered operations inside `evlist__remove()` to invoke
  `perf_evlist__remove()` before reference release.
- Patched an `evsel` reference leak inside `evlist__deliver_deferred_callchain()`.

2. Sashiko AI Review Cleanups
- Corrected the broken event name equality check in `gecko.py` to search
  for a substring match within the parsed event string.
- Fixed a latent `AttributeError` crash in `task-analyzer.py` by properly
  assigning the session instance.
- Safeguarded thread reporting in `check-perf-trace.py` by utilizing
  `sample_tid` instead of `sample_pid`, and wrapping the session thread
  resolution in a try-except block.

3. Omitted Minor Issues
- The minor review comments (such as permanent iterator exhaustion on
  `brstack`, or dead-code in `failed-syscalls-by-pid.py`) have been omitted
  because they do not affect correctness, lead to crashes, or require
  significant architectural rework.

---
v2 Changes
----------

1. String Match and Event Name Accuracy
- Replaced loose substring event matching across the script suite with exact
  matches or specific prefix constraints (syscalls:sys_exit_,
  evsel(skb:kfree_skb), etc.).
- Added getattr() safety checks to prevent script failures caused by
  unresolved attributes from older kernel traces.

2. OOM and Memory Protections
- Refactored netdev-times.py to compute and process network statistics
  chronologically on-the-fly, eliminating an unbounded in-memory list
  that caused Out-Of-Memory crashes on large files.
- Implemented threshold limits on intel-pt-events.py to cap memory allocation
  during event interleaving.
- Optimized export-to-sqlite.py to periodically commit database transactions
  (every 10,000 samples) to reduce temporary SQLite journal sizes.

3. Portability & Environment Independence
- Re-keyed internal tracking dictionaries in scripts like powerpc-hcalls.py to
  use thread PIDs instead of CPUs, ensuring correctness when threads migrate.
- Switched net_dropmonitor.py from host-specific /proc/kallsyms parsing to
  perf's built-in symbol resolution API. 
- Added the --iomem parameter to mem-phys-addr.py to support offline analysis
  of data collected on different architectures.

4. Standalone Scripting Improvements
- Patched builtin-script.c to ensure --input parameters are successfully passed
  down to standalone execution pipelines via execvp().
- Guarded against string buffer overflows during .py extension path resolving.

5. Code Cleanups
- Removed stale perl subdirectories from being detected by the TUI script
  browser.
- Ran the entire script suite through mypy and pylint to achieve strict static
  type checking and resolve unreferenced variables.

Ian Rogers (58):
  perf inject: Fix itrace branch stack synthesis
  perf arch arm: Sort includes and add missed explicit dependencies
  perf arch x86: Sort includes and add missed explicit dependencies
  perf tests: Sort includes and add missed explicit dependencies
  perf script: Sort includes and add missed explicit dependencies
  perf util: Sort includes and add missed explicit dependencies
  perf python: Add missed explicit dependencies
  perf evsel/evlist: Avoid unnecessary #includes
  perf data: Add open flag
  perf evlist: Add reference count
  perf evsel: Add reference count
  perf evlist: Add reference count checking
  perf python: Use evsel in sample in pyrf_event
  perf python: Add wrapper for perf_data file abstraction
  perf python: Add python session abstraction wrapping perf's session
  perf python: Add syscall name/id to convert syscall number and name
  perf python: Refactor and add accessors to sample event
  perf python: Add callchain support
  perf python: Add config file access
  perf python: Extend API for stat events in python.c
  perf python: Expose brstack in sample event
  perf python: Add perf.pyi stubs file
  perf python: Add LiveSession helper
  perf python: Move exported-sql-viewer.py and parallel-perf.py to
    tools/perf/python/
  perf stat-cpi: Port stat-cpi to use python module
  perf mem-phys-addr: Port mem-phys-addr to use python module
  perf syscall-counts: Port syscall-counts to use python module
  perf syscall-counts-by-pid: Port syscall-counts-by-pid to use python
    module
  perf futex-contention: Port futex-contention to use python module
  perf flamegraph: Port flamegraph to use python module
  perf gecko: Port gecko to use python module
  perf arm-cs-trace-disasm: Port arm-cs-trace-disasm to use python
    module
  perf check-perf-trace: Port check-perf-trace to use python module
  perf compaction-times: Port compaction-times to use python module
  perf event_analyzing_sample: Port event_analyzing_sample to use python
    module
  perf export-to-sqlite: Port export-to-sqlite to use python module
  perf export-to-postgresql: Port export-to-postgresql to use python
    module
  perf failed-syscalls-by-pid: Port failed-syscalls-by-pid to use python
    module
  perf intel-pt-events: Port intel-pt-events/libxed to use python module
  perf net_dropmonitor: Port net_dropmonitor to use python module
  perf netdev-times: Port netdev-times to use python module
  perf powerpc-hcalls: Port powerpc-hcalls to use python module
  perf sched-migration: Port sched-migration/SchedGui to use python
    module
  perf sctop: Port sctop to use python module
  perf stackcollapse: Port stackcollapse to use python module
  perf task-analyzer: Port task-analyzer to use python module
  perf failed-syscalls: Port failed-syscalls to use python module
  perf rw-by-file: Port rw-by-file to use python module
  perf rw-by-pid: Port rw-by-pid to use python module
  perf rwtop: Port rwtop to use python module
  perf wakeup-latency: Port wakeup-latency to use python module
  perf test: Migrate Intel PT virtual LBR test to use Python API
  perf: Remove libperl support, legacy Perl scripts and tests
  perf: Remove libpython support and legacy Python scripts
  perf Makefile: Update Python script installation path
  perf script: Refactor to support standalone scripts and remove legacy
    features
  perf Documentation: Update for standalone Python scripts and remove
    obsolete data
  perf python: Improve perf script -l descriptions

