perf machine: Protect the machine->threads with a rwlock

In addition to using refcounts for the struct thread lifetime
management, we need to protect access to machine->threads from
concurrent access.

That happens in 'perf top', where a thread processes events, inserting
and deleting entries from that rb_tree while another thread decays
hist_entries, that end up dropping references and ultimately deleting
threads from the rb_tree and releasing its resources when no further
hist_entry (or other data structures, like in 'perf sched') references
it.

So the rule is the same for refcounts + protected trees in the kernel,
get the tree lock, find object, bump the refcount, drop the tree lock,
return, use object, drop the refcount if no more use of it is needed,
keep it if storing it in some other data structure, drop when releasing
that data structure.

I.e. pair "t = machine__find(new)_thread()" with a "thread__put(t)", and
"perf_event__preprocess_sample(&al)" with "addr_location__put(&al)".

The addr_location__put() one is because as we return references to
several data structures, we may end up adding more reference counting
for the other data structures and then we'll drop it at
addr_location__put() time.

Acked-by: David Ahern <dsahern@gmail.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-bs9rt4n0jw3hi9f3zxyy3xln@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

29 files changed, 284 insertions, 114 deletions

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 71bf7451c0ca..b57a027fb200 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -84,6 +84,7 @@ static int process_sample_event(struct perf_tool *tool,
 {
 	struct perf_annotate *ann = container_of(tool, struct perf_annotate, tool);
 	struct addr_location al;
+	int ret = 0;
 
 	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
 		pr_warning("problem processing %d event, skipping it.\n",
@@ -92,15 +93,16 @@ static int process_sample_event(struct perf_tool *tool,
 	}
 
 	if (ann->cpu_list && !test_bit(sample->cpu, ann->cpu_bitmap))
-		return 0;
+		goto out_put;
 
 	if (!al.filtered && perf_evsel__add_sample(evsel, sample, &al, ann)) {
 		pr_warning("problem incrementing symbol count, "
 			   "skipping event\n");
-		return -1;
+		ret = -1;
 	}
-
-	return 0;
+out_put:
+	addr_location__put(&al);
+	return ret;
 }
 
 static int hist_entry__tty_annotate(struct hist_entry *he,
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index df6307b4050a..daaa7dca9c3b 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -328,6 +328,7 @@ static int diff__process_sample_event(struct perf_tool *tool __maybe_unused,
 {
 	struct addr_location al;
 	struct hists *hists = evsel__hists(evsel);
+	int ret = -1;
 
 	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
 		pr_warning("problem processing %d event, skipping it.\n",
@@ -338,7 +339,7 @@ static int diff__process_sample_event(struct perf_tool *tool __maybe_unused,
 	if (hists__add_entry(hists, &al, sample->period,
 			     sample->weight, sample->transaction)) {
 		pr_warning("problem incrementing symbol period, skipping event\n");
-		return -1;
+		goto out_put;
 	}
 
 	/*
@@ -350,8 +351,10 @@ static int diff__process_sample_event(struct perf_tool *tool __maybe_unused,
 	hists->stats.total_period += sample->period;
 	if (!al.filtered)
 		hists->stats.total_non_filtered_period += sample->period;
-
-	return 0;
+	ret = 0;
+out_put:
+	addr_location__put(&al);
+	return ret;
 }
 
 static struct perf_tool tool = {
diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
index d6a47e854b2b..52ec66b23607 100644
--- a/tools/perf/builtin-inject.c
+++ b/tools/perf/builtin-inject.c
@@ -365,6 +365,7 @@ static int perf_event__inject_buildid(struct perf_tool *tool,
 		}
 	}
 
+	thread__put(thread);
 repipe:
 	perf_event__repipe(tool, event, sample, machine);
 	return 0;
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index e628bf1a0c24..fe3fcb75f20a 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -906,6 +906,7 @@ static int process_sample_event(struct perf_tool *tool __maybe_unused,
 				struct perf_evsel *evsel,
 				struct machine *machine)
 {
+	int err = 0;
 	struct thread *thread = machine__findnew_thread(machine, sample->pid,
 							sample->tid);
 
@@ -919,10 +920,12 @@ static int process_sample_event(struct perf_tool *tool __maybe_unused,
 
 	if (evsel->handler != NULL) {
 		tracepoint_handler f = evsel->handler;
-		return f(evsel, sample);
+		err = f(evsel, sample);
 	}
 
