[PATCH 09/10] perf tests pmu-events: Test metricgroup descriptions and concurrent lookups

[Ian Rogers <irogers@google.com>]

Add metricgroups.json to arch/test/test_soc/cpu/ with prefix-overlapping
keys ("group1" and "group1A"), regenerate empty-pmu-events.c, and add a
"metricgroups" subtest to "PMU JSON event tests".

The new subtest verifies exact and prefix-overlapping metricgroup binary
searches via describe_metricgroup(), validates metricgroup descriptions
across all core and system metrics, and launches 8 concurrent worker
threads scanning core/sys events and metrics to stress-test lock-free
block decompression and shared dictionary initialization.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 .../arch/test/test_soc/cpu/metricgroups.json  |   4 +
 tools/perf/pmu-events/empty-pmu-events.c      |  84 +++----
 tools/perf/tests/pmu-events.c                 | 209 +++++++++++++++++-
 3 files changed, 259 insertions(+), 38 deletions(-)
 create mode 100644 tools/perf/pmu-events/arch/test/test_soc/cpu/metricgroups.json

diff --git a/tools/perf/pmu-events/arch/test/test_soc/cpu/metricgroups.json b/tools/perf/pmu-events/arch/test/test_soc/cpu/metricgroups.json
new file mode 100644
index 000000000000..2bf7e4fbff3c
--- /dev/null
+++ b/tools/perf/pmu-events/arch/test/test_soc/cpu/metricgroups.json
@@ -0,0 +1,4 @@
+{
+    "group1": "Test metricgroup 1",
+    "group1A": "Test metricgroup 1A"
+}
diff --git a/tools/perf/pmu-events/empty-pmu-events.c b/tools/perf/pmu-events/empty-pmu-events.c
index 46a32a4cb9d6..f4009e5c51ee 100644
--- a/tools/perf/pmu-events/empty-pmu-events.c
+++ b/tools/perf/pmu-events/empty-pmu-events.c
@@ -2605,42 +2605,50 @@ static const char big_c_string[] =
 /* offset=128790 */
 "M3\000\0001 / M3\000\000\000\000\000\000\000\000000"
 /* offset=128811 */
+"Test metricgroup 1\000"
+/* offset=128830 */
+"Test metricgroup 1A\000"
+/* offset=128850 */
 "backend_cycles_idle\000Default\000(stalled\\-cycles\\-backend / cpu\\-cycles if has_event(stalled\\-cycles\\-backend) else 0)\000backend_cycles_idle > 0.2\000Backend stalls per cycle\000\000\000\000\000\000001"
-/* offset=128985 */
+/* offset=129024 */
 "branch_frequency\000Default\000branches / (software@cpu\\-clock\\,name\\=cpu\\-clock@ if #target_cpu else software@task\\-clock\\,name\\=task\\-clock@)\000\000Branches per CPU second\000\0001000M/sec\000\000\000\000011"
-/* offset=129165 */
+/* offset=129204 */
 "branch_miss_rate\000Default\000branch\\-misses / branches\000branch_miss_rate > 0.05\000Branch miss rate\000\000100%\000\000\000\000001"
-/* offset=129269 */
+/* offset=129308 */
 "cache_miss_cycles\000group1\000dcache_miss_cpi + icache_miss_cycles\000\000\000\000\000\000\000\000000"
-/* offset=129341 */
+/* offset=129380 */
 "cs_per_second\000Default\000software@context\\-switches\\,name\\=context\\-switches@ * 1e9 / (software@cpu\\-clock\\,name\\=cpu\\-clock@ if #target_cpu else software@task\\-clock\\,name\\=task\\-clock@)\000\000Context switches per CPU second\000\0001cs/sec\000\000\000\000011"
-/* offset=129574 */
+/* offset=129613 */
 "cycles_frequency\000Default\000cpu\\-cycles / (software@cpu\\-clock\\,name\\=cpu\\-clock@ if #target_cpu else software@task\\-clock\\,name\\=task\\-clock@)\000\000Cycles per CPU second\000\0001GHz\000\000\000\000011"