 tools/build/Makefile.feature                  |    5 +-
 tools/build/feature/Makefile                  |   23 +-
 tools/build/feature/test-all.c                |    6 +-
 tools/build/feature/test-libperl.c            |   10 -
 tools/build/feature/test-libpython.c          |   10 -
 tools/build/feature/test-python-module.c      |   12 +
 tools/perf/Documentation/perf-check.txt       |    2 -
 tools/perf/Documentation/perf-script-perl.txt |  216 --
 .../perf/Documentation/perf-script-python.txt |  702 +-----
 tools/perf/Documentation/perf-script.txt      |   70 +-
 tools/perf/Makefile.config                    |   37 +-
 tools/perf/Makefile.perf                      |   22 +-
 tools/perf/arch/arm/util/cs-etm.c             |   36 +-
 tools/perf/arch/arm64/util/arm-spe.c          |    8 +-
 tools/perf/arch/arm64/util/hisi-ptt.c         |    2 +-
 tools/perf/arch/x86/tests/hybrid.c            |   22 +-
 tools/perf/arch/x86/tests/topdown.c           |    2 +-
 tools/perf/arch/x86/util/auxtrace.c           |    2 +-
 tools/perf/arch/x86/util/intel-bts.c          |   26 +-
 tools/perf/arch/x86/util/intel-pt.c           |   38 +-
 tools/perf/arch/x86/util/iostat.c             |    8 +-
 tools/perf/bench/evlist-open-close.c          |   29 +-
 tools/perf/bench/inject-buildid.c             |    9 +-
 tools/perf/builtin-annotate.c                 |    2 +-
 tools/perf/builtin-check.c                    |    3 +-
 tools/perf/builtin-ftrace.c                   |   14 +-
 tools/perf/builtin-inject.c                   |   81 +-
 tools/perf/builtin-kvm.c                      |   14 +-
 tools/perf/builtin-kwork.c                    |    8 +-
 tools/perf/builtin-lock.c                     |    2 +-
 tools/perf/builtin-record.c                   |   95 +-
 tools/perf/builtin-report.c                   |    6 +-
 tools/perf/builtin-sched.c                    |   26 +-
 tools/perf/builtin-script.c                   |  895 +++----
 tools/perf/builtin-stat.c                     |   81 +-
 tools/perf/builtin-top.c                      |  104 +-
 tools/perf/builtin-trace.c                    |   60 +-
 tools/perf/python/SchedGui.py                 |  219 ++
 tools/perf/python/arm-cs-trace-disasm.py      |  338 +++
 tools/perf/python/check-perf-trace.py         |  113 +
 tools/perf/python/compaction-times.py         |  326 +++
 tools/perf/python/counting.py                 |    1 +
 tools/perf/python/event_analyzing_sample.py   |  296 +++
 tools/perf/python/export-to-postgresql.py     |  697 ++++++
 tools/perf/python/export-to-sqlite.py         |  380 +++
 .../python/exported-sql-viewer.py             |    6 +-
 tools/perf/python/failed-syscalls-by-pid.py   |  119 +
 tools/perf/python/failed-syscalls.py          |   78 +
 tools/perf/python/flamegraph.py               |  250 ++
 tools/perf/python/futex-contention.py         |   87 +
 tools/perf/python/gecko.py                    |  380 +++
 tools/perf/python/intel-pt-events.py          |  435 ++++
 tools/perf/python/libxed.py                   |  122 +
 .../{scripts => }/python/mem-phys-addr.py     |   66 +-
 tools/perf/python/net_dropmonitor.py          |   58 +
 tools/perf/python/netdev-times.py             |  472 ++++
 .../{scripts => }/python/parallel-perf.py     |    0
 tools/perf/python/perf.pyi                    |  579 +++++
 tools/perf/python/perf_live.py                |   48 +
 tools/perf/python/powerpc-hcalls.py           |  211 ++
 tools/perf/python/rw-by-file.py               |  103 +
 tools/perf/python/rw-by-pid.py                |  158 ++
 tools/perf/python/rwtop.py                    |  219 ++
 tools/perf/python/sched-migration.py          |  469 ++++
 tools/perf/python/sctop.py                    |  174 ++
 tools/perf/python/stackcollapse.py            |  126 +
 tools/perf/python/stat-cpi.py                 |  151 ++
 tools/perf/python/syscall-counts-by-pid.py    |   88 +
 tools/perf/python/syscall-counts.py           |   72 +
 tools/perf/python/task-analyzer.py            |  547 ++++
 tools/perf/python/tracepoint.py               |    1 +
 tools/perf/python/twatch.py                   |    1 +
 tools/perf/python/wakeup-latency.py           |   88 +
 tools/perf/scripts/Build                      |    4 -
 tools/perf/scripts/perl/Perf-Trace-Util/Build |    9 -
 .../scripts/perl/Perf-Trace-Util/Context.c    |  122 -
 .../scripts/perl/Perf-Trace-Util/Context.xs   |   42 -
 .../scripts/perl/Perf-Trace-Util/Makefile.PL  |   18 -
 .../perf/scripts/perl/Perf-Trace-Util/README  |   59 -
 .../Perf-Trace-Util/lib/Perf/Trace/Context.pm |   55 -
 .../Perf-Trace-Util/lib/Perf/Trace/Core.pm    |  192 --
 .../Perf-Trace-Util/lib/Perf/Trace/Util.pm    |   94 -
 .../perf/scripts/perl/Perf-Trace-Util/typemap |    1 -
 .../scripts/perl/bin/check-perf-trace-record  |    2 -
 .../scripts/perl/bin/failed-syscalls-record   |    3 -
 .../scripts/perl/bin/failed-syscalls-report   |   10 -
 tools/perf/scripts/perl/bin/rw-by-file-record |    3 -
 tools/perf/scripts/perl/bin/rw-by-file-report |   10 -
 tools/perf/scripts/perl/bin/rw-by-pid-record  |    2 -
 tools/perf/scripts/perl/bin/rw-by-pid-report  |    3 -
 tools/perf/scripts/perl/bin/rwtop-record      |    2 -
 tools/perf/scripts/perl/bin/rwtop-report      |   20 -
 .../scripts/perl/bin/wakeup-latency-record    |    6 -
 .../scripts/perl/bin/wakeup-latency-report    |    3 -
 tools/perf/scripts/perl/check-perf-trace.pl   |  106 -
 tools/perf/scripts/perl/failed-syscalls.pl    |   47 -
 tools/perf/scripts/perl/rw-by-file.pl         |  106 -
 tools/perf/scripts/perl/rw-by-pid.pl          |  184 --
 tools/perf/scripts/perl/rwtop.pl              |  203 --
 tools/perf/scripts/perl/wakeup-latency.pl     |  107 -
 .../perf/scripts/python/Perf-Trace-Util/Build |    4 -
 .../scripts/python/Perf-Trace-Util/Context.c  |  225 --
 .../Perf-Trace-Util/lib/Perf/Trace/Core.py    |  116 -
 .../lib/Perf/Trace/EventClass.py              |   97 -
 .../lib/Perf/Trace/SchedGui.py                |  184 --
 .../Perf-Trace-Util/lib/Perf/Trace/Util.py    |   92 -
 .../scripts/python/arm-cs-trace-disasm.py     |  355 ---
 .../python/bin/compaction-times-record        |    2 -
 .../python/bin/compaction-times-report        |    4 -
 .../python/bin/event_analyzing_sample-record  |    8 -
 .../python/bin/event_analyzing_sample-report  |    3 -
 .../python/bin/export-to-postgresql-record    |    8 -
 .../python/bin/export-to-postgresql-report    |   29 -
 .../python/bin/export-to-sqlite-record        |    8 -
 .../python/bin/export-to-sqlite-report        |   29 -
 .../python/bin/failed-syscalls-by-pid-record  |    3 -
 .../python/bin/failed-syscalls-by-pid-report  |   10 -
 .../perf/scripts/python/bin/flamegraph-record |    2 -
 .../perf/scripts/python/bin/flamegraph-report |    3 -
 .../python/bin/futex-contention-record        |    2 -
 .../python/bin/futex-contention-report        |    4 -
 tools/perf/scripts/python/bin/gecko-record    |    2 -
 tools/perf/scripts/python/bin/gecko-report    |    7 -
 .../scripts/python/bin/intel-pt-events-record |   13 -
 .../scripts/python/bin/intel-pt-events-report |    3 -
 .../scripts/python/bin/mem-phys-addr-record   |   19 -
 .../scripts/python/bin/mem-phys-addr-report   |    3 -
 .../scripts/python/bin/net_dropmonitor-record |    2 -
 .../scripts/python/bin/net_dropmonitor-report |    4 -
 .../scripts/python/bin/netdev-times-record    |    8 -
 .../scripts/python/bin/netdev-times-report    |    5 -
 .../scripts/python/bin/powerpc-hcalls-record  |    2 -
 .../scripts/python/bin/powerpc-hcalls-report  |    2 -
 .../scripts/python/bin/sched-migration-record |    2 -
 .../scripts/python/bin/sched-migration-report |    3 -
 tools/perf/scripts/python/bin/sctop-record    |    3 -
 tools/perf/scripts/python/bin/sctop-report    |   24 -
 .../scripts/python/bin/stackcollapse-record   |    8 -
 .../scripts/python/bin/stackcollapse-report   |    3 -
 .../python/bin/syscall-counts-by-pid-record   |    3 -
 .../python/bin/syscall-counts-by-pid-report   |   10 -
 .../scripts/python/bin/syscall-counts-record  |    3 -
 .../scripts/python/bin/syscall-counts-report  |   10 -
 .../scripts/python/bin/task-analyzer-record   |    2 -
 .../scripts/python/bin/task-analyzer-report   |    3 -
 tools/perf/scripts/python/check-perf-trace.py |   84 -
 tools/perf/scripts/python/compaction-times.py |  311 ---
 .../scripts/python/event_analyzing_sample.py  |  192 --
 .../scripts/python/export-to-postgresql.py    | 1114 ---------
 tools/perf/scripts/python/export-to-sqlite.py |  799 ------
 .../scripts/python/failed-syscalls-by-pid.py  |   79 -
 tools/perf/scripts/python/flamegraph.py       |  267 --
 tools/perf/scripts/python/futex-contention.py |   57 -
 tools/perf/scripts/python/gecko.py            |  395 ---
 tools/perf/scripts/python/intel-pt-events.py  |  494 ----
 tools/perf/scripts/python/libxed.py           |  107 -
 tools/perf/scripts/python/net_dropmonitor.py  |   78 -
 tools/perf/scripts/python/netdev-times.py     |  473 ----
 tools/perf/scripts/python/powerpc-hcalls.py   |  202 --
 tools/perf/scripts/python/sched-migration.py  |  462 ----
 tools/perf/scripts/python/sctop.py            |   89 -
 tools/perf/scripts/python/stackcollapse.py    |  127 -
 tools/perf/scripts/python/stat-cpi.py         |   79 -
 .../scripts/python/syscall-counts-by-pid.py   |   75 -
 tools/perf/scripts/python/syscall-counts.py   |   65 -
 tools/perf/scripts/python/task-analyzer.py    |  934 -------
 tools/perf/tests/backward-ring-buffer.c       |   26 +-
 tools/perf/tests/code-reading.c               |   14 +-
 tools/perf/tests/dlfilter-test.c              |    8 +-
 tools/perf/tests/event-times.c                |    6 +-
 tools/perf/tests/event_update.c               |    4 +-
 tools/perf/tests/evsel-roundtrip-name.c       |    8 +-
 tools/perf/tests/evsel-tp-sched.c             |    4 +-
 tools/perf/tests/expand-cgroup.c              |   12 +-
 tools/perf/tests/hists_cumulate.c             |    2 +-
 tools/perf/tests/hists_filter.c               |    2 +-
 tools/perf/tests/hists_link.c                 |    2 +-
 tools/perf/tests/hists_output.c               |    2 +-
 tools/perf/tests/hwmon_pmu.c                  |   21 +-
 tools/perf/tests/keep-tracking.c              |   10 +-
 tools/perf/tests/make                         |    9 +-
 tools/perf/tests/mmap-basic.c                 |   42 +-
 tools/perf/tests/openat-syscall-all-cpus.c    |    6 +-
 tools/perf/tests/openat-syscall-tp-fields.c   |   26 +-
 tools/perf/tests/openat-syscall.c             |    6 +-
 tools/perf/tests/parse-events.c               |  139 +-
 tools/perf/tests/parse-metric.c               |    8 +-
 tools/perf/tests/parse-no-sample-id-all.c     |    2 +-
 tools/perf/tests/perf-record.c                |   38 +-
 tools/perf/tests/perf-time-to-tsc.c           |   12 +-
 tools/perf/tests/pfm.c                        |   12 +-
 tools/perf/tests/pmu-events.c                 |   11 +-
 tools/perf/tests/pmu.c                        |    4 +-
 tools/perf/tests/sample-parsing.c             |   39 +-
 .../perf/tests/shell/lib/perf_brstack_max.py  |   43 +
 tools/perf/tests/shell/script.sh              |    2 +-
 tools/perf/tests/shell/script_perl.sh         |  102 -
 tools/perf/tests/shell/script_python.sh       |  113 -
 .../tests/shell/test_arm_coresight_disasm.sh  |   12 +-
 tools/perf/tests/shell/test_intel_pt.sh       |   35 +-
 tools/perf/tests/shell/test_task_analyzer.sh  |   79 +-
 tools/perf/tests/sw-clock.c                   |   20 +-
 tools/perf/tests/switch-tracking.c            |   10 +-
 tools/perf/tests/task-exit.c                  |   20 +-
 tools/perf/tests/time-utils-test.c            |   14 +-
 tools/perf/tests/tool_pmu.c                   |    7 +-
 tools/perf/tests/topology.c                   |    4 +-
 tools/perf/ui/browsers/annotate.c             |    2 +-
 tools/perf/ui/browsers/hists.c                |   22 +-
 tools/perf/ui/browsers/scripts.c              |    7 +-
 tools/perf/util/Build                         |    1 -
 tools/perf/util/amd-sample-raw.c              |    2 +-
 tools/perf/util/annotate-data.c               |    2 +-