-	return 0;
+	thread__put(thread);
+
+	return err;
 }
 
 static struct perf_tool perf_kmem = {
diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c
index 1f9338f6109c..15fecd3dc5d8 100644
--- a/tools/perf/builtin-kvm.c
+++ b/tools/perf/builtin-kvm.c
@@ -651,6 +651,7 @@ static int process_sample_event(struct perf_tool *tool,
 				struct perf_evsel *evsel,
 				struct machine *machine)
 {
+	int err = 0;
 	struct thread *thread;
 	struct perf_kvm_stat *kvm = container_of(tool, struct perf_kvm_stat,
 						 tool);
@@ -666,9 +667,10 @@ static int process_sample_event(struct perf_tool *tool,
 	}
 
 	if (!handle_kvm_event(kvm, thread, evsel, sample))
-		return -1;
+		err = -1;
 
-	return 0;
+	thread__put(thread);
+	return err;
 }
 
 static int cpu_isa_config(struct perf_kvm_stat *kvm)
diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c
index d49c2ab85fc2..de16aaed516e 100644
--- a/tools/perf/builtin-lock.c
+++ b/tools/perf/builtin-lock.c
@@ -769,6 +769,7 @@ static void dump_threads(void)
 		t = perf_session__findnew(session, st->tid);
 		pr_info("%10d: %s\n", st->tid, thread__comm_str(t));
 		node = rb_next(node);
+		thread__put(t);
 	};
 }
 
@@ -810,6 +811,7 @@ static int process_sample_event(struct perf_tool *tool __maybe_unused,
 				struct perf_evsel *evsel,
 				struct machine *machine)
 {
+	int err = 0;
 	struct thread *thread = machine__findnew_thread(machine, sample->pid,
 							sample->tid);
 
@@ -821,10 +823,12 @@ static int process_sample_event(struct perf_tool *tool __maybe_unused,
 
 	if (evsel->handler != NULL) {
 		tracepoint_handler f = evsel->handler;
-		return f(evsel, sample);
+		err = f(evsel, sample);
 	}
 
-	return 0;
+	thread__put(thread);
+
+	return err;
 }
 
 static void sort_result(void)
diff --git a/tools/perf/builtin-mem.c b/tools/perf/builtin-mem.c
index 675216e08bfc..da2ec06f0742 100644
--- a/tools/perf/builtin-mem.c
+++ b/tools/perf/builtin-mem.c
@@ -74,7 +74,7 @@ dump_raw_samples(struct perf_tool *tool,
 	}
 
 	if (al.filtered || (mem->hide_unresolved && al.sym == NULL))
-		return 0;
+		goto out_put;
 
 	if (al.map != NULL)
 		al.map->dso->hit = 1;
@@ -103,7 +103,8 @@ dump_raw_samples(struct perf_tool *tool,
 		symbol_conf.field_sep,
 		al.map ? (al.map->dso ? al.map->dso->long_name : "???") : "???",
 		al.sym ? al.sym->name : "???");
-
+out_put:
+	addr_location__put(&al);
 	return 0;
 }
 
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 18cb0ff39b4e..8d5118f5cedd 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -142,7 +142,7 @@ static int process_sample_event(struct perf_tool *tool,
 		.hide_unresolved = rep->hide_unresolved,
 		.add_entry_cb = hist_iter__report_callback,
 	};
-	int ret;
+	int ret = 0;
 
 	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
 		pr_debug("problem processing %d event, skipping it.\n",
@@ -151,10 +151,10 @@ static int process_sample_event(struct perf_tool *tool,
 	}
 
 	if (rep->hide_unresolved && al.sym == NULL)
-		return 0;
+		goto out_put;
 
 	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
-		return 0;
+		goto out_put;
 
 	if (sort__mode == SORT_MODE__BRANCH)
 		iter.ops = &hist_iter_branch;
@@ -172,7 +172,8 @@ static int process_sample_event(struct perf_tool *tool,
 				   rep);
 	if (ret < 0)
 		pr_debug("problem adding hist entry, skipping event\n");
-
+out_put:
+	addr_location__put(&al);
 	return ret;
 }
 