-/* offset=129750 */
+/* offset=129789 */
 "dcache_miss_cpi\000\000l1d\\-loads\\-misses / inst_retired.any\000\000\000\000\000\000\000\000000"
-/* offset=129815 */
+/* offset=129854 */
 "dtlb_miss_rate\000Default3\000dTLB\\-load\\-misses / dTLB\\-loads\000dtlb_miss_rate > 0.05\000dTLB miss rate\000\000100%\000\000\000\000001"
-/* offset=129921 */
+/* offset=129960 */
 "frontend_cycles_idle\000Default\000(stalled\\-cycles\\-frontend / cpu\\-cycles if has_event(stalled\\-cycles\\-frontend) else 0)\000frontend_cycles_idle > 0.1\000Frontend stalls per cycle\000\000\000\000\000\000001"
-/* offset=130100 */
+/* offset=130139 */
+"group1\000"
+/* offset=130146 */
+"group1A\000"
+/* offset=130154 */
 "icache_miss_cycles\000\000l1i\\-loads\\-misses / inst_retired.any\000\000\000\000\000\000\000\000000"
-/* offset=130168 */
+/* offset=130222 */
 "insn_per_cycle\000Default\000instructions / cpu\\-cycles\000insn_per_cycle < 1\000Instructions Per Cycle\000\0001instructions\000\000\000\000001"
-/* offset=130281 */
+/* offset=130335 */
 "itlb_miss_rate\000Default3\000iTLB\\-load\\-misses / iTLB\\-loads\000itlb_miss_rate > 0.05\000iTLB miss rate\000\000100%\000\000\000\000001"
-/* offset=130387 */
+/* offset=130441 */
 "l1_prefetch_miss_rate\000Default4\000L1\\-dcache\\-prefetch\\-misses / L1\\-dcache\\-prefetches\000l1_prefetch_miss_rate > 0.05\000L1 prefetch miss rate\000\000100%\000\000\000\000001"
-/* offset=130535 */
+/* offset=130589 */
 "l1d_miss_rate\000Default2\000L1\\-dcache\\-load\\-misses / L1\\-dcache\\-loads\000l1d_miss_rate > 0.05\000L1D  miss rate\000\000100%\000\000\000\000001"
-/* offset=130651 */
+/* offset=130705 */
 "l1i_miss_rate\000Default3\000L1\\-icache\\-load\\-misses / L1\\-icache\\-loads\000l1i_miss_rate > 0.05\000L1I miss rate\000\000100%\000\000\000\000001"
-/* offset=130766 */
+/* offset=130820 */
 "llc_miss_rate\000Default2\000LLC\\-load\\-misses / LLC\\-loads\000llc_miss_rate > 0.05\000LLC miss rate\000\000100%\000\000\000\000001"
-/* offset=130867 */
+/* offset=130921 */
 "migrations_per_second\000Default\000software@cpu\\-migrations\\,name\\=cpu\\-migrations@ * 1e9 / (software@cpu\\-clock\\,name\\=cpu\\-clock@ if #target_cpu else software@task\\-clock\\,name\\=task\\-clock@)\000\000Process migrations to a new CPU per CPU second\000\0001migrations/sec\000\000\000\000011"
-/* offset=131127 */
+/* offset=131181 */
 "page_faults_per_second\000Default\000software@page\\-faults\\,name\\=page\\-faults@ * 1e9 / (software@cpu\\-clock\\,name\\=cpu\\-clock@ if #target_cpu else software@task\\-clock\\,name\\=task\\-clock@)\000\000Page faults per CPU second\000\0001faults/sec\000\000\000\000011"
-/* offset=131358 */
+/* offset=131412 */
 "stalled_cycles_per_instruction\000Default\000(max(stalled\\-cycles\\-frontend, stalled\\-cycles\\-backend) / instructions if has_event(stalled\\-cycles\\-frontend) & has_event(stalled\\-cycles\\-backend) else (stalled\\-cycles\\-frontend / instructions if has_event(stalled\\-cycles\\-frontend) else (stalled\\-cycles\\-backend / instructions if has_event(stalled\\-cycles\\-backend) else 0)))\000\000Max front or backend stalls per instruction\000\000\000\000\000\000001"
 ;
 
