perf stat: Support per-cluster aggregation

Some platforms have 'cluster' topology and CPUs in the cluster will
share resources like L3 Cache Tag (for HiSilicon Kunpeng SoC) or L2
cache (for Intel Jacobsville). Currently parsing and building cluster
topology have been supported since [1].

perf stat has already supported aggregation for other topologies like
die or socket, etc. It'll be useful to aggregate per-cluster to find
problems like L3T bandwidth contention.

This patch add support for "--per-cluster" option for per-cluster
aggregation. Also update the docs and related test. The output will
be like:

[root@localhost tmp]# perf stat -a -e LLC-load --per-cluster -- sleep 5

Performance counter stats for 'system wide':

S56-D0-CLS158 4 1,321,521,570 LLC-load
S56-D0-CLS594 4 794,211,453 LLC-load
S56-D0-CLS1030 4 41,623 LLC-load
S56-D0-CLS1466 4 41,646 LLC-load
S56-D0-CLS1902 4 16,863 LLC-load
S56-D0-CLS2338 4 15,721 LLC-load
S56-D0-CLS2774 4 22,671 LLC-load
[...]

On a legacy system without cluster or cluster support, the output will
be look like:
[root@localhost perf]# perf stat -a -e cycles --per-cluster -- sleep 1

Performance counter stats for 'system wide':

S56-D0-CLS0 64 18,011,485 cycles
S7182-D0-CLS0 64 16,548,835 cycles

Note that this patch doesn't mix the cluster information in the outputs
of --per-core to avoid breaking any tools/scripts using it.

Note that perf recently supports "--per-cache" aggregation, but it's not
the same with the cluster although cluster CPUs may share some cache
resources. For example on my machine all clusters within a die share the
same L3 cache:
$ cat /sys/devices/system/cpu/cpu0/cache/index3/shared_cpu_list
0-31
$ cat /sys/devices/system/cpu/cpu0/topology/cluster_cpus_list
0-3

[1] commit c5e22feffdd7 ("topology: Represent clusters of CPUs within a die")

Tested-by: Jie Zhan <[e-mail]>
Reviewed-by: Tim Chen <[e-mail]>
Reviewed-by: Ian Rogers <[e-mail]>
Signed-off-by: Yicong Yang <[e-mail]>
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Signed-off-by: Namhyung Kim <[e-mail]>
Link: https://lore.kernel.org/r/[e-mail]

tools/perf/Documentation/perf-stat.txt

···
 die number and the number of online processors on that die. This is
 useful to gauge the amount of aggregation.
 
+--per-cluster::
+Aggregate counts per processor cluster for system-wide mode measurement.  This
+is a useful mode to detect imbalance between clusters.  To enable this mode,
+use --per-cluster in addition to -a. (system-wide).  The output includes the
+cluster number and the number of online processors on that cluster. This is
+useful to gauge the amount of aggregation. The information of cluster ID and
+related CPUs can be gotten from /sys/devices/system/cpu/cpuX/topology/cluster_{id, cpus}.
+
 --per-cache::
 Aggregate counts per cache instance for system-wide mode measurements.  By
 default, the aggregation happens for the cache level at the highest index
···
 
 --per-die::
 Aggregate counts per processor die for system-wide mode measurements.
+
+--per-cluster::
+Aggregate counts perf processor cluster for system-wide mode measurements.
 
 --per-cache::
 Aggregate counts per cache instance for system-wide mode measurements.  By