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 5275bab70313..79273ecf92eb 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -770,7 +770,7 @@ static int replay_fork_event(struct perf_sched *sched,
 	if (child == NULL || parent == NULL) {
 		pr_debug("thread does not exist on fork event: child %p, parent %p\n",
 				 child, parent);
-		return 0;
+		goto out_put;
 	}
 
 	if (verbose) {
@@ -781,6 +781,9 @@ static int replay_fork_event(struct perf_sched *sched,
 
 	register_pid(sched, parent->tid, thread__comm_str(parent));
 	register_pid(sched, child->tid, thread__comm_str(child));
+out_put:
+	thread__put(child);
+	thread__put(parent);
 	return 0;
 }
 
@@ -957,7 +960,7 @@ static int latency_switch_event(struct perf_sched *sched,
 	struct work_atoms *out_events, *in_events;
 	struct thread *sched_out, *sched_in;
 	u64 timestamp0, timestamp = sample->time;
-	int cpu = sample->cpu;
+	int cpu = sample->cpu, err = -1;
 	s64 delta;
 
 	BUG_ON(cpu >= MAX_CPUS || cpu < 0);
@@ -976,15 +979,17 @@ static int latency_switch_event(struct perf_sched *sched,
 
 	sched_out = machine__findnew_thread(machine, -1, prev_pid);
 	sched_in = machine__findnew_thread(machine, -1, next_pid);
+	if (sched_out == NULL || sched_in == NULL)
+		goto out_put;
 
 	out_events = thread_atoms_search(&sched->atom_root, sched_out, &sched->cmp_pid);
 	if (!out_events) {
 		if (thread_atoms_insert(sched, sched_out))
-			return -1;
+			goto out_put;
 		out_events = thread_atoms_search(&sched->atom_root, sched_out, &sched->cmp_pid);
 		if (!out_events) {
 			pr_err("out-event: Internal tree error");
-			return -1;
+			goto out_put;
 		}
 	}
 	if (add_sched_out_event(out_events, sched_out_state(prev_state), timestamp))
@@ -993,22 +998,25 @@ static int latency_switch_event(struct perf_sched *sched,
 	in_events = thread_atoms_search(&sched->atom_root, sched_in, &sched->cmp_pid);
 	if (!in_events) {
 		if (thread_atoms_insert(sched, sched_in))
-			return -1;
+			goto out_put;
 		in_events = thread_atoms_search(&sched->atom_root, sched_in, &sched->cmp_pid);
 		if (!in_events) {
 			pr_err("in-event: Internal tree error");
-			return -1;
+			goto out_put;
 		}
 		/*
 		 * Take came in we have not heard about yet,
 		 * add in an initial atom in runnable state:
 		 */
 		if (add_sched_out_event(in_events, 'R', timestamp))
-			return -1;
+			goto out_put;
 	}
 	add_sched_in_event(in_events, timestamp);
-
-	return 0;
+	err = 0;
+out_put:
+	thread__put(sched_out);
+	thread__put(sched_in);
+	return err;
 }
 
 static int latency_runtime_event(struct perf_sched *sched,
@@ -1021,23 +1029,29 @@ static int latency_runtime_event(struct perf_sched *sched,
 	struct thread *thread = machine__findnew_thread(machine, -1, pid);
 	struct work_atoms *atoms = thread_atoms_search(&sched->atom_root, thread, &sched->cmp_pid);
 	u64 timestamp = sample->time;
-	int cpu = sample->cpu;
+	int cpu = sample->cpu, err = -1;
+
+	if (thread == NULL)
+		return -1;
 
 	BUG_ON(cpu >= MAX_CPUS || cpu < 0);
 	if (!atoms) {
 		if (thread_atoms_insert(sched, thread))
-			return -1;
+			goto out_put;
 		atoms = thread_atoms_search(&sched->atom_root, thread, &sched->cmp_pid);
 		if (!atoms) {
 			pr_err("in-event: Internal tree error");
-			return -1;
+			goto out_put;
 		}
 		if (add_sched_out_event(atoms, 'R', timestamp))
-			return -1;
+			goto out_put;
 	}
 
 	add_runtime_event(atoms, runtime, timestamp);
-	return 0;
+	err = 0;
+out_put:
+	thread__put(thread);
+	return err;
 }
 
 static int latency_wakeup_event(struct perf_sched *sched,
@@ -1050,19 +1064,22 @@ static int latency_wakeup_event(struct perf_sched *sched,
 	struct work_atom *atom;
 	struct thread *wakee;
 	u64 timestamp = sample->time;
+	int err = -1;
 
 	wakee = machine__findnew_thread(machine, -1, pid);
+	if (wakee == NULL)
+		return -1;
 	atoms = thread_atoms_search(&sched->atom_root, wakee, &sched->cmp_pid);
 	if (!atoms) {
 		if (thread_atoms_insert(sched, wakee))
-			return -1;
+			goto out_put;
 		atoms = thread_atoms_search(&sched->atom_root, wakee, &sched->cmp_pid);
 		if (!atoms) {
 			pr_err("wakeup-event: Internal tree error");
-			return -1;
+			goto out_put;
 		}
 		if (add_sched_out_event(atoms, 'S', timestamp))
-			return -1;
+			goto out_put;
 	}
 