@@ -5197,37 +5205,37 @@ static const struct compact_pmu_event pmu_metrics__common_default_core[] = {
 	/* CPUs_utilized\000Default\000(software@cpu\\-clock\\,name\\=cpu\\-clock@ if #targ... */
 	{ 127820 },
 	/* backend_cycles_idle\000Default\000(stalled\\-cycles\\-backend / cpu\\-cycles if ... */
-	{ 128811 },
+	{ 128850 },
 	/* branch_frequency\000Default\000branches / (software@cpu\\-clock\\,name\\=cpu\\-c... */
-	{ 128985 },
+	{ 129024 },
 	/* branch_miss_rate\000Default\000branch\\-misses / branches\000branch_miss_rate > ... */
-	{ 129165 },
+	{ 129204 },
 	/* cs_per_second\000Default\000software@context\\-switches\\,name\\=context\\-switc... */
-	{ 129341 },
+	{ 129380 },
 	/* cycles_frequency\000Default\000cpu\\-cycles / (software@cpu\\-clock\\,name\\=cpu... */
-	{ 129574 },
+	{ 129613 },
 	/* dtlb_miss_rate\000Default3\000dTLB\\-load\\-misses / dTLB\\-loads\000dtlb_miss_r... */
-	{ 129815 },
+	{ 129854 },
 	/* frontend_cycles_idle\000Default\000(stalled\\-cycles\\-frontend / cpu\\-cycles i... */
-	{ 129921 },
+	{ 129960 },
 	/* insn_per_cycle\000Default\000instructions / cpu\\-cycles\000insn_per_cycle < 1\0... */
-	{ 130168 },
+	{ 130222 },
 	/* itlb_miss_rate\000Default3\000iTLB\\-load\\-misses / iTLB\\-loads\000itlb_miss_r... */
-	{ 130281 },
+	{ 130335 },
 	/* l1_prefetch_miss_rate\000Default4\000L1\\-dcache\\-prefetch\\-misses / L1\\-dcac... */
-	{ 130387 },
+	{ 130441 },
 	/* l1d_miss_rate\000Default2\000L1\\-dcache\\-load\\-misses / L1\\-dcache\\-loads\0... */
-	{ 130535 },
+	{ 130589 },
 	/* l1i_miss_rate\000Default3\000L1\\-icache\\-load\\-misses / L1\\-icache\\-loads\0... */
-	{ 130651 },
+	{ 130705 },
 	/* llc_miss_rate\000Default2\000LLC\\-load\\-misses / LLC\\-loads\000llc_miss_rate ... */
-	{ 130766 },
+	{ 130820 },
 	/* migrations_per_second\000Default\000software@cpu\\-migrations\\,name\\=cpu\\-mig... */
-	{ 130867 },
+	{ 130921 },
 	/* page_faults_per_second\000Default\000software@page\\-faults\\,name\\=page\\-faul... */
-	{ 131127 },
+	{ 131181 },
 	/* stalled_cycles_per_instruction\000Default\000(max(stalled\\-cycles\\-frontend, s... */
-	{ 131358 },
+	{ 131412 },
 
 };
 
@@ -5338,11 +5346,11 @@ static const struct compact_pmu_event pmu_metrics__test_soc_cpu_default_core[] =
 	/* M3\000\0001 / M3\000\000\000\000\000\000\000\000000 */
 	{ 128790 },
 	/* cache_miss_cycles\000group1\000dcache_miss_cpi + icache_miss_cycles\000\000\000\... */
-	{ 129269 },
+	{ 129308 },
 	/* dcache_miss_cpi\000\000l1d\\-loads\\-misses / inst_retired.any\000\000\000\000\0... */
-	{ 129750 },
+	{ 129789 },
 	/* icache_miss_cycles\000\000l1i\\-loads\\-misses / inst_retired.any\000\000\000\00... */
-	{ 130100 },
+	{ 130154 },
 
 };
 
@@ -6034,6 +6042,8 @@ int pmu_metrics_table__iterate_tables(pmu_metrics_table_iter_t fn, void *data)
 }
 
 static const uint32_t metricgroups[][2] = {
+	{ 130139, 128811 }, /* group1\000 => Test metricgroup 1\000 */
+	{ 130146, 128830 }, /* group1A\000 => Test metricgroup 1A\000 */
 
 };
 
diff --git a/tools/perf/tests/pmu-events.c b/tools/perf/tests/pmu-events.c
index e6e4f4cf1fa2..a06525cafb48 100644
--- a/tools/perf/tests/pmu-events.c
+++ b/tools/perf/tests/pmu-events.c
@@ -5,7 +5,12 @@
 #include "pmus.h"
 #include "tests.h"
 #include <errno.h>
+#include <pthread.h>
+#include <stdbool.h>
+#include <stddef.h>
 #include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
 #include <linux/kernel.h>
 #include <linux/zalloc.h>
 #include "debug.h"
@@ -1094,6 +1099,202 @@ static int test__parsing_threshold(struct test_suite *test __maybe_unused,
 	return pmu_for_each_sys_metric(test__parsing_threshold_callback, NULL);
 }
 