tools/perf/builtin-stat.c

···
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-die", &stat_config.aggr_mode,
 		     "aggregate counts per processor die", AGGR_DIE),
+	OPT_SET_UINT(0, "per-cluster", &stat_config.aggr_mode,
+		     "aggregate counts per processor cluster", AGGR_CLUSTER),
 	OPT_CALLBACK_OPTARG(0, "per-cache", &stat_config.aggr_mode, &stat_config.aggr_level,
 			    "cache level", "aggregate count at this cache level (Default: LLC)",
 			    parse_cache_level),
···
 static const char *const aggr_mode__string[] = {
 	[AGGR_CORE] = "core",
 	[AGGR_CACHE] = "cache",
+	[AGGR_CLUSTER] = "cluster",
 	[AGGR_DIE] = "die",
 	[AGGR_GLOBAL] = "global",
 	[AGGR_NODE] = "node",
···
 						  struct perf_cpu cpu)
 {
 	return aggr_cpu_id__cache(cpu, /*data=*/NULL);
+}
+
+static struct aggr_cpu_id perf_stat__get_cluster(struct perf_stat_config *config __maybe_unused,
+						 struct perf_cpu cpu)
+{
+	return aggr_cpu_id__cluster(cpu, /*data=*/NULL);
 }
 
 static struct aggr_cpu_id perf_stat__get_core(struct perf_stat_config *config __maybe_unused,
···
 	return perf_stat__get_aggr(config, perf_stat__get_die, cpu);
 }
 
+static struct aggr_cpu_id perf_stat__get_cluster_cached(struct perf_stat_config *config,
+							struct perf_cpu cpu)
+{
+	return perf_stat__get_aggr(config, perf_stat__get_cluster, cpu);
+}
+
 static struct aggr_cpu_id perf_stat__get_cache_id_cached(struct perf_stat_config *config,
 							 struct perf_cpu cpu)
 {
···
 		return aggr_cpu_id__socket;
 	case AGGR_DIE:
 		return aggr_cpu_id__die;
+	case AGGR_CLUSTER:
+		return aggr_cpu_id__cluster;
 	case AGGR_CACHE:
 		return aggr_cpu_id__cache;
 	case AGGR_CORE:
···
 		return perf_stat__get_socket_cached;
 	case AGGR_DIE:
 		return perf_stat__get_die_cached;
+	case AGGR_CLUSTER:
+		return perf_stat__get_cluster_cached;
 	case AGGR_CACHE:
 		return perf_stat__get_cache_id_cached;
 	case AGGR_CORE:
···
 	return id;
 }
 
+static struct aggr_cpu_id perf_env__get_cluster_aggr_by_cpu(struct perf_cpu cpu,
+							    void *data)
+{
+	struct perf_env *env = data;
+	struct aggr_cpu_id id = aggr_cpu_id__empty();
+
+	if (cpu.cpu != -1) {
+		id.socket = env->cpu[cpu.cpu].socket_id;
+		id.die = env->cpu[cpu.cpu].die_id;
+		id.cluster = env->cpu[cpu.cpu].cluster_id;
+	}
+
+	return id;
+}
+
 static struct aggr_cpu_id perf_env__get_core_aggr_by_cpu(struct perf_cpu cpu, void *data)
 {
 	struct perf_env *env = data;
···
 
 	if (cpu.cpu != -1) {
 		/*
-		 * core_id is relative to socket and die,
-		 * we need a global id. So we set
-		 * socket, die id and core id
+		 * core_id is relative to socket, die and cluster, we need a
+		 * global id. So we set socket, die id, cluster id and core id.
 		 */
 		id.socket = env->cpu[cpu.cpu].socket_id;
 		id.die = env->cpu[cpu.cpu].die_id;
+		id.cluster = env->cpu[cpu.cpu].cluster_id;
 		id.core = env->cpu[cpu.cpu].core_id;
 	}
 
···
 	return perf_env__get_die_aggr_by_cpu(cpu, &perf_stat.session->header.env);
 }
 