 tools/perf/util/annotate.c                    |   10 +-
 tools/perf/util/arm-spe.c                     |    7 +-
 tools/perf/util/auxtrace.c                    |   14 +-
 tools/perf/util/block-info.c                  |    4 +-
 tools/perf/util/bpf_counter.c                 |    2 +-
 tools/perf/util/bpf_counter_cgroup.c          |   10 +-
 tools/perf/util/bpf_ftrace.c                  |    9 +-
 tools/perf/util/bpf_lock_contention.c         |   12 +-
 tools/perf/util/bpf_off_cpu.c                 |   44 +-
 tools/perf/util/bpf_trace_augment.c           |    8 +-
 tools/perf/util/cgroup.c                      |   26 +-
 tools/perf/util/cs-etm.c                      |    6 +-
 tools/perf/util/data-convert-bt.c             |    2 +-
 tools/perf/util/data.c                        |   26 +-
 tools/perf/util/data.h                        |    4 +-
 tools/perf/util/evlist.c                      |  487 ++--
 tools/perf/util/evlist.h                      |  273 +-
 tools/perf/util/evsel.c                       |  109 +-
 tools/perf/util/evsel.h                       |   35 +-
 tools/perf/util/expr.c                        |    2 +-
 tools/perf/util/header.c                      |   51 +-
 tools/perf/util/header.h                      |    2 +-
 tools/perf/util/intel-bts.c                   |    3 +-
 tools/perf/util/intel-pt.c                    |   13 +-
 tools/perf/util/intel-tpebs.c                 |    7 +-
 tools/perf/util/map.h                         |    9 +-
 tools/perf/util/metricgroup.c                 |   12 +-
 tools/perf/util/parse-events.c                |   10 +-
 tools/perf/util/parse-events.y                |    2 +-
 tools/perf/util/perf_api_probe.c              |   20 +-
 tools/perf/util/pfm.c                         |    4 +-
 tools/perf/util/print-events.c                |    2 +-
 tools/perf/util/print_insn.h                  |    5 +-
 tools/perf/util/python.c                      | 1854 ++++++++++++--
 tools/perf/util/record.c                      |   11 +-
 tools/perf/util/s390-sample-raw.c             |   19 +-
 tools/perf/util/sample-raw.c                  |    4 +-
 tools/perf/util/sample.c                      |   17 +-
 tools/perf/util/scripting-engines/Build       |    9 -
 .../util/scripting-engines/trace-event-perl.c |  773 ------
 .../scripting-engines/trace-event-python.c    | 2209 -----------------
 tools/perf/util/session.c                     |   59 +-
 tools/perf/util/sideband_evlist.c             |   40 +-
 tools/perf/util/sort.c                        |    2 +-
 tools/perf/util/stat-display.c                |    6 +-
 tools/perf/util/stat-shadow.c                 |   24 +-
 tools/perf/util/stat.c                        |   20 +-
 tools/perf/util/stream.c                      |    4 +-
 tools/perf/util/synthetic-events.c            |   36 +-
 tools/perf/util/synthetic-events.h            |    6 +-
 tools/perf/util/time-utils.c                  |   12 +-
 tools/perf/util/top.c                         |    4 +-
 tools/perf/util/trace-event-parse.c           |   65 -
 tools/perf/util/trace-event-scripting.c       |  410 ---
 tools/perf/util/trace-event.h                 |   75 +-
 268 files changed, 11291 insertions(+), 15964 deletions(-)
 delete mode 100644 tools/build/feature/test-libperl.c
 delete mode 100644 tools/build/feature/test-libpython.c
 create mode 100644 tools/build/feature/test-python-module.c
 delete mode 100644 tools/perf/Documentation/perf-script-perl.txt
 create mode 100755 tools/perf/python/SchedGui.py
 create mode 100755 tools/perf/python/arm-cs-trace-disasm.py
 create mode 100755 tools/perf/python/check-perf-trace.py
 create mode 100755 tools/perf/python/compaction-times.py
 create mode 100755 tools/perf/python/event_analyzing_sample.py
 create mode 100755 tools/perf/python/export-to-postgresql.py
 create mode 100755 tools/perf/python/export-to-sqlite.py
 rename tools/perf/{scripts => }/python/exported-sql-viewer.py (99%)
 create mode 100755 tools/perf/python/failed-syscalls-by-pid.py
 create mode 100755 tools/perf/python/failed-syscalls.py
 create mode 100755 tools/perf/python/flamegraph.py
 create mode 100755 tools/perf/python/futex-contention.py
 create mode 100755 tools/perf/python/gecko.py
 create mode 100755 tools/perf/python/intel-pt-events.py
 create mode 100755 tools/perf/python/libxed.py
 rename tools/perf/{scripts => }/python/mem-phys-addr.py (73%)
 mode change 100644 => 100755
 create mode 100755 tools/perf/python/net_dropmonitor.py
 create mode 100755 tools/perf/python/netdev-times.py
 rename tools/perf/{scripts => }/python/parallel-perf.py (100%)
 create mode 100644 tools/perf/python/perf.pyi
 create mode 100755 tools/perf/python/perf_live.py
 create mode 100755 tools/perf/python/powerpc-hcalls.py
 create mode 100755 tools/perf/python/rw-by-file.py
 create mode 100755 tools/perf/python/rw-by-pid.py
 create mode 100755 tools/perf/python/rwtop.py
 create mode 100755 tools/perf/python/sched-migration.py
 create mode 100755 tools/perf/python/sctop.py
 create mode 100755 tools/perf/python/stackcollapse.py
 create mode 100755 tools/perf/python/stat-cpi.py
 create mode 100755 tools/perf/python/syscall-counts-by-pid.py
 create mode 100755 tools/perf/python/syscall-counts.py
 create mode 100755 tools/perf/python/task-analyzer.py
 create mode 100755 tools/perf/python/wakeup-latency.py
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/Build
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/Context.c
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/Context.xs
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/Makefile.PL
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/README
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Context.pm
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Core.pm
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm
 delete mode 100644 tools/perf/scripts/perl/Perf-Trace-Util/typemap
 delete mode 100644 tools/perf/scripts/perl/bin/check-perf-trace-record
 delete mode 100644 tools/perf/scripts/perl/bin/failed-syscalls-record
 delete mode 100644 tools/perf/scripts/perl/bin/failed-syscalls-report
 delete mode 100644 tools/perf/scripts/perl/bin/rw-by-file-record
 delete mode 100644 tools/perf/scripts/perl/bin/rw-by-file-report
 delete mode 100644 tools/perf/scripts/perl/bin/rw-by-pid-record
 delete mode 100644 tools/perf/scripts/perl/bin/rw-by-pid-report
 delete mode 100644 tools/perf/scripts/perl/bin/rwtop-record
 delete mode 100644 tools/perf/scripts/perl/bin/rwtop-report
 delete mode 100644 tools/perf/scripts/perl/bin/wakeup-latency-record
 delete mode 100644 tools/perf/scripts/perl/bin/wakeup-latency-report
 delete mode 100644 tools/perf/scripts/perl/check-perf-trace.pl
 delete mode 100644 tools/perf/scripts/perl/failed-syscalls.pl
 delete mode 100644 tools/perf/scripts/perl/rw-by-file.pl
 delete mode 100644 tools/perf/scripts/perl/rw-by-pid.pl
 delete mode 100644 tools/perf/scripts/perl/rwtop.pl
 delete mode 100644 tools/perf/scripts/perl/wakeup-latency.pl
 delete mode 100644 tools/perf/scripts/python/Perf-Trace-Util/Build
 delete mode 100644 tools/perf/scripts/python/Perf-Trace-Util/Context.c
 delete mode 100644 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py
 delete mode 100755 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/EventClass.py
 delete mode 100644 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/SchedGui.py
 delete mode 100644 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py
 delete mode 100755 tools/perf/scripts/python/arm-cs-trace-disasm.py
 delete mode 100644 tools/perf/scripts/python/bin/compaction-times-record
 delete mode 100644 tools/perf/scripts/python/bin/compaction-times-report
 delete mode 100644 tools/perf/scripts/python/bin/event_analyzing_sample-record
 delete mode 100644 tools/perf/scripts/python/bin/event_analyzing_sample-report
 delete mode 100644 tools/perf/scripts/python/bin/export-to-postgresql-record
 delete mode 100644 tools/perf/scripts/python/bin/export-to-postgresql-report
 delete mode 100644 tools/perf/scripts/python/bin/export-to-sqlite-record
 delete mode 100644 tools/perf/scripts/python/bin/export-to-sqlite-report
 delete mode 100644 tools/perf/scripts/python/bin/failed-syscalls-by-pid-record
 delete mode 100644 tools/perf/scripts/python/bin/failed-syscalls-by-pid-report
 delete mode 100755 tools/perf/scripts/python/bin/flamegraph-record
 delete mode 100755 tools/perf/scripts/python/bin/flamegraph-report
 delete mode 100644 tools/perf/scripts/python/bin/futex-contention-record
 delete mode 100644 tools/perf/scripts/python/bin/futex-contention-report
 delete mode 100644 tools/perf/scripts/python/bin/gecko-record
 delete mode 100755 tools/perf/scripts/python/bin/gecko-report
 delete mode 100644 tools/perf/scripts/python/bin/intel-pt-events-record
 delete mode 100644 tools/perf/scripts/python/bin/intel-pt-events-report
 delete mode 100644 tools/perf/scripts/python/bin/mem-phys-addr-record
 delete mode 100644 tools/perf/scripts/python/bin/mem-phys-addr-report
 delete mode 100755 tools/perf/scripts/python/bin/net_dropmonitor-record
 delete mode 100755 tools/perf/scripts/python/bin/net_dropmonitor-report
 delete mode 100644 tools/perf/scripts/python/bin/netdev-times-record
 delete mode 100644 tools/perf/scripts/python/bin/netdev-times-report
 delete mode 100644 tools/perf/scripts/python/bin/powerpc-hcalls-record
 delete mode 100644 tools/perf/scripts/python/bin/powerpc-hcalls-report
 delete mode 100644 tools/perf/scripts/python/bin/sched-migration-record
 delete mode 100644 tools/perf/scripts/python/bin/sched-migration-report
 delete mode 100644 tools/perf/scripts/python/bin/sctop-record
 delete mode 100644 tools/perf/scripts/python/bin/sctop-report
 delete mode 100755 tools/perf/scripts/python/bin/stackcollapse-record
 delete mode 100755 tools/perf/scripts/python/bin/stackcollapse-report
 delete mode 100644 tools/perf/scripts/python/bin/syscall-counts-by-pid-record
 delete mode 100644 tools/perf/scripts/python/bin/syscall-counts-by-pid-report
 delete mode 100644 tools/perf/scripts/python/bin/syscall-counts-record
 delete mode 100644 tools/perf/scripts/python/bin/syscall-counts-report
 delete mode 100755 tools/perf/scripts/python/bin/task-analyzer-record
 delete mode 100755 tools/perf/scripts/python/bin/task-analyzer-report
 delete mode 100644 tools/perf/scripts/python/check-perf-trace.py
 delete mode 100644 tools/perf/scripts/python/compaction-times.py
 delete mode 100644 tools/perf/scripts/python/event_analyzing_sample.py
 delete mode 100644 tools/perf/scripts/python/export-to-postgresql.py
 delete mode 100644 tools/perf/scripts/python/export-to-sqlite.py
 delete mode 100644 tools/perf/scripts/python/failed-syscalls-by-pid.py
 delete mode 100755 tools/perf/scripts/python/flamegraph.py
 delete mode 100644 tools/perf/scripts/python/futex-contention.py
 delete mode 100644 tools/perf/scripts/python/gecko.py
 delete mode 100644 tools/perf/scripts/python/intel-pt-events.py
 delete mode 100644 tools/perf/scripts/python/libxed.py
 delete mode 100755 tools/perf/scripts/python/net_dropmonitor.py
 delete mode 100644 tools/perf/scripts/python/netdev-times.py
 delete mode 100644 tools/perf/scripts/python/powerpc-hcalls.py
 delete mode 100644 tools/perf/scripts/python/sched-migration.py
 delete mode 100644 tools/perf/scripts/python/sctop.py
 delete mode 100755 tools/perf/scripts/python/stackcollapse.py
 delete mode 100644 tools/perf/scripts/python/stat-cpi.py
 delete mode 100644 tools/perf/scripts/python/syscall-counts-by-pid.py
 delete mode 100644 tools/perf/scripts/python/syscall-counts.py
 delete mode 100755 tools/perf/scripts/python/task-analyzer.py
 create mode 100644 tools/perf/tests/shell/lib/perf_brstack_max.py
 delete mode 100755 tools/perf/tests/shell/script_perl.sh
 delete mode 100755 tools/perf/tests/shell/script_python.sh
 delete mode 100644 tools/perf/util/scripting-engines/Build
 delete mode 100644 tools/perf/util/scripting-engines/trace-event-perl.c
 delete mode 100644 tools/perf/util/scripting-engines/trace-event-python.c
 delete mode 100644 tools/perf/util/trace-event-scripting.c

