[PATCH V3 3/5] perf stat: Support per-die aggregation

From: kan . liang
Date: Tue Jun 04 2019 - 18:55:22 EST

From: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>

It is useful to aggregate counts per die. E.g. Uncore becomes die-scope
on Xeon Cascade Lake-AP.

Introduce a new option "--per-die" to support per-die aggregation.

The global id for each core has been changed to socket + die id + core
id. The global id for each die is socket + die id.

Add die information for per-core aggregation. The output of per-core
aggregation will be changed from "S0-C0" to "S0-D0-C0". Any scripts
which rely on the output format of per-core aggregation probably be
broken.

For perf stat record/report, there is no die information when processing
the old perf.data. The per-die result will be the same as per-socket.

Signed-off-by: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
---

Changes since V2:
- Fix the comments
- Drop "Please upgrade the perf tool." in warning
- Check core number in warning as well
- Fix a bug in cpu_map__get_core(). The cpu_map__get_die() is already
return the combination of socket + die id.

tools/perf/Documentation/perf-stat.txt | 10 ++++
tools/perf/builtin-stat.c | 88 +++++++++++++++++++++++++++++++---
tools/perf/util/cpumap.c | 57 +++++++++++++++++++---
tools/perf/util/cpumap.h | 9 +++-
tools/perf/util/stat-display.c | 29 +++++++++--
tools/perf/util/stat-shadow.c | 1 +
tools/perf/util/stat.c | 1 +
tools/perf/util/stat.h | 1 +
8 files changed, 178 insertions(+), 18 deletions(-)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 1e312c2..930c51c0 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -200,6 +200,13 @@ use --per-socket in addition to -a. (system-wide). The output includes the
socket number and the number of online processors on that socket. This is
useful to gauge the amount of aggregation.

+--per-die::
+Aggregate counts per processor die for system-wide mode measurements. This
+is a useful mode to detect imbalance between dies. To enable this mode,
+use --per-die in addition to -a. (system-wide). The output includes the
+die number and the number of online processors on that die. This is
+useful to gauge the amount of aggregation.
+
--per-core::
Aggregate counts per physical processor for system-wide mode measurements. This
is a useful mode to detect imbalance between physical cores. To enable this mode,
@@ -239,6 +246,9 @@ Input file name.
--per-socket::
Aggregate counts per processor socket for system-wide mode measurements.

+--per-die::
+Aggregate counts per processor die for system-wide mode measurements.
+
--per-core::
Aggregate counts per physical processor for system-wide mode measurements.

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 24b8e69..b367021 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -777,6 +777,8 @@ static struct option stat_options[] = {
"stop workload and print counts after a timeout period in ms (>= 10ms)"),
OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
"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-core", &stat_config.aggr_mode,
"aggregate counts per physical processor core", AGGR_CORE),
OPT_SET_UINT(0, "per-thread", &stat_config.aggr_mode,
@@ -801,6 +803,12 @@ static int perf_stat__get_socket(struct perf_stat_config *config __maybe_unused,
return cpu_map__get_socket(map, cpu, NULL);
}

+static int perf_stat__get_die(struct perf_stat_config *config __maybe_unused,
+ struct cpu_map *map, int cpu)
+{
+ return cpu_map__get_die(map, cpu, NULL);
+}
+
static int perf_stat__get_core(struct perf_stat_config *config __maybe_unused,
struct cpu_map *map, int cpu)
{
@@ -841,6 +849,12 @@ static int perf_stat__get_socket_cached(struct perf_stat_config *config,
return perf_stat__get_aggr(config, perf_stat__get_socket, map, idx);
}

+static int perf_stat__get_die_cached(struct perf_stat_config *config,
+ struct cpu_map *map, int idx)
+{
+ return perf_stat__get_aggr(config, perf_stat__get_die, map, idx);
+}
+
static int perf_stat__get_core_cached(struct perf_stat_config *config,
struct cpu_map *map, int idx)
{
@@ -871,6 +885,13 @@ static int perf_stat_init_aggr_mode(void)
}
stat_config.aggr_get_id = perf_stat__get_socket_cached;
break;
+ case AGGR_DIE:
+ if (cpu_map__build_die_map(evsel_list->cpus, &stat_config.aggr_map)) {
+ perror("cannot build die map");
+ return -1;
+ }
+ stat_config.aggr_get_id = perf_stat__get_die_cached;
+ break;
case AGGR_CORE:
if (cpu_map__build_core_map(evsel_list->cpus, &stat_config.aggr_map)) {
perror("cannot build core map");
@@ -936,21 +957,56 @@ static int perf_env__get_socket(struct cpu_map *map, int idx, void *data)
return cpu == -1 ? -1 : env->cpu[cpu].socket_id;
}