+static struct aggr_cpu_id perf_stat__get_cluster_file(struct perf_stat_config *config __maybe_unused,
+						      struct perf_cpu cpu)
+{
+	return perf_env__get_cluster_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+}
+
 static struct aggr_cpu_id perf_stat__get_cache_file(struct perf_stat_config *config __maybe_unused,
 						    struct perf_cpu cpu)
 {
···
 		return perf_env__get_socket_aggr_by_cpu;
 	case AGGR_DIE:
 		return perf_env__get_die_aggr_by_cpu;
+	case AGGR_CLUSTER:
+		return perf_env__get_cluster_aggr_by_cpu;
 	case AGGR_CACHE:
 		return perf_env__get_cache_aggr_by_cpu;
 	case AGGR_CORE:
···
 		return perf_stat__get_socket_file;
 	case AGGR_DIE:
 		return perf_stat__get_die_file;
+	case AGGR_CLUSTER:
+		return perf_stat__get_cluster_file;
 	case AGGR_CACHE:
 		return perf_stat__get_cache_file;
 	case AGGR_CORE:
···
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-die", &perf_stat.aggr_mode,
 		     "aggregate counts per processor die", AGGR_DIE),
+	OPT_SET_UINT(0, "per-cluster", &perf_stat.aggr_mode,
+		     "aggregate counts perf processor cluster", AGGR_CLUSTER),
 	OPT_CALLBACK_OPTARG(0, "per-cache", &perf_stat.aggr_mode, &perf_stat.aggr_level,
 			    "cache level",
 			    "aggregate count at this cache level (Default: LLC)",

tools/perf/tests/shell/lib/perf_json_output_lint.py

···
 ap.add_argument('--per-core', action='store_true')
 ap.add_argument('--per-thread', action='store_true')
 ap.add_argument('--per-cache', action='store_true')
+ap.add_argument('--per-cluster', action='store_true')
 ap.add_argument('--per-die', action='store_true')
 ap.add_argument('--per-node', action='store_true')
 ap.add_argument('--per-socket', action='store_true')
···
       'cgroup': lambda x: True,
       'cpu': lambda x: isint(x),
       'cache': lambda x: True,
+      'cluster': lambda x: True,
       'die': lambda x: True,
       'event': lambda x: True,
       'event-runtime': lambda x: isfloat(x),
···
     expected_items = 7
   elif args.interval or args.per_thread or args.system_wide_no_aggr:
     expected_items = 8
-  elif args.per_core or args.per_socket or args.per_node or args.per_die or args.per_cache:
+  elif args.per_core or args.per_socket or args.per_node or args.per_die or args.per_cluster or args.per_cache:
     expected_items = 9
   else:
     # If no option is specified, don't check the number of items.

tools/perf/tests/shell/lib/stat_output.sh

···
 	echo "[Success]"
 }
 
+check_per_cluster()
+{
+	echo -n "Checking $1 output: per cluster "
+	if ParanoidAndNotRoot 0
+	then
+		echo "[Skip] paranoid and not root"
+		return
+	fi
+	perf stat --per-cluster -a $2 true
+	echo "[Success]"
+}
+
 check_per_die()
 {
 	echo -n "Checking $1 output: per die "

tools/perf/tests/shell/stat+csv_output.sh

···
 	;; "--per-socket")	exp=8
 	;; "--per-node")	exp=8
 	;; "--per-die")		exp=8
+	;; "--per-cluster")	exp=8
 	;; "--per-cache")	exp=8
 	esac
 
···
 	check_system_wide_no_aggr "CSV" "$perf_cmd"
 	check_per_core "CSV" "$perf_cmd"
 	check_per_cache_instance "CSV" "$perf_cmd"
+	check_per_cluster "CSV" "$perf_cmd"
 	check_per_die "CSV" "$perf_cmd"
 	check_per_socket "CSV" "$perf_cmd"
 else

tools/perf/tests/shell/stat+json_output.sh

···
 	echo "[Success]"
 }
 
+check_per_cluster()
+{
+	echo -n "Checking json output: per cluster "
+	if ParanoidAndNotRoot 0
+	then
+		echo "[Skip] paranoia and not root"
+		return
+	fi
+	perf stat -j --per-cluster -a true 2>&1 | $PYTHON $pythonchecker --per-cluster
+	echo "[Success]"
+}
+
 check_per_die()
 {
 	echo -n "Checking json output: per die "
···
 	check_system_wide_no_aggr
 	check_per_core
 	check_per_cache_instance
+	check_per_cluster
 	check_per_die
 	check_per_socket
 else