+static int check_event_strings_cb(const struct pmu_event *pe,
+				  const struct pmu_events_table *table __maybe_unused,
+				  void *data __maybe_unused)
+{
+	if (!pe->pmu || !pe->name || (!pe->event && !pe->desc))
+		return TEST_FAIL;
+	return 0;
+}
+
+static int check_find_btb_cb(const struct pmu_event *pe,
+			     const struct pmu_events_table *table __maybe_unused,
+			     void *data)
+{
+	int *matched = data;
+
+	if (!pe->event || strcmp(pe->event, "event=0x8a"))
+		return TEST_FAIL;
+	(*matched)++;
+	return 0;
+}
+
+static int check_find_ipc_cb(const struct pmu_metric *pm,
+			     const struct pmu_metrics_table *table __maybe_unused,
+			     void *data)
+{
+	int *matched = data;
+
+	if (!pm->metric_group || strcmp(pm->metric_group, "group1"))
+		return TEST_FAIL;
+	(*matched)++;
+	return 0;
+}
+
+static int check_metricgroup_strings_cb(const struct pmu_metric *pm,
+					const struct pmu_metrics_table *table __maybe_unused,
+					void *data __maybe_unused)
+{
+	char *dup, *group, *saveptr = NULL;
+
+	if (!pm->pmu || !pm->metric_expr)
+		return TEST_FAIL;
+	if (!pm->metric_group)
+		return 0;
+
+	dup = strdup(pm->metric_group);
+	if (!dup)
+		return -ENOMEM;
+
+	for (group = strtok_r(dup, ";", &saveptr); group;
+	     group = strtok_r(NULL, ";", &saveptr)) {
+		const char *desc = describe_metricgroup(group);
+
+		if (desc && desc[0] == '\0') {
+			free(dup);
+			return TEST_FAIL;
+		}
+		if (!strcmp(group, "group1") &&
+		    (!desc || strcmp(desc, "Test metricgroup 1"))) {
+			free(dup);
+			return TEST_FAIL;
+		}
+	}
+	free(dup);
+	return 0;
+}
+
+struct concurrent_lookup_gate {
+	pthread_mutex_t mutex;
+	pthread_cond_t cond;
+	bool start;
+};
+
+struct concurrent_lookup_worker {
+	struct concurrent_lookup_gate *gate;
+	int err;
+};
+
+static int run_lookup_checks(void)
+{
+	const struct pmu_events_table *etest;
+	const struct pmu_metrics_table *mtest;
+	const char *desc;
+	int matched, err;
+
+	desc = describe_metricgroup("group1");
+	if (!desc || strcmp(desc, "Test metricgroup 1")) {
+		pr_err("Unexpected group1 description: %s\n", desc ?: "(null)");
+		return TEST_FAIL;
+	}
+
+	desc = describe_metricgroup("group1A");
+	if (!desc || strcmp(desc, "Test metricgroup 1A")) {
+		pr_err("Unexpected group1A description: %s\n", desc ?: "(null)");
+		return TEST_FAIL;
+	}
+
+	if (describe_metricgroup("group1_nonexistent") || describe_metricgroup("")) {
+		pr_err("Unexpected match for nonexistent metricgroup\n");
+		return TEST_FAIL;
+	}
+
+	if (describe_metricgroup("BadSpec") && !describe_metricgroup("Bad")) {
+		pr_err("Prefix metricgroup 'Bad' missing when 'BadSpec' is present\n");
+		return TEST_FAIL;
+	}
+
+	etest = find_core_events_table("testarch", "testcpu");
+	mtest = find_core_metrics_table("testarch", "testcpu");
+	if (!etest || !mtest)
+		return TEST_FAIL;
+
+	matched = 0;
+	err = pmu_events_table__find_event(etest, NULL, "bp_l1_btb_correct",
+					   check_find_btb_cb, &matched);
+	if (err || matched != 1)
+		return TEST_FAIL;
+
+	matched = 0;
+	err = pmu_metrics_table__find_metric(mtest, NULL, "IPC",
+					     check_find_ipc_cb, &matched);
+	if (err || matched != 1)
+		return TEST_FAIL;
+