+static int perf_env__get_die(struct cpu_map *map, int idx, void *data)
+{
+ struct perf_env *env = data;
+ int die = -1, cpu = perf_env__get_cpu(env, map, idx);
+
+ if (cpu != -1) {
+ /*
+ * Encode socket in bit range 15:8
+ * die_id is relative to socket,
+ * we need a global id. So we combine
+ * socket + die id
+ */
+ if (WARN_ONCE(env->cpu[cpu].socket_id >> 8, "The socket id number is too big.\n"))
+ return -1;
+
+ if (WARN_ONCE(env->cpu[cpu].die_id >> 8, "The die id number is too big.\n"))
+ return -1;
+
+ die = (env->cpu[cpu].socket_id << 8) |
+ (env->cpu[cpu].die_id & 0xff);
+ }
+
+ return die;
+}
+
static int perf_env__get_core(struct cpu_map *map, int idx, void *data)
{
struct perf_env *env = data;
int core = -1, cpu = perf_env__get_cpu(env, map, idx);

if (cpu != -1) {
- int socket_id = env->cpu[cpu].socket_id;
-
/*
- * Encode socket in upper 16 bits
- * core_id is relative to socket, and
+ * Encode socket in bit range 31:24
+ * encode die id in bit range 23:16
+ * core_id is relative to socket and die,
* we need a global id. So we combine
- * socket + core id.
+ * socket + die id + core id
*/
- core = (socket_id << 16) | (env->cpu[cpu].core_id & 0xffff);
+ if (WARN_ONCE(env->cpu[cpu].socket_id >> 8, "The socket id number is too big.\n"))
+ return -1;
+
+ if (WARN_ONCE(env->cpu[cpu].die_id >> 8, "The die id number is too big.\n"))
+ return -1;
+
+ if (WARN_ONCE(env->cpu[cpu].core_id >> 16, "The core id number is too big.\n"))
+ return -1;
+
+ core = (env->cpu[cpu].socket_id << 24) |
+ (env->cpu[cpu].die_id << 16) |
+ (env->cpu[cpu].core_id & 0xffff);
}

return core;
@@ -962,6 +1018,12 @@ static int perf_env__build_socket_map(struct perf_env *env, struct cpu_map *cpus
return cpu_map__build_map(cpus, sockp, perf_env__get_socket, env);
}

+static int perf_env__build_die_map(struct perf_env *env, struct cpu_map *cpus,
+ struct cpu_map **diep)
+{
+ return cpu_map__build_map(cpus, diep, perf_env__get_die, env);
+}
+
static int perf_env__build_core_map(struct perf_env *env, struct cpu_map *cpus,
struct cpu_map **corep)
{
@@ -973,6 +1035,11 @@ static int perf_stat__get_socket_file(struct perf_stat_config *config __maybe_un
{
return perf_env__get_socket(map, idx, &perf_stat.session->header.env);
}
+static int perf_stat__get_die_file(struct perf_stat_config *config __maybe_unused,
+ struct cpu_map *map, int idx)
+{
+ return perf_env__get_die(map, idx, &perf_stat.session->header.env);
+}

static int perf_stat__get_core_file(struct perf_stat_config *config __maybe_unused,
struct cpu_map *map, int idx)
@@ -992,6 +1059,13 @@ static int perf_stat_init_aggr_mode_file(struct perf_stat *st)
}
stat_config.aggr_get_id = perf_stat__get_socket_file;
break;
+ case AGGR_DIE:
+ if (perf_env__build_die_map(env, evsel_list->cpus, &stat_config.aggr_map)) {
+ perror("cannot build die map");
+ return -1;
+ }
+ stat_config.aggr_get_id = perf_stat__get_die_file;
+ break;
case AGGR_CORE:
if (perf_env__build_core_map(env, evsel_list->cpus, &stat_config.aggr_map)) {
perror("cannot build core map");
@@ -1542,6 +1616,8 @@ static int __cmd_report(int argc, const char **argv)
OPT_STRING('i', "input", &input_name, "file", "input file name"),
OPT_SET_UINT(0, "per-socket", &perf_stat.aggr_mode,
"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-core", &perf_stat.aggr_mode,
"aggregate counts per physical processor core", AGGR_CORE),
OPT_SET_UINT('A', "no-aggr", &perf_stat.aggr_mode,
diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c
index 7db1365..b56646c 100644
--- a/tools/perf/util/cpumap.c
+++ b/tools/perf/util/cpumap.c
@@ -380,6 +380,39 @@ int cpu_map__get_die_id(int cpu)
return ret ?: value;
}