-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply

* [PATCH v5 01/58] perf inject: Fix itrace branch stack synthesis
From: Ian Rogers @ 2026-04-24 16:46 UTC (permalink / raw)
  To: acme, adrian.hunter, james.clark, leo.yan, namhyung, tmricht
  Cc: alice.mei.rogers, dapeng1.mi, linux-arm-kernel, linux-kernel,
	linux-perf-users, mingo, peterz, Ian Rogers
In-Reply-To: <20260424164721.2229025-1-irogers@google.com>

When using "perf inject --itrace=L" to synthesize branch stacks from
AUX data, several issues caused failures:

1. The synthesized samples were delivered without the
   PERF_SAMPLE_BRANCH_STACK flag if it was not in the original event's
   sample_type. Fixed by using sample_type | evsel->synth_sample_type
   in intel_pt_deliver_synth_event.

2. The record layout was misaligned because of inconsistent handling
   of PERF_SAMPLE_BRANCH_HW_INDEX. Fixed by explicitly writing nr and
   hw_idx in perf_event__synthesize_sample.

3. Modifying evsel->core.attr.sample_type early in __cmd_inject caused
   parse failures for subsequent records in the input file. Fixed by
   moving this modification to just before writing the header.

4. perf_event__repipe_sample was narrowed to only synthesize samples
   when branch stack injection was requested, and restored the use of
   perf_inject__cut_auxtrace_sample as a fallback to preserve
   functionality.

Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <irogers@google.com>
---
Issues fixed in v2:

1. Potential Heap Overflow in perf_event__repipe_sample : Addressed by
   adding a check that prints an error and returns -EFAULT if the
   calculated event size exceeds PERF_SAMPLE_MAX_SIZE , as you
   requested.

2. Header vs Payload Mismatch in __cmd_inject : Addressed by narrowing
   the condition so that HEADER_BRANCH_STACK is only set in the file
   header if add_last_branch was true.

3. NULL Pointer Dereference in intel-pt.c : Addressed by updating the
   condition in intel_pt_do_synth_pebs_sample to fill sample.
   branch_stack if it was synthesized, even if not in the original
   sample_type .

4. Unsafe Reads for events lacking HW_INDEX in synthetic-events.c :
   Addressed by using the perf_sample__branch_entries() macro and
   checking sample->no_hw_idx .

5. Size mismatch in perf_event__sample_event_size : Addressed by
   passing branch_sample_type to it and conditioning the hw_idx size on
   PERF_SAMPLE_BRANCH_HW_INDEX .
---
 tools/perf/bench/inject-buildid.c  |  9 ++--
 tools/perf/builtin-inject.c        | 77 ++++++++++++++++++++++++++++--
 tools/perf/tests/dlfilter-test.c   |  8 +++-
 tools/perf/tests/sample-parsing.c  |  5 +-
 tools/perf/util/arm-spe.c          |  7 ++-
 tools/perf/util/cs-etm.c           |  6 ++-
 tools/perf/util/intel-bts.c        |  3 +-
 tools/perf/util/intel-pt.c         | 13 +++--
 tools/perf/util/synthetic-events.c | 25 +++++++---
 tools/perf/util/synthetic-events.h |  6 ++-
 10 files changed, 129 insertions(+), 30 deletions(-)

diff --git a/tools/perf/bench/inject-buildid.c b/tools/perf/bench/inject-buildid.c
index aad572a78d7f..bfd2c5ec9488 100644
--- a/tools/perf/bench/inject-buildid.c
+++ b/tools/perf/bench/inject-buildid.c
@@ -228,9 +228,12 @@ static ssize_t synthesize_sample(struct bench_data *data, struct bench_dso *dso,
 
 	event.header.type = PERF_RECORD_SAMPLE;
 	event.header.misc = PERF_RECORD_MISC_USER;
-	event.header.size = perf_event__sample_event_size(&sample, bench_sample_type, 0);
-
-	perf_event__synthesize_sample(&event, bench_sample_type, 0, &sample);
+	event.header.size = perf_event__sample_event_size(&sample, bench_sample_type,
+							   /*read_format=*/0,
+							   /*branch_sample_type=*/0);
+	perf_event__synthesize_sample(&event, bench_sample_type,
+				      /*read_format=*/0,
+				      /*branch_sample_type=*/0, &sample);
 
 	return writen(data->input_pipe[1], &event, event.header.size);
 }
diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
index f174bc69cec4..88c0ef4f5ff1 100644
--- a/tools/perf/builtin-inject.c
+++ b/tools/perf/builtin-inject.c
@@ -375,7 +375,59 @@ static int perf_event__repipe_sample(const struct perf_tool *tool,
 
 	build_id__mark_dso_hit(tool, event, sample, evsel, machine);
 