tools/perf/tests/shell/stat+std_output.sh

···
 	;; "--per-node")	prefix=3
 	;; "--per-die")		prefix=3
 	;; "--per-cache")	prefix=3
+	;; "--per-cluster")	prefix=3
 	esac
 
 	while read line
···
 	check_system_wide_no_aggr "STD" "$perf_cmd"
 	check_per_core "STD" "$perf_cmd"
 	check_per_cache_instance "STD" "$perf_cmd"
+	check_per_cluster "STD" "$perf_cmd"
 	check_per_die "STD" "$perf_cmd"
 	check_per_socket "STD" "$perf_cmd"
 else

tools/perf/util/cpumap.c

···
 		return a->socket - b->socket;
 	else if (a->die != b->die)
 		return a->die - b->die;
+	else if (a->cluster != b->cluster)
+		return a->cluster - b->cluster;
 	else if (a->cache_lvl != b->cache_lvl)
 		return a->cache_lvl - b->cache_lvl;
 	else if (a->cache != b->cache)
···
 	return id;
 }
 
+int cpu__get_cluster_id(struct perf_cpu cpu)
+{
+	int value, ret = cpu__get_topology_int(cpu.cpu, "cluster_id", &value);
+
+	return ret ?: value;
+}
+
+struct aggr_cpu_id aggr_cpu_id__cluster(struct perf_cpu cpu, void *data)
+{
+	int cluster = cpu__get_cluster_id(cpu);
+	struct aggr_cpu_id id;
+
+	/* There is no cluster_id on legacy system. */
+	if (cluster == -1)
+		cluster = 0;
+
+	id = aggr_cpu_id__die(cpu, data);
+	if (aggr_cpu_id__is_empty(&id))
+		return id;
+
+	id.cluster = cluster;
+	return id;
+}
+
 int cpu__get_core_id(struct perf_cpu cpu)
 {
 	int value, ret = cpu__get_topology_int(cpu.cpu, "core_id", &value);
···
 	struct aggr_cpu_id id;
 	int core = cpu__get_core_id(cpu);
 
-	/* aggr_cpu_id__die returns a struct with socket and die set. */
-	id = aggr_cpu_id__die(cpu, data);
+	/* aggr_cpu_id__die returns a struct with socket die, and cluster set. */
+	id = aggr_cpu_id__cluster(cpu, data);
 	if (aggr_cpu_id__is_empty(&id))
 		return id;
 
···
 		a->node == b->node &&
 		a->socket == b->socket &&
 		a->die == b->die &&
+		a->cluster == b->cluster &&
 		a->cache_lvl == b->cache_lvl &&
 		a->cache == b->cache &&
 		a->core == b->core &&
···
 		a->node == -1 &&
 		a->socket == -1 &&
 		a->die == -1 &&
+		a->cluster == -1 &&
 		a->cache_lvl == -1 &&
 		a->cache == -1 &&
 		a->core == -1 &&
···
 		.node = -1,
 		.socket = -1,
 		.die = -1,
+		.cluster = -1,
 		.cache_lvl = -1,
 		.cache = -1,
 		.core = -1,