+int cpu_map__get_die(struct cpu_map *map, int idx, void *data)
+{
+ int cpu, die, s;
+
+ if (idx > map->nr)
+ return -1;
+
+ cpu = map->map[idx];
+
+ die = cpu_map__get_die_id(cpu);
+ /* There is no die_id on legacy system. */
+ if (die == -1)
+ die = 0;
+
+ s = cpu_map__get_socket(map, idx, data);
+ if (s == -1)
+ return -1;
+
+ /*
+ * Encode socket in bit range 15:8
+ * die_id is relative to socket, and
+ * we need a global id. So we combine
+ * socket + die id
+ */
+ if (WARN_ONCE(die >> 8, "The die id number is too big.\n"))
+ return -1;
+
+ if (WARN_ONCE(s >> 8, "The socket id number is too big.\n"))
+ return -1;
+
+ return (s << 8) | (die & 0xff);
+}
+
int cpu_map__get_core_id(int cpu)
{
int value, ret = cpu__get_topology_int(cpu, "core_id", &value);
@@ -388,7 +421,7 @@ int cpu_map__get_core_id(int cpu)

int cpu_map__get_core(struct cpu_map *map, int idx, void *data)
{
- int cpu, s;
+ int cpu, s_die;

if (idx > map->nr)
return -1;
@@ -397,17 +430,22 @@ int cpu_map__get_core(struct cpu_map *map, int idx, void *data)

cpu = cpu_map__get_core_id(cpu);

- s = cpu_map__get_socket(map, idx, data);
- if (s == -1)
+ /* s_die is the combination of socket + die id */
+ s_die = cpu_map__get_die(map, idx, data);
+ if (s_die == -1)
return -1;

/*
- * encode socket in upper 16 bits
- * core_id is relative to socket, and
+ * encode socket in bit range 31:24
+ * encode die id in bit range 23:16
+ * core_id is relative to socket and die,
* we need a global id. So we combine
- * socket+ core id
+ * socket + die id + core id
*/
- return (s << 16) | (cpu & 0xffff);
+ if (WARN_ONCE(cpu >> 16, "The core id number is too big.\n"))
+ return -1;
+
+ return (s_die << 16) | (cpu & 0xffff);
}

int cpu_map__build_socket_map(struct cpu_map *cpus, struct cpu_map **sockp)
@@ -415,6 +453,11 @@ int cpu_map__build_socket_map(struct cpu_map *cpus, struct cpu_map **sockp)
return cpu_map__build_map(cpus, sockp, cpu_map__get_socket, NULL);
}

+int cpu_map__build_die_map(struct cpu_map *cpus, struct cpu_map **diep)
+{
+ return cpu_map__build_map(cpus, diep, cpu_map__get_die, NULL);
+}
+
int cpu_map__build_core_map(struct cpu_map *cpus, struct cpu_map **corep)
{
return cpu_map__build_map(cpus, corep, cpu_map__get_core, NULL);
diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h
index 6762ff9..1265f0e 100644
--- a/tools/perf/util/cpumap.h
+++ b/tools/perf/util/cpumap.h
@@ -26,9 +26,11 @@ size_t cpu_map__fprintf(struct cpu_map *map, FILE *fp);
int cpu_map__get_socket_id(int cpu);
int cpu_map__get_socket(struct cpu_map *map, int idx, void *data);
int cpu_map__get_die_id(int cpu);
+int cpu_map__get_die(struct cpu_map *map, int idx, void *data);
int cpu_map__get_core_id(int cpu);
int cpu_map__get_core(struct cpu_map *map, int idx, void *data);
int cpu_map__build_socket_map(struct cpu_map *cpus, struct cpu_map **sockp);
+int cpu_map__build_die_map(struct cpu_map *cpus, struct cpu_map **diep);
int cpu_map__build_core_map(struct cpu_map *cpus, struct cpu_map **corep);
const struct cpu_map *cpu_map__online(void); /* thread unsafe */