-	if (inject->itrace_synth_opts.set && sample->aux_sample.size) {
+	if (inject->itrace_synth_opts.set &&
+	    (inject->itrace_synth_opts.last_branch ||
+	     inject->itrace_synth_opts.add_last_branch)) {
+		union perf_event *event_copy = (void *)inject->event_copy;
+		struct branch_stack dummy_bs = { .nr = 0 };
+		int err;
+		size_t sz;
+		u64 orig_type = evsel->core.attr.sample_type;
+		u64 orig_branch_type = evsel->core.attr.branch_sample_type;
+
+		if (event_copy == NULL) {
+			inject->event_copy = malloc(PERF_SAMPLE_MAX_SIZE);
+			if (!inject->event_copy)
+				return -ENOMEM;
+
+			event_copy = (void *)inject->event_copy;
+		}
+
+		if (!sample->branch_stack)
+			sample->branch_stack = &dummy_bs;
+
+		if (inject->itrace_synth_opts.add_last_branch) {
+			/* Temporarily add in type bits for synthesis. */
+			evsel->core.attr.sample_type |= PERF_SAMPLE_BRANCH_STACK;
+			evsel->core.attr.branch_sample_type |= PERF_SAMPLE_BRANCH_HW_INDEX;
+			evsel->core.attr.sample_type &= ~PERF_SAMPLE_AUX;
+		}
+
+		sz = perf_event__sample_event_size(sample, evsel->core.attr.sample_type,
+						   evsel->core.attr.read_format,
+						   evsel->core.attr.branch_sample_type);
+
+		if (sz > PERF_SAMPLE_MAX_SIZE) {
+			pr_err("Sample size %zu exceeds max size %d\n", sz, PERF_SAMPLE_MAX_SIZE);
+			return -EFAULT;
+		}
+
+		event_copy->header.type = PERF_RECORD_SAMPLE;
+		event_copy->header.size = sz;
+
+		err = perf_event__synthesize_sample(event_copy, evsel->core.attr.sample_type,
+						    evsel->core.attr.read_format,
+						    evsel->core.attr.branch_sample_type, sample);
+
+		evsel->core.attr.sample_type = orig_type;
+		evsel->core.attr.branch_sample_type = orig_branch_type;
+
+		if (err) {
+			pr_err("Failed to synthesize sample\n");
+			return err;
+		}
+		event = event_copy;
+	} else if (inject->itrace_synth_opts.set && sample->aux_sample.size) {
 		event = perf_inject__cut_auxtrace_sample(inject, event, sample);
 		if (IS_ERR(event))
 			return PTR_ERR(event);
@@ -464,7 +516,8 @@ static int perf_event__convert_sample_callchain(const struct perf_tool *tool,
 	sample_type &= ~(PERF_SAMPLE_STACK_USER | PERF_SAMPLE_REGS_USER);
 
 	perf_event__synthesize_sample(event_copy, sample_type,
-				      evsel->core.attr.read_format, sample);
+				      evsel->core.attr.read_format,
+				      evsel->core.attr.branch_sample_type, sample);
 	return perf_event__repipe_synth(tool, event_copy);
 }
 
@@ -1100,7 +1153,8 @@ static int perf_inject__sched_stat(const struct perf_tool *tool,
 	sample_sw.period = sample->period;
 	sample_sw.time	 = sample->time;
 	perf_event__synthesize_sample(event_sw, evsel->core.attr.sample_type,
-				      evsel->core.attr.read_format, &sample_sw);
+				      evsel->core.attr.read_format,
+				      evsel->core.attr.branch_sample_type, &sample_sw);
 	build_id__mark_dso_hit(tool, event_sw, &sample_sw, evsel, machine);
 	ret = perf_event__repipe(tool, event_sw, &sample_sw, machine);
 	perf_sample__exit(&sample_sw);
@@ -2434,12 +2488,25 @@ static int __cmd_inject(struct perf_inject *inject)
 		 * synthesized hardware events, so clear the feature flag.
 		 */
 		if (inject->itrace_synth_opts.set) {
+			struct evsel *evsel;
+
 			perf_header__clear_feat(&session->header,
 						HEADER_AUXTRACE);
-			if (inject->itrace_synth_opts.last_branch ||
-			    inject->itrace_synth_opts.add_last_branch)
+
+			evlist__for_each_entry(session->evlist, evsel) {
+				evsel->core.attr.sample_type &= ~PERF_SAMPLE_AUX;
+			}
+
+			if (inject->itrace_synth_opts.add_last_branch) {
 				perf_header__set_feat(&session->header,
 						      HEADER_BRANCH_STACK);
+
+				evlist__for_each_entry(session->evlist, evsel) {
+					evsel->core.attr.sample_type |= PERF_SAMPLE_BRANCH_STACK;
+					evsel->core.attr.branch_sample_type |=
+						PERF_SAMPLE_BRANCH_HW_INDEX;
+				}
+			}
 		}
 
 		/*
diff --git a/tools/perf/tests/dlfilter-test.c b/tools/perf/tests/dlfilter-test.c
index e63790c61d53..204663571943 100644
--- a/tools/perf/tests/dlfilter-test.c
+++ b/tools/perf/tests/dlfilter-test.c
@@ -188,8 +188,12 @@ static int write_sample(struct test_data *td, u64 sample_type, u64 id, pid_t pid
 
 	event->header.type = PERF_RECORD_SAMPLE;
 	event->header.misc = PERF_RECORD_MISC_USER;
-	event->header.size = perf_event__sample_event_size(&sample, sample_type, 0);
-	err = perf_event__synthesize_sample(event, sample_type, 0, &sample);
+	event->header.size = perf_event__sample_event_size(&sample, sample_type,
+							   /*read_format=*/0,
+							   /*branch_sample_type=*/0);
+	err = perf_event__synthesize_sample(event, sample_type,
+					    /*read_format=*/0,
+					    /*branch_sample_type=*/0, &sample);
 	if (err)
 		return test_result("perf_event__synthesize_sample() failed", TEST_FAIL);
 
diff --git a/tools/perf/tests/sample-parsing.c b/tools/perf/tests/sample-parsing.c
index a7327c942ca2..55f0b73ca20e 100644
--- a/tools/perf/tests/sample-parsing.c
+++ b/tools/perf/tests/sample-parsing.c
@@ -310,7 +310,8 @@ static int do_test(u64 sample_type, u64 sample_regs, u64 read_format)
 		sample.read.one.lost  = 1;
 	}
 
-	sz = perf_event__sample_event_size(&sample, sample_type, read_format);
+	sz = perf_event__sample_event_size(&sample, sample_type, read_format,
+					   evsel.core.attr.branch_sample_type);
 	bufsz = sz + 4096; /* Add a bit for overrun checking */
 	event = malloc(bufsz);
 	if (!event) {
@@ -324,7 +325,7 @@ static int do_test(u64 sample_type, u64 sample_regs, u64 read_format)
 	event->header.size = sz;
 
 	err = perf_event__synthesize_sample(event, sample_type, read_format,
-					    &sample);
+					    evsel.core.attr.branch_sample_type, &sample);
 	if (err) {
 		pr_debug("%s failed for sample_type %#"PRIx64", error %d\n",
 			 "perf_event__synthesize_sample", sample_type, err);
diff --git a/tools/perf/util/arm-spe.c b/tools/perf/util/arm-spe.c
index e5835042acdf..c4ed9f10e731 100644
--- a/tools/perf/util/arm-spe.c
+++ b/tools/perf/util/arm-spe.c
@@ -484,8 +484,11 @@ static void arm_spe__prep_branch_stack(struct arm_spe_queue *speq)
 
 static int arm_spe__inject_event(union perf_event *event, struct perf_sample *sample, u64 type)
 {
-	event->header.size = perf_event__sample_event_size(sample, type, 0);
-	return perf_event__synthesize_sample(event, type, 0, sample);
+	event->header.type = PERF_RECORD_SAMPLE;
+	event->header.size = perf_event__sample_event_size(sample, type, /*read_format=*/0,
+							   /*branch_sample_type=*/0);
+	return perf_event__synthesize_sample(event, type, /*read_format=*/0,
+					     /*branch_sample_type=*/0, sample);
 }
 
 static inline int
diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c
index 8a639d2e51a4..1ebc1a6a5e75 100644
--- a/tools/perf/util/cs-etm.c
+++ b/tools/perf/util/cs-etm.c
@@ -1425,8 +1425,10 @@ static void cs_etm__update_last_branch_rb(struct cs_etm_queue *etmq,
 static int cs_etm__inject_event(union perf_event *event,
 			       struct perf_sample *sample, u64 type)
 {
-	event->header.size = perf_event__sample_event_size(sample, type, 0);
-	return perf_event__synthesize_sample(event, type, 0, sample);
+	event->header.size = perf_event__sample_event_size(sample, type, /*read_format=*/0,
+							   /*branch_sample_type=*/0);
+	return perf_event__synthesize_sample(event, type, /*read_format=*/0,
+					     /*branch_sample_type=*/0, sample);
 }
 