tools/perf/util/cpumap.h

···
 	int socket;
 	/** The die id as read from /sys/devices/system/cpu/cpuX/topology/die_id. */
 	int die;
+	/** The cluster id as read from /sys/devices/system/cpu/cpuX/topology/cluster_id */
+	int cluster;
 	/** The cache level as read from /sys/devices/system/cpu/cpuX/cache/indexY/level */
 	int cache_lvl;
 	/**
···
  */
 int cpu__get_die_id(struct perf_cpu cpu);
 /**
+ * cpu__get_cluster_id - Returns the cluster id as read from
+ * /sys/devices/system/cpu/cpuX/topology/cluster_id for the given CPU
+ */
+int cpu__get_cluster_id(struct perf_cpu cpu);
+/**
  * cpu__get_core_id - Returns the core id as read from
  * /sys/devices/system/cpu/cpuX/topology/core_id for the given CPU.
  */
···
  */
 struct aggr_cpu_id aggr_cpu_id__die(struct perf_cpu cpu, void *data);
 /**
- * aggr_cpu_id__core - Create an aggr_cpu_id with the core, die and socket
- * populated with the core, die and socket for cpu. The function signature is
- * compatible with aggr_cpu_id_get_t.
+ * aggr_cpu_id__cluster - Create an aggr_cpu_id with cluster, die and socket
+ * populated with the cluster, die and socket for cpu. The function signature
+ * is compatible with aggr_cpu_id_get_t.
+ */
+struct aggr_cpu_id aggr_cpu_id__cluster(struct perf_cpu cpu, void *data);
+/**
+ * aggr_cpu_id__core - Create an aggr_cpu_id with the core, cluster, die and
+ * socket populated with the core, die and socket for cpu. The function
+ * signature is compatible with aggr_cpu_id_get_t.
  */
 struct aggr_cpu_id aggr_cpu_id__core(struct perf_cpu cpu, void *data);
 /**

tools/perf/util/env.h

···
 struct cpu_topology_map {
 	int	socket_id;
 	int	die_id;
+	int	cluster_id;
 	int	core_id;
 };
 

tools/perf/util/stat-display.c

···
 		snprintf(buf, sizeof(buf), "S%d-D%d-L%d-ID%d",
 			 id.socket, id.die, id.cache_lvl, id.cache);
 		break;
+	case AGGR_CLUSTER:
+		snprintf(buf, sizeof(buf), "S%d-D%d-CLS%d", id.socket, id.die, id.cluster);
+		break;
 	case AGGR_DIE:
 		snprintf(buf, sizeof(buf), "S%d-D%d", id.socket, id.die);
 		break;
···
 		fprintf(config->output, "S%d-D%d-L%d-ID%d%s%d%s",
 			id.socket, id.die, id.cache_lvl, id.cache, sep, aggr_nr, sep);
 		break;
+	case AGGR_CLUSTER:
+		fprintf(config->output, "S%d-D%d-CLS%d%s%d%s",
+			id.socket, id.die, id.cluster, sep, aggr_nr, sep);
+		break;
 	case AGGR_DIE:
 		fprintf(output, "S%d-D%d%s%d%s",
 			id.socket, id.die, sep, aggr_nr, sep);
···
 	case AGGR_CACHE:
 		fprintf(output, "\"cache\" : \"S%d-D%d-L%d-ID%d\", \"aggregate-number\" : %d, ",
 			id.socket, id.die, id.cache_lvl, id.cache, aggr_nr);
+		break;
+	case AGGR_CLUSTER:
+		fprintf(output, "\"cluster\" : \"S%d-D%d-CLS%d\", \"aggregate-number\" : %d, ",
+			id.socket, id.die, id.cluster, aggr_nr);
 		break;
 	case AGGR_DIE:
 		fprintf(output, "\"die\" : \"S%d-D%d\", \"aggregate-number\" : %d, ",
···
 	case AGGR_NODE:
 	case AGGR_SOCKET:
 	case AGGR_DIE:
+	case AGGR_CLUSTER:
 	case AGGR_CACHE:
 	case AGGR_CORE:
 		fprintf(output, "#%*s %-*s cpus",
···
 	switch (config->aggr_mode) {
 	case AGGR_CORE:
 	case AGGR_CACHE:
+	case AGGR_CLUSTER:
 	case AGGR_DIE:
 	case AGGR_SOCKET:
 	case AGGR_NODE:

tools/perf/util/stat.h

···
 	AGGR_GLOBAL,
 	AGGR_SOCKET,
 	AGGR_DIE,
+	AGGR_CLUSTER,
 	AGGR_CACHE,
 	AGGR_CORE,
 	AGGR_THREAD,