@@ -44,7 +46,12 @@ static inline int cpu_map__socket(struct cpu_map *sock, int s)

static inline int cpu_map__id_to_socket(int id)
{
- return id >> 16;
+ return id >> 24;
+}
+
+static inline int cpu_map__id_to_die(int id)
+{
+ return (id >> 16) & 0xff;
}

static inline int cpu_map__id_to_cpu(int id)
diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
index 4c53bae..a6b9de3 100644
--- a/tools/perf/util/stat-display.c
+++ b/tools/perf/util/stat-display.c
@@ -69,8 +69,9 @@ static void aggr_printout(struct perf_stat_config *config,
{
switch (config->aggr_mode) {
case AGGR_CORE:
- fprintf(config->output, "S%d-C%*d%s%*d%s",
+ fprintf(config->output, "S%d-D%d-C%*d%s%*d%s",
cpu_map__id_to_socket(id),
+ cpu_map__id_to_die(id),
config->csv_output ? 0 : -8,
cpu_map__id_to_cpu(id),
config->csv_sep,
@@ -78,6 +79,16 @@ static void aggr_printout(struct perf_stat_config *config,
nr,
config->csv_sep);
break;
+ case AGGR_DIE:
+ fprintf(config->output, "S%d-D%*d%s%*d%s",
+ cpu_map__id_to_socket(id << 16),
+ config->csv_output ? 0 : -8,
+ cpu_map__id_to_die(id << 16),
+ config->csv_sep,
+ config->csv_output ? 0 : 4,
+ nr,
+ config->csv_sep);
+ break;
case AGGR_SOCKET:
fprintf(config->output, "S%*d%s%*d%s",
config->csv_output ? 0 : -5,
@@ -89,8 +100,9 @@ static void aggr_printout(struct perf_stat_config *config,
break;
case AGGR_NONE:
if (evsel->percore) {
- fprintf(config->output, "S%d-C%*d%s",
+ fprintf(config->output, "S%d-D%d-C%*d%s",
cpu_map__id_to_socket(id),
+ cpu_map__id_to_die(id),
config->csv_output ? 0 : -5,
cpu_map__id_to_cpu(id), config->csv_sep);
} else {
@@ -407,6 +419,7 @@ static void printout(struct perf_stat_config *config, int id, int nr,
[AGGR_THREAD] = 1,
[AGGR_NONE] = 1,
[AGGR_SOCKET] = 2,
+ [AGGR_DIE] = 2,
[AGGR_CORE] = 2,
};

@@ -879,7 +892,8 @@ static void print_no_aggr_metric(struct perf_stat_config *config,
}

static int aggr_header_lens[] = {
- [AGGR_CORE] = 18,
+ [AGGR_CORE] = 24,
+ [AGGR_DIE] = 18,
[AGGR_SOCKET] = 12,
[AGGR_NONE] = 6,
[AGGR_THREAD] = 24,
@@ -888,6 +902,7 @@ static int aggr_header_lens[] = {

static const char *aggr_header_csv[] = {
[AGGR_CORE] = "core,cpus,",
+ [AGGR_DIE] = "die,cpus",
[AGGR_SOCKET] = "socket,cpus",
[AGGR_NONE] = "cpu,",
[AGGR_THREAD] = "comm-pid,",
@@ -954,8 +969,13 @@ static void print_interval(struct perf_stat_config *config,
if (!metric_only)
fprintf(output, " counts %*s events\n", unit_width, "unit");
break;
+ case AGGR_DIE:
+ fprintf(output, "# time die cpus");
+ if (!metric_only)
+ fprintf(output, " counts %*s events\n", unit_width, "unit");
+ break;
case AGGR_CORE:
- fprintf(output, "# time core cpus");
+ fprintf(output, "# time core cpus");
if (!metric_only)
fprintf(output, " counts %*s events\n", unit_width, "unit");
break;
@@ -1165,6 +1185,7 @@ perf_evlist__print_counters(struct perf_evlist *evlist,

switch (config->aggr_mode) {
case AGGR_CORE:
+ case AGGR_DIE:
case AGGR_SOCKET:
print_aggr(config, evlist, prefix);
break;
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index 83d8094..027b09a 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -12,6 +12,7 @@
/*
* AGGR_GLOBAL: Use CPU 0
* AGGR_SOCKET: Use first CPU of socket
+ * AGGR_DIE: Use first CPU of die
* AGGR_CORE: Use first CPU of core
* AGGR_NONE: Use matching CPU
* AGGR_THREAD: Not supported?
diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
index c3115d9..d91fe75 100644
--- a/tools/perf/util/stat.c
+++ b/tools/perf/util/stat.c
@@ -272,6 +272,7 @@ process_counter_values(struct perf_stat_config *config, struct perf_evsel *evsel
switch (config->aggr_mode) {
case AGGR_THREAD:
case AGGR_CORE:
+ case AGGR_DIE:
case AGGR_SOCKET:
case AGGR_NONE:
if (!evsel->snapshot)
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index 2f9c915..7032dd1 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -44,6 +44,7 @@ enum aggr_mode {
AGGR_NONE,
AGGR_GLOBAL,
AGGR_SOCKET,
+ AGGR_DIE,
AGGR_CORE,
AGGR_THREAD,
AGGR_UNSET,
--
2.7.4