+	err = pmu_for_each_core_event(check_event_strings_cb, NULL);
+	if (err)
+		return err;
+
+	err = pmu_for_each_sys_event(check_event_strings_cb, NULL);
+	if (err)
+		return err;
+
+	err = pmu_for_each_core_metric(check_metricgroup_strings_cb, NULL);
+	if (err)
+		return err;
+
+	return pmu_for_each_sys_metric(check_metricgroup_strings_cb, NULL);
+}
+
+static void *concurrent_lookup_thread(void *arg)
+{
+	struct concurrent_lookup_worker *worker = arg;
+
+	pthread_mutex_lock(&worker->gate->mutex);
+	while (!worker->gate->start)
+		pthread_cond_wait(&worker->gate->cond, &worker->gate->mutex);
+	pthread_mutex_unlock(&worker->gate->mutex);
+
+	worker->err = run_lookup_checks();
+	return NULL;
+}
+
+/*
+ * Verify metricgroup binary search and release 8 worker threads simultaneously
+ * to stress-test lock-free on-demand decompression of ~8 KiB string blocks in
+ * pmu_events__get_string() (big_c_string_cache[]) and shared dictionary setup.
+ */
+static int test__metricgroups(struct test_suite *test __maybe_unused,
+			      int subtest __maybe_unused)
+{
+	struct concurrent_lookup_gate gate = {
+		.mutex = PTHREAD_MUTEX_INITIALIZER,
+		.cond = PTHREAD_COND_INITIALIZER,
+		.start = false,
+	};
+	struct concurrent_lookup_worker workers[8];
+	pthread_t threads[ARRAY_SIZE(workers)];
+	size_t created = 0;
+	int ret = TEST_OK;
+
+	for (size_t i = 0; i < ARRAY_SIZE(workers); i++) {
+		workers[i].gate = &gate;
+		workers[i].err = 0;
+		if (pthread_create(&threads[i], NULL, concurrent_lookup_thread,
+				   &workers[i]) != 0) {
+			ret = TEST_FAIL;
+			break;
+		}
+		created++;
+	}
+
+	pthread_mutex_lock(&gate.mutex);
+	gate.start = true;
+	pthread_cond_broadcast(&gate.cond);
+	pthread_mutex_unlock(&gate.mutex);
+
+	for (size_t i = 0; i < created; i++) {
+		pthread_join(threads[i], NULL);
+		if (workers[i].err && ret == TEST_OK)
+			ret = workers[i].err;
+	}
+
+	pthread_cond_destroy(&gate.cond);
+	pthread_mutex_destroy(&gate.mutex);
+	return ret;
+}
+
 struct populate_cb_data {
 	struct test_case *test_cases;
 	size_t curr;
@@ -1145,7 +1346,7 @@ static struct test_case pmu_events_tests[];
 static int setup_pmu_events_suite(struct test_suite *suite)
 {
 	size_t num_tables = 0;
-	size_t num_fixed_tests = 4;
+	size_t num_fixed_tests = 5;
 	size_t tests_per_table = 2;
 	size_t total_tests;
 	struct test_case *test_cases;
@@ -1171,6 +1372,11 @@ static int setup_pmu_events_suite(struct test_suite *suite)
 		.desc = "PMU event table sanity",
 		.run_case = test__pmu_event_table,
 	};
+	test_cases[curr++] = (struct test_case){
+		.name = "metricgroups",
+		.desc = "PMU metricgroup descriptions and concurrent lookups",
+		.run_case = test__metricgroups,
+	};
 	test_cases[curr++] = (struct test_case){
 		.name = "aliases",
 		.desc = "PMU event map aliases",
@@ -1208,6 +1414,7 @@ static int setup_pmu_events_suite(struct test_suite *suite)
 
 static struct test_case pmu_events_tests[] = {
 	TEST_CASE("PMU event table sanity", pmu_event_table),
+	TEST_CASE("PMU metricgroup descriptions and concurrent lookups", metricgroups),
 	TEST_CASE("PMU event map aliases", aliases),
 	TEST_CASE_REASON("Parsing of PMU event table metrics", parsing,
 			 "some metrics failed"),
-- 
2.56.1.703.gf848ed807c-goog