 	BUG_ON(list_empty(&atoms->work_list));
@@ -1081,17 +1098,21 @@ static int latency_wakeup_event(struct perf_sched *sched,
 	 * skip in this case.
 	 */
 	if (sched->profile_cpu == -1 && atom->state != THREAD_SLEEPING)
-		return 0;
+		goto out_ok;
 
 	sched->nr_timestamps++;
 	if (atom->sched_out_time > timestamp) {
 		sched->nr_unordered_timestamps++;
-		return 0;
+		goto out_ok;
 	}
 
 	atom->state = THREAD_WAIT_CPU;
 	atom->wake_up_time = timestamp;
-	return 0;
+out_ok:
+	err = 0;
+out_put:
+	thread__put(wakee);
+	return err;
 }
 
 static int latency_migrate_task_event(struct perf_sched *sched,
@@ -1104,6 +1125,7 @@ static int latency_migrate_task_event(struct perf_sched *sched,
 	struct work_atoms *atoms;
 	struct work_atom *atom;
 	struct thread *migrant;
+	int err = -1;
 
 	/*
 	 * Only need to worry about migration when profiling one CPU.
@@ -1112,18 +1134,20 @@ static int latency_migrate_task_event(struct perf_sched *sched,
 		return 0;
 
 	migrant = machine__findnew_thread(machine, -1, pid);
+	if (migrant == NULL)
+		return -1;
 	atoms = thread_atoms_search(&sched->atom_root, migrant, &sched->cmp_pid);
 	if (!atoms) {
 		if (thread_atoms_insert(sched, migrant))
-			return -1;
+			goto out_put;
 		register_pid(sched, migrant->tid, thread__comm_str(migrant));
 		atoms = thread_atoms_search(&sched->atom_root, migrant, &sched->cmp_pid);
 		if (!atoms) {
 			pr_err("migration-event: Internal tree error");
-			return -1;
+			goto out_put;
 		}
 		if (add_sched_out_event(atoms, 'R', timestamp))
-			return -1;
+			goto out_put;
 	}
 
 	BUG_ON(list_empty(&atoms->work_list));
@@ -1135,8 +1159,10 @@ static int latency_migrate_task_event(struct perf_sched *sched,
 
 	if (atom->sched_out_time > timestamp)
 		sched->nr_unordered_timestamps++;
-
-	return 0;
+	err = 0;
+out_put:
+	thread__put(migrant);
+	return err;
 }
 
 static void output_lat_thread(struct perf_sched *sched, struct work_atoms *work_list)
@@ -1330,8 +1356,10 @@ static int map_switch_event(struct perf_sched *sched, struct perf_evsel *evsel,
 	}
 
 	sched_in = machine__findnew_thread(machine, -1, next_pid);
+	if (sched_in == NULL)
+		return -1;
 
-	sched->curr_thread[this_cpu] = sched_in;
+	sched->curr_thread[this_cpu] = thread__get(sched_in);
 
 	printf("  ");
 
@@ -1381,6 +1409,8 @@ static int map_switch_event(struct perf_sched *sched, struct perf_evsel *evsel,
 		printf("\n");
 	}
 
+	thread__put(sched_in);
+
 	return 0;
 }
 
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 6805098e3751..24809787369f 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -607,13 +607,14 @@ static int process_sample_event(struct perf_tool *tool __maybe_unused,
 	}
 
 	if (al.filtered)
-		return 0;
+		goto out_put;
 
 	if (cpu_list && !test_bit(sample->cpu, cpu_bitmap))
-		return 0;
+		goto out_put;
 
 	scripting_ops->process_event(event, sample, evsel, &al);
-
+out_put:
+	addr_location__put(&al);
 	return 0;
 }
 
@@ -681,8 +682,8 @@ static int process_comm_event(struct perf_tool *tool,
 	print_sample_start(sample, thread, evsel);
 	perf_event__fprintf(event, stdout);
 	ret = 0;
-
 out:
+	thread__put(thread);
 	return ret;
 }
 
@@ -713,6 +714,7 @@ static int process_fork_event(struct perf_tool *tool,
 	}
 	print_sample_start(sample, thread, evsel);
 	perf_event__fprintf(event, stdout);
+	thread__put(thread);
 
 	return 0;
 }
@@ -721,6 +723,7 @@ static int process_exit_event(struct perf_tool *tool,
 			      struct perf_sample *sample,
 			      struct machine *machine)
 {
+	int err = 0;
 	struct thread *thread;
 	struct perf_script *script = container_of(tool, struct perf_script, tool);
 	struct perf_session *session = script->session;
@@ -742,9 +745,10 @@ static int process_exit_event(struct perf_tool *tool,
 	perf_event__fprintf(event, stdout);
 
 	if (perf_event__process_exit(tool, event, sample, machine) < 0)
-		return -1;
+		err = -1;
 
-	return 0;
+	thread__put(thread);
+	return err;
 }
 
 static int process_mmap_event(struct perf_tool *tool,
@@ -774,7 +778,7 @@ static int process_mmap_event(struct perf_tool *tool,
 	}
 	print_sample_start(sample, thread, evsel);
 	perf_event__fprintf(event, stdout);
-
+	thread__put(thread);
 	return 0;
 }
 
@@ -805,7 +809,7 @@ static int process_mmap2_event(struct perf_tool *tool,
 	}
 	print_sample_start(sample, thread, evsel);
 	perf_event__fprintf(event, stdout);
-
+	thread__put(thread);
 	return 0;
 }
 
diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c
index e50fe1187b0b..3b884e37ab8b 100644
--- a/tools/perf/builtin-timechart.c
+++ b/tools/perf/builtin-timechart.c
@@ -523,7 +523,7 @@ static const char *cat_backtrace(union perf_event *event,
 				 * Discard all.
 				 */
 				zfree(&p);
-				goto exit;
+				goto exit_put;
 			}
 			continue;
 		}
@@ -538,7 +538,8 @@ static const char *cat_backtrace(union perf_event *event,
 		else
 			fprintf(f, "..... %016" PRIx64 "\n", ip);
 	}
-
+exit_put:
+	addr_location__put(&al);
 exit:
 	fclose(f);
 
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 1cb3436276d1..232658337f18 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -793,7 +793,7 @@ static void perf_event__process_sample(struct perf_tool *tool,
 		pthread_mutex_unlock(&hists->lock);
 	}
 
-	return;
+	addr_location__put(&al);
 }
 
 static void perf_top__mmap_read_idx(struct perf_top *top, int idx)
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index d1139b6bd534..bb05e44f8f0c 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -1712,7 +1712,7 @@ static int trace__sys_enter(struct trace *trace, struct perf_evsel *evsel,
 	void *args;
 	size_t printed = 0;
 	struct thread *thread;
-	int id = perf_evsel__sc_tp_uint(evsel, id, sample);
+	int id = perf_evsel__sc_tp_uint(evsel, id, sample), err = -1;
 	struct syscall *sc = trace__syscall_info(trace, evsel, id);
 	struct thread_trace *ttrace;
 
@@ -1725,14 +1725,14 @@ static int trace__sys_enter(struct trace *trace, struct perf_evsel *evsel,
 	thread = machine__findnew_thread(trace->host, sample->pid, sample->tid);
 	ttrace = thread__trace(thread, trace->output);
 	if (ttrace == NULL)
-		return -1;
+		goto out_put;
 
 	args = perf_evsel__sc_tp_ptr(evsel, args, sample);
 
 	if (ttrace->entry_str == NULL) {
 		ttrace->entry_str = malloc(1024);
 		if (!ttrace->entry_str)
-			return -1;
+			goto out_put;
 	}
 
 	if (!trace->summary_only)
@@ -1757,8 +1757,10 @@ static int trace__sys_enter(struct trace *trace, struct perf_evsel *evsel,
 		thread__put(trace->current);
 		trace->current = thread__get(thread);
 	}
-
-	return 0;
+	err = 0;
+out_put:
+	thread__put(thread);
+	return err;
 }
 
 static int trace__sys_exit(struct trace *trace, struct perf_evsel *evsel,
@@ -1768,7 +1770,7 @@ static int trace__sys_exit(struct trace *trace, struct perf_evsel *evsel,
 	long ret;
 	u64 duration = 0;
 	struct thread *thread;
-	int id = perf_evsel__sc_tp_uint(evsel, id, sample);
+	int id = perf_evsel__sc_tp_uint(evsel, id, sample), err = -1;
 	struct syscall *sc = trace__syscall_info(trace, evsel, id);
 	struct thread_trace *ttrace;
 
@@ -1781,7 +1783,7 @@ static int trace__sys_exit(struct trace *trace, struct perf_evsel *evsel,
 	thread = machine__findnew_thread(trace->host, sample->pid, sample->tid);
 	ttrace = thread__trace(thread, trace->output);
 	if (ttrace == NULL)
-		return -1;
+		goto out_put;
 
 	if (trace->summary)
 		thread__update_stats(ttrace, id, sample);
@@ -1835,8 +1837,10 @@ signed_print:
 	fputc('\n', trace->output);
 out:
 	ttrace->entry_pending = false;
-
-	return 0;
+	err = 0;
+out_put:
+	thread__put(thread);
+	return err;
 }
 
 static int trace__vfs_getname(struct trace *trace, struct perf_evsel *evsel,
@@ -1863,6 +1867,7 @@ static int trace__sched_stat_runtime(struct trace *trace, struct perf_evsel *evs
 
 	ttrace->runtime_ms += runtime_ms;
 	trace->runtime_ms += runtime_ms;
+	thread__put(thread);
 	return 0;
 
 out_dump:
@@ -1872,6 +1877,7 @@ out_dump:
 	       (pid_t)perf_evsel__intval(evsel, sample, "pid"),
 	       runtime,
 	       perf_evsel__intval(evsel, sample, "vruntime"));
+	thread__put(thread);
 	return 0;
 }
 
@@ -1924,11 +1930,12 @@ static int trace__pgfault(struct trace *trace,
 	struct addr_location al;
 	char map_type = 'd';
 	struct thread_trace *ttrace;
+	int err = -1;
 
 	thread = machine__findnew_thread(trace->host, sample->pid, sample->tid);
 	ttrace = thread__trace(thread, trace->output);
 	if (ttrace == NULL)
-		return -1;
+		goto out_put;
 
 	if (evsel->attr.config == PERF_COUNT_SW_PAGE_FAULTS_MAJ)
 		ttrace->pfmaj++;
@@ -1936,7 +1943,7 @@ static int trace__pgfault(struct trace *trace,
 		ttrace->pfmin++;
 
 	if (trace->summary_only)
-		return 0;
+		goto out;
 
 	thread__find_addr_location(thread, cpumode, MAP__FUNCTION,
 			      sample->ip, &al);
@@ -1967,8 +1974,11 @@ static int trace__pgfault(struct trace *trace,
 	print_location(trace->output, sample, &al, true, false);
 
 	fprintf(trace->output, " (%c%c)\n", map_type, al.level);
-
-	return 0;
+out:
+	err = 0;
+out_put:
+	thread__put(thread);
+	return err;
 }
 
 static bool skip_sample(struct trace *trace, struct perf_sample *sample)
diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
index ca0e480e741b..e2a432b67d52 100644
--- a/tools/perf/tests/code-reading.c
+++ b/tools/perf/tests/code-reading.c
@@ -248,6 +248,7 @@ static int process_sample_event(struct machine *machine,
 	struct perf_sample sample;
 	struct thread *thread;
 	u8 cpumode;
+	int ret;
 
 	if (perf_evlist__parse_sample(evlist, event, &sample)) {
 		pr_debug("perf_evlist__parse_sample failed\n");
@@ -262,7 +263,9 @@ static int process_sample_event(struct machine *machine,
 
 	cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
 
-	return read_object_code(sample.ip, READLEN, cpumode, thread, state);
+	ret = read_object_code(sample.ip, READLEN, cpumode, thread, state);
+	thread__put(thread);
+	return ret;
 }
 
 static int process_event(struct machine *machine, struct perf_evlist *evlist,
@@ -457,13 +460,13 @@ static int do_test_code_reading(bool try_kcore)
 	thread = machine__findnew_thread(machine, pid, pid);
 	if (!thread) {
 		pr_debug("machine__findnew_thread failed\n");
-		goto out_err;
+		goto out_put;
 	}
 
 	cpus = cpu_map__new(NULL);
 	if (!cpus) {
 		pr_debug("cpu_map__new failed\n");
-		goto out_err;
+		goto out_put;
 	}
 
 	while (1) {
@@ -472,7 +475,7 @@ static int do_test_code_reading(bool try_kcore)
 		evlist = perf_evlist__new();
 		if (!evlist) {
 			pr_debug("perf_evlist__new failed\n");
-			goto out_err;
+			goto out_put;
 		}
 
 		perf_evlist__set_maps(evlist, cpus, threads);
@@ -485,7 +488,7 @@ static int do_test_code_reading(bool try_kcore)
 		ret = parse_events(evlist, str, NULL);
 		if (ret < 0) {
 			pr_debug("parse_events failed\n");
-			goto out_err;
+			goto out_put;
 		}
 
 		perf_evlist__config(evlist, &opts);
@@ -506,7 +509,7 @@ static int do_test_code_reading(bool try_kcore)
 				continue;
 			}
 			pr_debug("perf_evlist__open failed\n");
-			goto out_err;
+			goto out_put;
 		}
 		break;
 	}
@@ -514,7 +517,7 @@ static int do_test_code_reading(bool try_kcore)
 	ret = perf_evlist__mmap(evlist, UINT_MAX, false);
 	if (ret < 0) {
 		pr_debug("perf_evlist__mmap failed\n");
-		goto out_err;
+		goto out_put;
 	}
 
 	perf_evlist__enable(evlist);
@@ -525,7 +528,7 @@ static int do_test_code_reading(bool try_kcore)
 
 	ret = process_events(machine, evlist, &state);
 	if (ret < 0)
-		goto out_err;
+		goto out_put;
 
 	if (!have_vmlinux && !have_kcore && !try_kcore)
 		err = TEST_CODE_READING_NO_KERNEL_OBJ;
@@ -535,7 +538,10 @@ static int do_test_code_reading(bool try_kcore)
 		err = TEST_CODE_READING_NO_ACCESS;
 	else
 		err = TEST_CODE_READING_OK;
+out_put:
+	thread__put(thread);
 out_err:
+
 	if (evlist) {
 		perf_evlist__delete(evlist);
 	} else {
diff --git a/tools/perf/tests/dwarf-unwind.c b/tools/perf/tests/dwarf-unwind.c
index 0bf06bec68c7..9b748e1ad46e 100644
--- a/tools/perf/tests/dwarf-unwind.c
+++ b/tools/perf/tests/dwarf-unwind.c
@@ -170,6 +170,7 @@ int test__dwarf_unwind(void)
 	}
 
 	err = krava_1(thread);
+	thread__put(thread);
 
  out:
 	machine__delete_threads(machine);
diff --git a/tools/perf/tests/hists_common.c b/tools/perf/tests/hists_common.c
index a62c09134516..456f884eb27b 100644
--- a/tools/perf/tests/hists_common.c
+++ b/tools/perf/tests/hists_common.c
@@ -96,6 +96,7 @@ struct machine *setup_fake_machine(struct machines *machines)
 			goto out;
 
 		thread__set_comm(thread, fake_threads[i].comm, 0);
+		thread__put(thread);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(fake_mmap_info); i++) {
diff --git a/tools/perf/tests/hists_cumulate.c b/tools/perf/tests/hists_cumulate.c
index b08a95a5ca1a..620f626e5b35 100644
--- a/tools/perf/tests/hists_cumulate.c
+++ b/tools/perf/tests/hists_cumulate.c
@@ -105,8 +105,10 @@ static int add_hist_entries(struct hists *hists, struct machine *machine)
 			goto out;
 
 		if (hist_entry_iter__add(&iter, &al, evsel, &sample,
-					 PERF_MAX_STACK_DEPTH, NULL) < 0)
+					 PERF_MAX_STACK_DEPTH, NULL) < 0) {
+			addr_location__put(&al);
 			goto out;
+		}
 
 		fake_samples[i].thread = al.thread;
 		fake_samples[i].map = al.map;
diff --git a/tools/perf/tests/hists_filter.c b/tools/perf/tests/hists_filter.c
index 108488cd71fa..82e1ee52e024 100644
--- a/tools/perf/tests/hists_filter.c
+++ b/tools/perf/tests/hists_filter.c
@@ -82,8 +82,10 @@ static in