 
diff --git a/tools/perf/util/intel-bts.c b/tools/perf/util/intel-bts.c
index 382255393fb3..0b18ebd13f7c 100644
--- a/tools/perf/util/intel-bts.c
+++ b/tools/perf/util/intel-bts.c
@@ -303,7 +303,8 @@ static int intel_bts_synth_branch_sample(struct intel_bts_queue *btsq,
 		event.sample.header.size = bts->branches_event_size;
 		ret = perf_event__synthesize_sample(&event,
 						    bts->branches_sample_type,
-						    0, &sample);
+						    /*read_format=*/0, /*branch_sample_type=*/0,
+						    &sample);
 		if (ret)
 			return ret;
 	}
diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
index fc9eec8b54b8..2dce6106c038 100644
--- a/tools/perf/util/intel-pt.c
+++ b/tools/perf/util/intel-pt.c
@@ -1731,8 +1731,12 @@ static void intel_pt_prep_b_sample(struct intel_pt *pt,
 static int intel_pt_inject_event(union perf_event *event,
 				 struct perf_sample *sample, u64 type)
 {
-	event->header.size = perf_event__sample_event_size(sample, type, 0);
-	return perf_event__synthesize_sample(event, type, 0, sample);
+	event->header.type = PERF_RECORD_SAMPLE;
+	event->header.size = perf_event__sample_event_size(sample, type, /*read_format=*/0,
+							   /*branch_sample_type=*/0);
+
+	return perf_event__synthesize_sample(event, type, /*read_format=*/0,
+					     /*branch_sample_type=*/0, sample);
 }
 
 static inline int intel_pt_opt_inject(struct intel_pt *pt,
@@ -2486,7 +2490,7 @@ static int intel_pt_do_synth_pebs_sample(struct intel_pt_queue *ptq, struct evse
 		intel_pt_add_xmm(intr_regs, pos, items, regs_mask);
 	}
 
-	if (sample_type & PERF_SAMPLE_BRANCH_STACK) {
+	if ((sample_type | evsel->synth_sample_type) & PERF_SAMPLE_BRANCH_STACK) {
 		if (items->mask[INTEL_PT_LBR_0_POS] ||
 		    items->mask[INTEL_PT_LBR_1_POS] ||
 		    items->mask[INTEL_PT_LBR_2_POS]) {
@@ -2557,7 +2561,8 @@ static int intel_pt_do_synth_pebs_sample(struct intel_pt_queue *ptq, struct evse
 		sample.transaction = txn;
 	}
 
-	ret = intel_pt_deliver_synth_event(pt, event, &sample, sample_type);
+	ret = intel_pt_deliver_synth_event(pt, event, &sample,
+					   sample_type | evsel->synth_sample_type);
 	perf_sample__exit(&sample);
 	return ret;
 }
diff --git a/tools/perf/util/synthetic-events.c b/tools/perf/util/synthetic-events.c
index 85bee747f4cd..2461f25a4d7d 100644
--- a/tools/perf/util/synthetic-events.c
+++ b/tools/perf/util/synthetic-events.c
@@ -1455,7 +1455,8 @@ int perf_event__synthesize_stat_round(const struct perf_tool *tool,
 	return process(tool, (union perf_event *) &event, NULL, machine);
 }
 
-size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type, u64 read_format)
+size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type, u64 read_format,
+				     u64 branch_sample_type)
 {
 	size_t sz, result = sizeof(struct perf_record_sample);
 
@@ -1515,8 +1516,10 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type,
 
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		sz = sample->branch_stack->nr * sizeof(struct branch_entry);
-		/* nr, hw_idx */
-		sz += 2 * sizeof(u64);
+		/* nr */
+		sz += sizeof(u64);
+		if (branch_sample_type & PERF_SAMPLE_BRANCH_HW_INDEX)
+			sz += sizeof(u64);
 		result += sz;
 	}
 
@@ -1605,7 +1608,7 @@ static __u64 *copy_read_group_values(__u64 *array, __u64 read_format,
 }
 
 int perf_event__synthesize_sample(union perf_event *event, u64 type, u64 read_format,
-				  const struct perf_sample *sample)
+				  u64 branch_sample_type, const struct perf_sample *sample)
 {
 	__u64 *array;
 	size_t sz;
@@ -1719,9 +1722,17 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type, u64 read_fo
 
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		sz = sample->branch_stack->nr * sizeof(struct branch_entry);
-		/* nr, hw_idx */
-		sz += 2 * sizeof(u64);
-		memcpy(array, sample->branch_stack, sz);
+
+		*array++ = sample->branch_stack->nr;
+
+		if (branch_sample_type & PERF_SAMPLE_BRANCH_HW_INDEX) {
+			if (sample->no_hw_idx)
+				*array++ = 0;
+			else
+				*array++ = sample->branch_stack->hw_idx;
+		}
+
+		memcpy(array, perf_sample__branch_entries((struct perf_sample *)sample), sz);
 		array = (void *)array + sz;
 	}
 
diff --git a/tools/perf/util/synthetic-events.h b/tools/perf/util/synthetic-events.h
index b0edad0c3100..8c7f49f9ccf5 100644
--- a/tools/perf/util/synthetic-events.h
+++ b/tools/perf/util/synthetic-events.h
@@ -81,7 +81,8 @@ int perf_event__synthesize_mmap_events(const struct perf_tool *tool, union perf_
 int perf_event__synthesize_modules(const struct perf_tool *tool, perf_event__handler_t process, struct machine *machine);
 int perf_event__synthesize_namespaces(const struct perf_tool *tool, union perf_event *event, pid_t pid, pid_t tgid, perf_event__handler_t process, struct machine *machine);
 int perf_event__synthesize_cgroups(const struct perf_tool *tool, perf_event__handler_t process, struct machine *machine);
-int perf_event__synthesize_sample(union perf_event *event, u64 type, u64 read_format, const struct perf_sample *sample);
+int perf_event__synthesize_sample(union perf_event *event, u64 type, u64 read_format,
+				  u64 branch_sample_type, const struct perf_sample *sample);
 int perf_event__synthesize_stat_config(const struct perf_tool *tool, struct perf_stat_config *config, perf_event__handler_t process, struct machine *machine);
 int perf_event__synthesize_stat_events(struct perf_stat_config *config, const struct perf_tool *tool, struct evlist *evlist, perf_event__handler_t process, bool attrs);
 int perf_event__synthesize_stat_round(const struct perf_tool *tool, u64 time, u64 type, perf_event__handler_t process, struct machine *machine);
@@ -97,7 +98,8 @@ void perf_event__synthesize_final_bpf_metadata(struct perf_session *session,
 
 int perf_tool__process_synth_event(const struct perf_tool *tool, union perf_event *event, struct machine *machine, perf_event__handler_t process);
 
-size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type, u64 read_format);
+size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type,
+				     u64 read_format, u64 branch_sample_type);
 
 int __machine__synthesize_threads(struct machine *machine, const struct perf_tool *tool,
 				  struct target *target, struct perf_thread_map *threads,
-- 
2.54.0.545.g6539524ca2-goog



^ permalink raw reply related

* [PATCH] arm64/entry: Fix arm64-specific rseq brokenness (was: Re: [REGRESSION] rseq: refactoring in v6.19 broke everyone on arm64) and tcmalloc everywhere
From: Mark Rutland @ 2026-04-24 16:45 UTC (permalink / raw)
  To: Mathias Stearn, Linus Torvalds, Catalin Marinas, Will Deacon,
	Thomas Gleixner, Mathieu Desnoyers, Peter Zijlstra
  Cc: Boqun Feng, Paul E. McKenney, Chris Kennelly, Dmitry Vyukov,
	regressions, linux-kernel, linux-arm-kernel, Ingo Molnar,
	Jinjie Ruan, Blake Oler
In-Reply-To: <CAHnCjA25b+nO2n5CeifknSKHssJpPrjnf+dtr7UgzRw4Zgu=oA@mail.gmail.com>

Patch for the arm64-specific issue below. This doesn't fix the generic
cpu_id_start issue, but it brings arm64 into line with everyone else,
and it's the shape we'll need going forwards for other stuff anyway.

I've given it light testing with Mathias's reproducer and the
kselftests, which all pass.

I've also pushed it to my arm64/rseq branch:

  https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/log/?h=arm64/rseq

Mark.

---->8----
From 79b65cbbfa20aa2cb0bc248591fab5459cdc101b Mon Sep 17 00:00:00 2001
From: Mark Rutland <mark.rutland@arm.com>
Date: Thu, 23 Apr 2026 16:51:12 +0100
Subject: [PATCH] arm64/entry: Fix arm64-specific rseq brokenness

Mathias Stearn reports that since v6.19, there are two big issues
affecting rseq:

(1) On arm64 specifically, rseq critical sections aren't aborted when
    they should be.

(2) The 'cpu_id_start' field is no longer written by the kernel in all
    cases it used to be, including some cases where TCMalloc depends on
    the kernel clobbering the field.

This patch fixes issue #1. This patch DOES NOT fix issue #2, which will
need to be addressed by other patches.

The arm64-specific brokenness is a result of commits:

  2fc0e4b4126c ("rseq: Record interrupt from user space")
  39a167560a61 ("rseq: Optimize event setting")

The first commit failed to add a call to rseq_note_user_irq_entry() on
arm64. Thus arm64 never sets rseq_event::user_irq to record that it may
be necessary to abort an active rseq critical section upon return to
userspace. On its own, this commit had no functional impact as the value
of rseq_event::user_irq was not consumed.

The second commit relied upon rseq_event::user_irq to determine whether
or not to bother to perform rseq work when returning to userspace. As
rseq_event::user_irq wasn't set on arm64, this work would be skipped,
and consequently an active rseq critical section would not be aborted.

Fix this by giving arm64 syscall-specific entry/exit paths, and
performing the relevant logic in syscall and non-syscall paths,
including calling rseq_note_user_irq_entry() for non-syscall entry.

Currently arm64 cannot use syscall_enter_from_user_mode(),
syscall_exit_to_user_mode(), and irqentry_exit_to_user_mode(), due to
ordering constraints with exception masking, and risk of ABI breakage
for syscall tracing/audit/etc. For the moment the entry/exit logic is
left as arm64-specific, but mirroring the generic code.

I intend to follow up with refactoring/cleanup, as we did for kernel
mode entry paths in commit:

  041aa7a85390 ("entry: Split preemption from irqentry_exit_to_kernel_mode()")

... which will allow arm64 to use the GENERIC_IRQ_ENTRY functions directly.

Fixes: 39a167560a61 ("rseq: Optimize event setting")
Reported-by: Mathias Stearn <mathias@mongodb.com>
Link: https://lore.kernel.org/regressions/CAHnCjA25b+nO2n5CeifknSKHssJpPrjnf+dtr7UgzRw4Zgu=oA@mail.gmail.com/
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Chris Kennelly <ckennelly@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will@kernel.org>
---
 arch/arm64/kernel/entry-common.c | 29 ++++++++++++++++++++++-------
 include/linux/irq-entry-common.h |  8 --------
 include/linux/rseq_entry.h       | 19 -------------------
 3 files changed, 22 insertions(+), 34 deletions(-)

diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c
index cb54335465f66..65ade1f1544f6 100644
--- a/arch/arm64/kernel/entry-common.c
+++ b/arch/arm64/kernel/entry-common.c
@@ -62,6 +62,12 @@ static void noinstr arm64_exit_to_kernel_mode(struct pt_regs *regs,
 	irqentry_exit_to_kernel_mode_after_preempt(regs, state);
 }
 
+static __always_inline void arm64_syscall_enter_from_user_mode(struct pt_regs *regs)
+{
+	enter_from_user_mode(regs);
+	mte_disable_tco_entry(current);
+}
+
 /*
  * Handle IRQ/context state management when entering from user mode.
  * Before this function is called it is not safe to call regular kernel code,
@@ -70,20 +76,29 @@ static void noinstr arm64_exit_to_kernel_mode(struct pt_regs *regs,
 static __always_inline void arm64_enter_from_user_mode(struct pt_regs *regs)
 {
 	enter_from_user_mode(regs);
+	rseq_note_user_irq_entry();
 	mte_disable_tco_entry(current);
 	sme_enter_from_user_mode();
 }
 
+static __always_inline void arm64_syscall_exit_to_user_mode(struct pt_regs *regs)
+{
+	local_irq_disable();
+	syscall_exit_to_user_mode_prepare(regs);
+	local_daif_mask();
+	mte_check_tfsr_exit();
+	exit_to_user_mode();
+}
+
 /*
  * Handle IRQ/context state management when exiting to user mode.
  * After this function returns it is not safe to call regular kernel code,
  * instrumentable code, or any code which may trigger an exception.
  */
-
 static __always_inline void arm64_exit_to_user_mode(struct pt_regs *regs)
 {
 	local_irq_disable();
-	exit_to_user_mode_prepare_legacy(regs);
+	irqentry_exit_to_user_mode_prepare(regs);
 	local_daif_mask();
 	sme_exit_to_user_mode();
 	mte_check_tfsr_exit();
@@ -92,7 +107,7 @@ static __always_inline void arm64_exit_to_user_mode(struct pt_regs *regs)
 
 asmlinkage void noinstr asm_exit_to_user_mode(struct pt_regs *regs)
 {
-	arm64_exit_to_user_mode(regs);
+	arm64_syscall_exit_to_user_mode(regs);
 }
 
 /*
@@ -716,12 +731,12 @@ static void noinstr el0_brk64(struct pt_regs *regs, unsigned long esr)
 
 static void noinstr el0_svc(struct pt_regs *regs)
 {
-	arm64_enter_from_user_mode(regs);
+	arm64_syscall_enter_from_user_mode(regs);
 	cortex_a76_erratum_1463225_svc_handler();
 	fpsimd_syscall_enter();
 	local_daif_restore(DAIF_PROCCTX);
 	do_el0_svc(regs);
-	arm64_exit_to_user_mode(regs);
+	arm64_syscall_exit_to_user_mode(regs);
 	fpsimd_syscall_exit();
 }
 
@@ -868,11 +883,11 @@ static void noinstr el0_cp15(struct pt_regs *regs, unsigned long esr)
 
 static void noinstr el0_svc_compat(struct pt_regs *regs)
 {
-	arm64_enter_from_user_mode(regs);
+	arm64_syscall_enter_from_user_mode(regs);
 	cortex_a76_erratum_1463225_svc_handler();
 	local_daif_restore(DAIF_PROCCTX);
 	do_el0_svc_compat(regs);
-	arm64_exit_to_user_mode(regs);
+	arm64_syscall_exit_to_user_mode(regs);
 }
 
 static void noinstr el0_bkpt32(struct pt_regs *regs, unsigned long esr)
diff --git a/include/linux/irq-entry-common.h b/include/linux/irq-entry-common.h
index 167fba7dbf043..1fabf0f5ea8e7 100644
--- a/include/linux/irq-entry-common.h
+++ b/include/linux/irq-entry-common.h
@@ -218,14 +218,6 @@ static __always_inline void __exit_to_user_mode_validate(void)
 	lockdep_sys_exit();
 }
 
-/* Temporary workaround to keep ARM64 alive */
-static __always_inline void exit_to_user_mode_prepare_legacy(struct pt_regs *regs)
-{
-	__exit_to_user_mode_prepare(regs, EXIT_TO_USER_MODE_WORK);
-	rseq_exit_to_user_mode_legacy();
-	__exit_to_user_mode_validate();
-}
-
 /**
  * syscall_exit_to_user_mode_prepare - call exit_to_user_mode_loop() if required
  * @regs:	Pointer to pt_regs on entry stack
diff --git a/include/linux/rseq_entry.h b/include/linux/rseq_entry.h
index f11ebd34f8b95..a3762410c4ab6 100644
--- a/include/linux/rseq_entry.h
+++ b/include/linux/rseq_entry.h
@@ -753,24 +753,6 @@ static __always_inline void rseq_irqentry_exit_to_user_mode(void)
 	ev->events = 0;
 }
 
-/* Required to keep ARM64 working */
-static __always_inline void rseq_exit_to_user_mode_legacy(void)
-{
-	struct rseq_event *ev = &current->rseq.event;
-
-	rseq_stat_inc(rseq_stats.exit);
-
-	if (static_branch_unlikely(&rseq_debug_enabled))
-		WARN_ON_ONCE(ev->sched_switch);
-
-	/*
-	 * Ensure that event (especially user_irq) is cleared when the
-	 * interrupt did not result in a schedule and therefore the
-	 * rseq processing did not clear it.
-	 */
-	ev->events = 0;
-}
-
 void __rseq_debug_syscall_return(struct pt_regs *regs);
 
 static __always_inline void rseq_debug_syscall_return(struct pt_regs *regs)
@@ -786,7 +768,6 @@ static inline bool rseq_exit_to_user_mode_restart(struct pt_regs *regs, unsigned
 }
 static inline void rseq_syscall_exit_to_user_mode(void) { }
 static inline void rseq_irqentry_exit_to_user_mode(void) { }
-static inline void rseq_exit_to_user_mode_legacy(void) { }
 static inline void rseq_debug_syscall_return(struct pt_regs *regs) { }
 static inline bool rseq_grant_slice_extension(unsigned long ti_work, unsigned long mask) { return false; }
 #endif /* !CONFIG_RSEQ */
-- 
2.30.2



^ permalink raw reply related

* Re: [PATCH v2] kselftest/arm64: Fix build failure with GCC-15
From: Catalin Marinas @ 2026-04-24 16:44 UTC (permalink / raw)
  To: Mark Brown
  Cc: Leo Yan, Will Deacon, Shuah Khan, Thiago Jung Bauermann,
	linux-arm-kernel, linux-kselftest, linux-kernel
In-Reply-To: <63409143-f4eb-48c3-89de-1aef4fb57381@sirena.org.uk>

On Fri, Apr 24, 2026 at 05:07:42PM +0100, Mark Brown wrote:
> On Fri, Apr 24, 2026 at 04:51:24PM +0100, Catalin Marinas wrote:
> > On Wed, Apr 22, 2026 at 06:42:54PM +0100, Leo Yan wrote:
> 
> > > Building on Debian sid with GCC 15 fails:
> 
> > I think a better fix is to always define struct user_gcs and only
> > conditionally define NT_ARM_GCS (IOW, move the #endif higher).
> 
> I've not actually double checked that everything is wired up properly
> but I believe these days this should actually pick up asm/ptrace.h from
> the headers_install target so a current kernel copy.  We ought to be
> able to remove the local definition of struct user_gcs I think, there's
> still some weirdness with the NT_ definitions I can't remember but the
> struct should be fine.

OK, so it does look like it picks the kernel uapi/asm/ptrace.h. It
builds fine on Debian stable (no GCS anywhere) with including
asm/ptrace.h and removing struct user_gcs.

But I think we should include asm/ptrace.h in libc-gcs.h and not the
gcs-util.h header (for NT_ARM_GCS it's fine to keep in gcs-util.h).

-- 
Catalin


^ permalink raw reply

* Re: [PATCH] iommu/arm-smmu-v3: Allow disabling Stage 1 translation
From: Jason Gunthorpe @ 2026-04-24 16:39 UTC (permalink / raw)
  To: Will Deacon
  Cc: Evangelos Petrongonas, Robin Murphy, Joerg Roedel, Nicolin Chen,
	Pranjal Shrivastava, Lu Baolu, linux-arm-kernel, iommu,
	linux-kernel, nh-open-source, Zeev Zilberman
In-Reply-To: <aeuT1-TB6dOT5ZQ2@willie-the-truck>

On Fri, Apr 24, 2026 at 05:01:27PM +0100, Will Deacon wrote:
> On Fri, Apr 24, 2026 at 12:42:56PM -0300, Jason Gunthorpe wrote:
> > On Fri, Apr 24, 2026 at 04:16:17PM +0100, Will Deacon wrote:
> > > > > > STE/CD is pretty simple now, there is only one place to put the CMO
> > > > > > and the ordering is all handled with that shared code. We no longer
> > > > > > care about ordering beyond all the writes must be visible to HW before
> > > > > > issuing the CMDQ invalidation command - which is the same environment
> > > > > > as the pagetable.
> > > > > 
> > > > > You presumably rely on 64-bit single-copy atomicity for hitless updates,
> > > > > no?
> > > > 
> > > > Yes, just like the page table does..
> > > > 
> > > > I hope that's not a problem or we have a issue with the PTW :)
> > > 
> > > You trimmed the part from my reply where I think we _do_ have an issue
> > > with the PTW. Here it is again:
> > > 
> > >   The non-coherent case looks more fragile, because I don't _think_ the
> > >   architecture provides any ordering or atomicity guarantees about cache
> > >   cleaning to the PoC. Presumably, the correct sequence would be to write
> > >   the PTE with the valid bit clear, do the CMO (with completion barrier),
> > >   *then* write the bottom byte with the valid bit set and do another CMO.
> > 
> > I wasn't sure if you are being serious.
> > 
> > CMO + barriers must provide an ordering guarentee about cache cleaning
> > to POC otherwise the entire Linux DMA API is broken. dma_sync must
> > order with following device DMA. IMHO that's not negotiable for Linux.
> 
> The problem is with concurrent DMA (from the page-table walker) and I
> don't see anything that guarantees that in the CPU architecture. I don't
> think the streaming DMA API pretends to handle that case, does it? It
> relies on a pretty rigid ownership concept from what I understand.

I think you pointed out two things, ordering and tearing.

Ordering is OK. If I write a PTE, dma_sync, then command a device to
use that IOVA the PTW must observe the new PTE value. Otherwise
dma_sync isn't doing what Linux requires.

Tearing is a different issue, if the device uses the IOVA and races
with the PTE write changing it then you say maybe it can mis-read it
with tearing.

However, this race only happens if the PTE is currently non-valid or
being changed to non-valid. Meaning randomly you will be getting an
invalid IOVA event.

In non-coherent mode we don't allow SVA and we don't allow VFIO. Only
the DMA API and drivers open coding things.

For VFIO and SVA, yes, we need the HW to work and properly, userspcae
can trigger invalid IOVA, we can't tolerate a corrupted PTE.

In embedded I suppose you could make an argument you don't care about
it since invalid IOVA would have to be caused by a buggy kernel
driver, it should never happen, and thus this is really a debug
feature. If the race will never be hit in a working system maybe it is
fine to leave it as is.

Would be good to document this detail :)

> Of course I'd rather that the architecture said that our current code
> is fine, but if it doesn't then I don't have much choice, really. At the
> very least, we should minimise the number of places where we rely on
> non-architected behaviour and so keeping the CDs and STEs non-cacheable
> remains my preference.

So, I am convinced, PTW has that escape above that doesn't apply to
STE/CD. Those can be accessed truely at any time and we can't ever
leave a 64 bit value in a strange state.

Jason


^ permalink raw reply

* Re: [PATCH] clk: bcm: rpi: Mark VEC clock as CLK_IGNORE_UNUSED
From: Stefan Wahren @ 2026-04-24 16:32 UTC (permalink / raw)
  To: Maíra Canal, Michael Turquette, Stephen Boyd,
	Florian Fainelli, Broadcom internal kernel review list,
	Mark Brown, Maxime Ripard, Dom Cobley, Dave Stevenson
  Cc: linux-clk, linux-rpi-kernel, linux-arm-kernel, kernel-dev
In-Reply-To: <20260401111416.562279-2-mcanal@igalia.com>

Am 01.04.26 um 13:13 schrieb Maíra Canal:
> On Raspberry Pi 3B, the VEC clock is used by the VideoCore firmware
> display driver, which remains active until the vc4 driver loads and
> sends NOTIFY_DISPLAY_DONE. If this clock is disabled during boot, a bus
> lockup happens and the firmware becomes unresponsive, causing a complete
> system lockup.
>
> Mark the VEC clock with CLK_IGNORE_UNUSED so it survives the unused
> clock disablement and remains available until the vc4 driver takes over
> display management.
>
> Fixes: 672299736af6 ("clk: bcm: rpi: Manage clock rate in prepare/unprepare callbacks")
> Reported-by: Mark Brown <broonie@kernel.org>
> Signed-off-by: Maíra Canal <mcanal@igalia.com>

FWIW:

Reviewed-by: Stefan Wahren <wahrenst@gmx.net>



^ permalink raw reply

* Re: [PATCH 3/8] firmware: sysfb: Make CONFIG_SYSFB a user-selectable option
From: Javier Martinez Canillas @ 2026-04-24 16:24 UTC (permalink / raw)
  To: Thomas Zimmermann, Arnd Bergmann, Ard Biesheuvel,
	Ilias Apalodimas, Huacai Chen, WANG Xuerui, Maarten Lankhorst,
	Maxime Ripard, Dave Airlie, Simona Vetter, K. Y. Srinivasan,
	Haiyang Zhang, Wei Liu, Dexuan Cui, longli, Helge Deller
  Cc: linux-arm-kernel, loongarch, linux-efi, linux-riscv, dri-devel,
	linux-hyperv, linux-fbdev
In-Reply-To: <0156562f-5fcf-47ce-8fea-03345f2c3fe6@suse.de>

Thomas Zimmermann <tzimmermann@suse.de> writes:

Hello,

[...]

>>>>>> On Thu, Apr 2, 2026, at 11:09, Thomas Zimmermann wrote:
>>>>>> I don't really like this part of the series and would prefer
>>>>>> to keep CONFIG_SYSFB hidden as much as possible as an x86

I tend to agree with Arnd here, I'm also not seeing that much value on
making this symbol user selectable. For now I would just keep it hidden.

[...]

>> Yes, I saw that as well and don't have an immediate idea for how
>> to best do it. I saw that you already abstracted the access to
>> the screen_info members in drm_sysfb_screen_info.c, which I think
>> is a step in that direction.
>>
>> I also noticed that efidrm is mostly a subset of vesadrm, so
>> in theory they could be merged back into an x86 drm driver
>> along with the drm_sysfb_screen_info helpers, and have a non-x86
>> driver that constructs a drm_sysfb_device directly from the
>> EFI structures.
>
> I would not want to have a unifed driver for all-things-screen_info. The 
> code that can easily be shared is already in the sysfb helpers. But I 
> don't mind adding a separate driver for EFI's Graphics Output Protocol.

I agree. It is much more maintainable if we have dedicated DRM drivers that
use shared helpers, than attempting to have a driver for different platforms.

As Thomas explained, the maintance effort is small on the DRM side and he has
done a lot of work to split simpledrm in efidrm, vesadrm and ofdrm.

-- 
Best regards,

Javier Martinez Canillas
Core Platforms
Red Hat



^ permalink raw reply

* Re: [PATCH 15/15] arm64: dts: ti: beagley-ai: Enable HDMI display and audio
From: Robert Nelson @ 2026-04-24 16:16 UTC (permalink / raw)
  To: Tomi Valkeinen
  Cc: Maarten Lankhorst, Maxime Ripard, Thomas Zimmermann, David Airlie,
	Simona Vetter, Rob Herring, Krzysztof Kozlowski, Conor Dooley,
	Lee Jones, Aradhya Bhatia, Nishanth Menon, Vignesh Raghavendra,
	Swamil Jain, Devarsh Thakkar, Louis Chauvet, devicetree,
	dri-devel, linux-kernel, linux-arm-kernel, Andrew Davis
In-Reply-To: <20260420-beagley-ai-display-v1-15-f628543dfd14@ideasonboard.com>

On Mon, Apr 20, 2026 at 8:04 AM Tomi Valkeinen
<tomi.valkeinen@ideasonboard.com> wrote:
>
> From: Andrew Davis <afd@ti.com>
>
> Enable HDMI support for BeagleY-AI platform. The display controller used is
> TIDSS and the HDMI bridge used is IT66122.
>
> Based on DT by: Robert Nelson <robertcnelson@gmail.com>
> Signed-off-by: Andrew Davis <afd@ti.com>
> Signed-off-by: Swamil Jain <s-jain1@ti.com>
> [tomi.valkeinen: cosmetic fixes]
> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>

Tested-by: Robert Nelson <robertcnelson@gmail.com>

Thank you for getting the display back end working on j722s family!

After this goes in, I need to submit the usb/usb-hub changes to enable
USB support...

Regards,

-- 
Robert Nelson
https://rcn-ee.com/


^ permalink raw reply


This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox