summaryrefslogtreecommitdiff
path: root/tools/perf/builtin-stat.c
diff options
context:
space:
mode:
Diffstat (limited to 'tools/perf/builtin-stat.c')
-rw-r--r--tools/perf/builtin-stat.c1327
1 files changed, 742 insertions, 585 deletions
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 07b48f6df48e..ab40d85fb125 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -46,6 +46,7 @@
#include "util/parse-events.h"
#include "util/pmus.h"
#include "util/pmu.h"
+#include "util/tool_pmu.h"
#include "util/event.h"
#include "util/evlist.h"
#include "util/evsel.h"
@@ -70,8 +71,10 @@
#include "util/bpf_counter.h"
#include "util/iostat.h"
#include "util/util.h"
+#include "util/intel-tpebs.h"
#include "asm/bug.h"
+#include <linux/list_sort.h>
#include <linux/time64.h>
#include <linux/zalloc.h>
#include <api/fs/fs.h>
@@ -94,8 +97,17 @@
#include <perf/evlist.h>
#include <internal/threadmap.h>
+#ifdef HAVE_BPF_SKEL
+#include "util/bpf_skel/bperf_cgroup.h"
+#endif
+
#define DEFAULT_SEPARATOR " "
-#define FREEZE_ON_SMI_PATH "devices/cpu/freeze_on_smi"
+#define FREEZE_ON_SMI_PATH "bus/event_source/devices/cpu/freeze_on_smi"
+
+struct rusage_stats {
+ struct stats ru_utime_usec_stat;
+ struct stats ru_stime_usec_stat;
+};
static void print_counters(struct timespec *ts, int argc, const char **argv);
@@ -106,11 +118,7 @@ static struct parse_events_option_args parse_events_option_args = {
static bool all_counters_use_bpf = true;
-static struct target target = {
- .uid = UINT_MAX,
-};
-
-#define METRIC_ONLY_LEN 20
+static struct target target;
static volatile sig_atomic_t child_pid = -1;
static int detailed_run = 0;
@@ -130,6 +138,7 @@ static bool interval_count;
static const char *output_name;
static int output_fd;
static char *metrics;
+static struct rusage_stats ru_stats;
struct perf_stat {
bool record;
@@ -149,39 +158,33 @@ static struct perf_stat perf_stat;
static volatile sig_atomic_t done = 0;
-static struct perf_stat_config stat_config = {
- .aggr_mode = AGGR_GLOBAL,
- .aggr_level = MAX_CACHE_LVL + 1,
- .scale = true,
- .unit_width = 4, /* strlen("unit") */
- .run_count = 1,
- .metric_only_len = METRIC_ONLY_LEN,
- .walltime_nsecs_stats = &walltime_nsecs_stats,
- .ru_stats = &ru_stats,
- .big_num = true,
- .ctl_fd = -1,
- .ctl_fd_ack = -1,
- .iostat_run = false,
+/* Options set from the command line. */
+struct opt_aggr_mode {
+ bool node, socket, die, cluster, cache, core, thread, no_aggr;
};
-static bool cpus_map_matched(struct evsel *a, struct evsel *b)
-{
- if (!a->core.cpus && !b->core.cpus)
- return true;
-
- if (!a->core.cpus || !b->core.cpus)
- return false;
-
- if (perf_cpu_map__nr(a->core.cpus) != perf_cpu_map__nr(b->core.cpus))
- return false;
-
- for (int i = 0; i < perf_cpu_map__nr(a->core.cpus); i++) {
- if (perf_cpu_map__cpu(a->core.cpus, i).cpu !=
- perf_cpu_map__cpu(b->core.cpus, i).cpu)
- return false;
- }
-
- return true;
+/* Turn command line option into most generic aggregation mode setting. */
+static enum aggr_mode opt_aggr_mode_to_aggr_mode(struct opt_aggr_mode *opt_mode)
+{
+ enum aggr_mode mode = AGGR_GLOBAL;
+
+ if (opt_mode->node)
+ mode = AGGR_NODE;
+ if (opt_mode->socket)
+ mode = AGGR_SOCKET;
+ if (opt_mode->die)
+ mode = AGGR_DIE;
+ if (opt_mode->cluster)
+ mode = AGGR_CLUSTER;
+ if (opt_mode->cache)
+ mode = AGGR_CACHE;
+ if (opt_mode->core)
+ mode = AGGR_CORE;
+ if (opt_mode->thread)
+ mode = AGGR_THREAD;
+ if (opt_mode->no_aggr)
+ mode = AGGR_NONE;
+ return mode;
}
static void evlist__check_cpu_maps(struct evlist *evlist)
@@ -194,7 +197,7 @@ static void evlist__check_cpu_maps(struct evlist *evlist)
/* Check that leader matches cpus with each member. */
if (leader == evsel)
continue;
- if (cpus_map_matched(leader, evsel))
+ if (perf_cpu_map__equal(leader->core.cpus, evsel->core.cpus))
continue;
/* If there's mismatch disable the group and warn user. */
@@ -236,10 +239,10 @@ static inline void diff_timespec(struct timespec *r, struct timespec *a,
static void perf_stat__reset_stats(void)
{
evlist__reset_stats(evsel_list);
- perf_stat__reset_shadow_stats();
+ memset(stat_config.walltime_nsecs_stats, 0, sizeof(*stat_config.walltime_nsecs_stats));
}
-static int process_synthesized_event(struct perf_tool *tool __maybe_unused,
+static int process_synthesized_event(const struct perf_tool *tool __maybe_unused,
union perf_event *event,
struct perf_sample *sample __maybe_unused,
struct machine *machine __maybe_unused)
@@ -275,45 +278,48 @@ static int evsel__write_stat_event(struct evsel *counter, int cpu_map_idx, u32 t
process_synthesized_event, NULL);
}
-static int read_single_counter(struct evsel *counter, int cpu_map_idx,
- int thread, struct timespec *rs)
+static int read_single_counter(struct evsel *counter, int cpu_map_idx, int thread)
{
- switch(counter->tool_event) {
- case PERF_TOOL_DURATION_TIME: {
- u64 val = rs->tv_nsec + rs->tv_sec*1000000000ULL;
- struct perf_counts_values *count =
- perf_counts(counter->counts, cpu_map_idx, thread);
- count->ena = count->run = val;
- count->val = val;
- return 0;
- }
- case PERF_TOOL_USER_TIME:
- case PERF_TOOL_SYSTEM_TIME: {
- u64 val;
- struct perf_counts_values *count =
- perf_counts(counter->counts, cpu_map_idx, thread);
- if (counter->tool_event == PERF_TOOL_USER_TIME)
- val = ru_stats.ru_utime_usec_stat.mean;
- else
- val = ru_stats.ru_stime_usec_stat.mean;
- count->ena = count->run = val;
- count->val = val;
- return 0;
+ int err = evsel__read_counter(counter, cpu_map_idx, thread);
+
+ /*
+ * Reading user and system time will fail when the process
+ * terminates. Use the wait4 values in that case.
+ */
+ if (err && cpu_map_idx == 0 &&
+ (evsel__tool_event(counter) == TOOL_PMU__EVENT_USER_TIME ||
+ evsel__tool_event(counter) == TOOL_PMU__EVENT_SYSTEM_TIME)) {
+ struct perf_counts_values *count =
+ perf_counts(counter->counts, cpu_map_idx, thread);
+ struct perf_counts_values *old_count = NULL;
+ u64 val;
+
+ if (counter->prev_raw_counts)
+ old_count = perf_counts(counter->prev_raw_counts, cpu_map_idx, thread);
+
+ if (evsel__tool_event(counter) == TOOL_PMU__EVENT_USER_TIME)
+ val = ru_stats.ru_utime_usec_stat.mean;
+ else
+ val = ru_stats.ru_stime_usec_stat.mean;
+
+ count->val = val;
+ if (old_count) {
+ count->run = old_count->run + 1;
+ count->ena = old_count->ena + 1;
+ } else {
+ count->run++;
+ count->ena++;
}
- default:
- case PERF_TOOL_NONE:
- return evsel__read_counter(counter, cpu_map_idx, thread);
- case PERF_TOOL_MAX:
- /* This should never be reached */
- return 0;
+ return 0;
}
+ return err;
}
/*
* Read out the results of a single counter:
* do not aggregate counts across CPUs in system-wide mode
*/
-static int read_counter_cpu(struct evsel *counter, struct timespec *rs, int cpu_map_idx)
+static int read_counter_cpu(struct evsel *counter, int cpu_map_idx)
{
int nthreads = perf_thread_map__nr(evsel_list->core.threads);
int thread;
@@ -331,7 +337,7 @@ static int read_counter_cpu(struct evsel *counter, struct timespec *rs, int cpu_
* (via evsel__read_counter()) and sets their count->loaded.
*/
if (!perf_counts__is_loaded(counter->counts, cpu_map_idx, thread) &&
- read_single_counter(counter, cpu_map_idx, thread, rs)) {
+ read_single_counter(counter, cpu_map_idx, thread)) {
counter->counts->scaled = -1;
perf_counts(counter->counts, cpu_map_idx, thread)->ena = 0;
perf_counts(counter->counts, cpu_map_idx, thread)->run = 0;
@@ -360,7 +366,7 @@ static int read_counter_cpu(struct evsel *counter, struct timespec *rs, int cpu_
return 0;
}
-static int read_affinity_counters(struct timespec *rs)
+static int read_counters_with_affinity(void)
{
struct evlist_cpu_iterator evlist_cpu_itr;
struct affinity saved_affinity, *affinity;
@@ -381,10 +387,11 @@ static int read_affinity_counters(struct timespec *rs)
if (evsel__is_bpf(counter))
continue;
- if (!counter->err) {
- counter->err = read_counter_cpu(counter, rs,
- evlist_cpu_itr.cpu_map_idx);
- }
+ if (evsel__is_tool(counter))
+ continue;
+
+ if (!counter->err)
+ counter->err = read_counter_cpu(counter, evlist_cpu_itr.cpu_map_idx);
}
if (affinity)
affinity__cleanup(&saved_affinity);
@@ -408,16 +415,46 @@ static int read_bpf_map_counters(void)
return 0;
}
-static int read_counters(struct timespec *rs)
+static int read_tool_counters(void)
{
- if (!stat_config.stop_read_counter) {
- if (read_bpf_map_counters() ||
- read_affinity_counters(rs))
- return -1;
+ struct evsel *counter;
+
+ evlist__for_each_entry(evsel_list, counter) {
+ int idx;
+
+ if (!evsel__is_tool(counter))
+ continue;
+
+ perf_cpu_map__for_each_idx(idx, counter->core.cpus) {
+ if (!counter->err)
+ counter->err = read_counter_cpu(counter, idx);
+ }
}
return 0;
}
+static int read_counters(void)
+{
+ int ret;
+
+ if (stat_config.stop_read_counter)
+ return 0;
+
+ // Read all BPF counters first.
+ ret = read_bpf_map_counters();
+ if (ret)
+ return ret;
+
+ // Read non-BPF and non-tool counters next.
+ ret = read_counters_with_affinity();
+ if (ret)
+ return ret;
+
+ // Read the tool counters last. This way the duration_time counter
+ // should always be greater than any other counter's enabled time.
+ return read_tool_counters();
+}
+
static void process_counters(void)
{
struct evsel *counter;
@@ -443,7 +480,7 @@ static void process_interval(void)
evlist__reset_aggr_stats(evsel_list);
- if (read_counters(&rs) == 0)
+ if (read_counters() == 0)
process_counters();
if (STAT_RECORD) {
@@ -451,8 +488,8 @@ static void process_interval(void)
pr_err("failed to write stat round event\n");
}
- init_stats(&walltime_nsecs_stats);
- update_stats(&walltime_nsecs_stats, stat_config.interval * 1000000ULL);
+ init_stats(stat_config.walltime_nsecs_stats);
+ update_stats(stat_config.walltime_nsecs_stats, stat_config.interval * 1000000ULL);
print_counters(&rs, 0, NULL);
}
@@ -627,39 +664,34 @@ static int dispatch_events(bool forks, int timeout, int interval, int *times)
enum counter_recovery {
COUNTER_SKIP,
COUNTER_RETRY,
- COUNTER_FATAL,
};
-static enum counter_recovery stat_handle_error(struct evsel *counter)
+static enum counter_recovery stat_handle_error(struct evsel *counter, int err)
{
char msg[BUFSIZ];
+
+ assert(!counter->supported);
+
/*
* PPC returns ENXIO for HW counters until 2.6.37
* (behavior changed with commit b0a873e).
*/
- if (errno == EINVAL || errno == ENOSYS ||
- errno == ENOENT || errno == EOPNOTSUPP ||
- errno == ENXIO) {
- if (verbose > 0)
- ui__warning("%s event is not supported by the kernel.\n",
- evsel__name(counter));
- counter->supported = false;
- /*
- * errored is a sticky flag that means one of the counter's
- * cpu event had a problem and needs to be reexamined.
- */
- counter->errored = true;
-
- if ((evsel__leader(counter) != counter) ||
- !(counter->core.leader->nr_members > 1))
- return COUNTER_SKIP;
- } else if (evsel__fallback(counter, errno, msg, sizeof(msg))) {
+ if (err == EINVAL || err == ENOSYS || err == ENOENT || err == ENXIO) {
+ if (verbose > 0) {
+ evsel__open_strerror(counter, &target, err, msg, sizeof(msg));
+ ui__warning("%s event is not supported by the kernel.\n%s\n",
+ evsel__name(counter), msg);
+ }
+ return COUNTER_SKIP;
+ }
+ if (evsel__fallback(counter, &target, err, msg, sizeof(msg))) {
if (verbose > 0)
ui__warning("%s\n", msg);
+ counter->supported = true;
return COUNTER_RETRY;
- } else if (target__has_per_thread(&target) &&
- evsel_list->core.threads &&
- evsel_list->core.threads->err_thread != -1) {
+ }
+ if (target__has_per_thread(&target) && err != EOPNOTSUPP &&
+ evsel_list->core.threads && evsel_list->core.threads->err_thread != -1) {
/*
* For global --per-thread case, skip current
* error thread.
@@ -667,23 +699,85 @@ static enum counter_recovery stat_handle_error(struct evsel *counter)
if (!thread_map__remove(evsel_list->core.threads,
evsel_list->core.threads->err_thread)) {
evsel_list->core.threads->err_thread = -1;
+ counter->supported = true;
return COUNTER_RETRY;
}
- } else if (counter->skippable) {
- if (verbose > 0)
- ui__warning("skipping event %s that kernel failed to open .\n",
- evsel__name(counter));
- counter->supported = false;
- counter->errored = true;
- return COUNTER_SKIP;
}
+ if (verbose > 0) {
+ evsel__open_strerror(counter, &target, err, msg, sizeof(msg));
+ ui__warning(err == EOPNOTSUPP
+ ? "%s event is not supported by the kernel.\n%s\n"
+ : "skipping event %s that kernel failed to open.\n%s\n",
+ evsel__name(counter), msg);
+ }
+ return COUNTER_SKIP;
+}
- evsel__open_strerror(counter, &target, errno, msg, sizeof(msg));
- ui__error("%s\n", msg);
+static int create_perf_stat_counter(struct evsel *evsel,
+ struct perf_stat_config *config,
+ int cpu_map_idx)
+{
+ struct perf_event_attr *attr = &evsel->core.attr;
+ struct evsel *leader = evsel__leader(evsel);
- if (child_pid != -1)
- kill(child_pid, SIGTERM);
- return COUNTER_FATAL;
+ /* Reset supported flag as creating a stat counter is retried. */
+ attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED |
+ PERF_FORMAT_TOTAL_TIME_RUNNING;
+
+ /*
+ * The event is part of non trivial group, let's enable
+ * the group read (for leader) and ID retrieval for all
+ * members.
+ */
+ if (leader->core.nr_members > 1)
+ attr->read_format |= PERF_FORMAT_ID|PERF_FORMAT_GROUP;
+
+ attr->inherit = !config->no_inherit && list_empty(&evsel->bpf_counter_list);
+
+ /*
+ * Some events get initialized with sample_(period/type) set,
+ * like tracepoints. Clear it up for counting.
+ */
+ attr->sample_period = 0;
+
+ if (config->identifier)
+ attr->sample_type = PERF_SAMPLE_IDENTIFIER;
+
+ if (config->all_user) {
+ attr->exclude_kernel = 1;
+ attr->exclude_user = 0;
+ }
+
+ if (config->all_kernel) {
+ attr->exclude_kernel = 0;
+ attr->exclude_user = 1;
+ }
+
+ /*
+ * Disabling all counters initially, they will be enabled
+ * either manually by us or by kernel via enable_on_exec
+ * set later.
+ */
+ if (evsel__is_group_leader(evsel)) {
+ attr->disabled = 1;
+
+ if (target__enable_on_exec(&target))
+ attr->enable_on_exec = 1;
+ }
+
+ return evsel__open_per_cpu_and_thread(evsel, evsel__cpus(evsel), cpu_map_idx,
+ evsel->core.threads);
+}
+
+static void update_rusage_stats(const struct rusage *rusage)
+{
+ const u64 us_to_ns = 1000;
+ const u64 s_to_ns = 1000000000;
+
+ update_stats(&ru_stats.ru_utime_usec_stat,
+ (rusage->ru_utime.tv_usec * us_to_ns + rusage->ru_utime.tv_sec * s_to_ns));
+ update_stats(&ru_stats.ru_stime_usec_stat,
+ (rusage->ru_stime.tv_usec * us_to_ns + rusage->ru_stime.tv_sec * s_to_ns));
}
static int __run_perf_stat(int argc, const char **argv, int run_idx)
@@ -700,8 +794,8 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
bool is_pipe = STAT_RECORD ? perf_stat.data.is_pipe : false;
struct evlist_cpu_iterator evlist_cpu_itr;
struct affinity saved_affinity, *affinity = NULL;
- int err;
- bool second_pass = false;
+ int err, open_err = 0;
+ bool second_pass = false, has_supported_counters;
if (forks) {
if (evlist__prepare_workload(evsel_list, &target, argv, is_pipe, workload_exec_failed_signal) < 0) {
@@ -712,15 +806,19 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
}
if (!cpu_map__is_dummy(evsel_list->core.user_requested_cpus)) {
- if (affinity__setup(&saved_affinity) < 0)
- return -1;
+ if (affinity__setup(&saved_affinity) < 0) {
+ err = -1;
+ goto err_out;
+ }
affinity = &saved_affinity;
}
evlist__for_each_entry(evsel_list, counter) {
counter->reset_group = false;
- if (bpf_counter__load(counter, &target))
- return -1;
+ if (bpf_counter__load(counter, &target)) {
+ err = -1;
+ goto err_out;
+ }
if (!(evsel__is_bperf(counter)))
all_counters_use_bpf = false;
}
@@ -737,14 +835,17 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
if (target.use_bpf)
break;
- if (counter->reset_group || counter->errored)
+ if (counter->reset_group || !counter->supported)
continue;
if (evsel__is_bperf(counter))
continue;
-try_again:
- if (create_perf_stat_counter(counter, &stat_config, &target,
- evlist_cpu_itr.cpu_map_idx) < 0) {
+ while (true) {
+ if (create_perf_stat_counter(counter, &stat_config,
+ evlist_cpu_itr.cpu_map_idx) == 0)
+ break;
+
+ open_err = errno;
/*
* Weak group failed. We cannot just undo this here
* because earlier CPUs might be in group mode, and the kernel
@@ -752,28 +853,19 @@ try_again:
* it to later.
* Don't close here because we're in the wrong affinity.
*/
- if ((errno == EINVAL || errno == EBADF) &&
+ if ((open_err == EINVAL || open_err == EBADF) &&
evsel__leader(counter) != counter &&
counter->weak_group) {
evlist__reset_weak_group(evsel_list, counter, false);
assert(counter->reset_group);
+ counter->supported = true;
second_pass = true;
- continue;
- }
-
- switch (stat_handle_error(counter)) {
- case COUNTER_FATAL:
- return -1;
- case COUNTER_RETRY:
- goto try_again;
- case COUNTER_SKIP:
- continue;
- default:
break;
}
+ if (stat_handle_error(counter, open_err) != COUNTER_RETRY)
+ break;
}
- counter->supported = true;
}
if (second_pass) {
@@ -786,7 +878,7 @@ try_again:
evlist__for_each_cpu(evlist_cpu_itr, evsel_list, affinity) {
counter = evlist_cpu_itr.evsel;
- if (!counter->reset_group && !counter->errored)
+ if (!counter->reset_group && counter->supported)
continue;
perf_evsel__close_cpu(&counter->core, evlist_cpu_itr.cpu_map_idx);
@@ -797,43 +889,54 @@ try_again:
if (!counter->reset_group)
continue;
-try_again_reset:
- pr_debug2("reopening weak %s\n", evsel__name(counter));
- if (create_perf_stat_counter(counter, &stat_config, &target,
- evlist_cpu_itr.cpu_map_idx) < 0) {
-
- switch (stat_handle_error(counter)) {
- case COUNTER_FATAL:
- return -1;
- case COUNTER_RETRY:
- goto try_again_reset;
- case COUNTER_SKIP:
- continue;
- default:
+
+ while (true) {
+ pr_debug2("reopening weak %s\n", evsel__name(counter));
+ if (create_perf_stat_counter(counter, &stat_config,
+ evlist_cpu_itr.cpu_map_idx) == 0)
+ break;
+
+ open_err = errno;
+ if (stat_handle_error(counter, open_err) != COUNTER_RETRY)
break;
- }
}
- counter->supported = true;
}
}
affinity__cleanup(affinity);
+ affinity = NULL;
+ has_supported_counters = false;
evlist__for_each_entry(evsel_list, counter) {
if (!counter->supported) {
perf_evsel__free_fd(&counter->core);
continue;
}
+ has_supported_counters = true;
l = strlen(counter->unit);
if (l > stat_config.unit_width)
stat_config.unit_width = l;
if (evsel__should_store_id(counter) &&
- evsel__store_ids(counter, evsel_list))
- return -1;
+ evsel__store_ids(counter, evsel_list)) {
+ err = -1;
+ goto err_out;
+ }
+ }
+ if (!has_supported_counters && !stat_config.null_run) {
+ if (open_err) {
+ evsel__open_strerror(evlist__first(evsel_list), &target, open_err,
+ msg, sizeof(msg));
+ }
+ ui__error("No supported events found.\n%s\n", msg);
+
+ if (child_pid != -1)
+ kill(child_pid, SIGTERM);
+ err = -1;
+ goto err_out;
}
- if (evlist__apply_filters(evsel_list, &counter)) {
+ if (evlist__apply_filters(evsel_list, &counter, &target)) {
pr_err("failed to set filter \"%s\" on event %s with %d (%s)\n",
counter->filter, evsel__name(counter), errno,
str_error_r(errno, msg, sizeof(msg)));
@@ -851,20 +954,23 @@ try_again_reset:
}
if (err < 0)
- return err;
+ goto err_out;
err = perf_event__synthesize_stat_events(&stat_config, NULL, evsel_list,
process_synthesized_event, is_pipe);
if (err < 0)
- return err;
+ goto err_out;
+
}
if (target.initial_delay) {
pr_info(EVLIST_DISABLED_MSG);
} else {
err = enable_counters();
- if (err)
- return -1;
+ if (err) {
+ err = -1;
+ goto err_out;
+ }
}
/* Exec the command, if any */
@@ -874,8 +980,10 @@ try_again_reset:
if (target.initial_delay > 0) {
usleep(target.initial_delay * USEC_PER_MSEC);
err = enable_counters();
- if (err)
- return -1;
+ if (err) {
+ err = -1;
+ goto err_out;
+ }
pr_info(EVLIST_ENABLED_MSG);
}
@@ -895,13 +1003,24 @@ try_again_reset:
if (workload_exec_errno) {
const char *emsg = str_error_r(workload_exec_errno, msg, sizeof(msg));
pr_err("Workload failed: %s\n", emsg);
- return -1;
+ err = -1;
+ goto err_out;
}
- if (WIFSIGNALED(status))
+ if (WIFSIGNALED(status)) {
+ /*
+ * We want to indicate failure to stop a repeat run,
+ * hence negative. We want the value to be the exit code
+ * of perf, which for termination by a signal is 128
+ * plus the signal number.
+ */
+ err = 0 - (128 + WTERMSIG(status));
psignal(WTERMSIG(status), argv[0]);
+ } else {
+ err = WEXITSTATUS(status);
+ }
} else {
- status = dispatch_events(forks, timeout, interval, &times);
+ err = dispatch_events(forks, timeout, interval, &times);
}
disable_counters();
@@ -914,15 +1033,15 @@ try_again_reset:
if (interval && stat_config.summary) {
stat_config.interval = 0;
stat_config.stop_read_counter = true;
- init_stats(&walltime_nsecs_stats);
- update_stats(&walltime_nsecs_stats, t1 - t0);
+ init_stats(stat_config.walltime_nsecs_stats);
+ update_stats(stat_config.walltime_nsecs_stats, t1 - t0);
evlist__copy_prev_raw_counts(evsel_list);
evlist__reset_prev_raw_counts(evsel_list);
evlist__reset_aggr_stats(evsel_list);
} else {
- update_stats(&walltime_nsecs_stats, t1 - t0);
- update_rusage_stats(&ru_stats, &stat_config.ru_data);
+ update_stats(stat_config.walltime_nsecs_stats, t1 - t0);
+ update_rusage_stats(&stat_config.ru_data);
}
/*
@@ -931,7 +1050,7 @@ try_again_reset:
* avoid arbitrary skew, we must read all counters before closing any
* group leaders.
*/
- if (read_counters(&(struct timespec) { .tv_nsec = t1-t0 }) == 0)
+ if (read_counters() == 0)
process_counters();
/*
@@ -941,9 +1060,24 @@ try_again_reset:
if (!STAT_RECORD)
evlist__close(evsel_list);
- return WEXITSTATUS(status);
+ return err;
+
+err_out:
+ if (forks)
+ evlist__cancel_workload(evsel_list);
+
+ affinity__cleanup(affinity);
+ return err;
}
+/*
+ * Returns -1 for fatal errors which signifies to not continue
+ * when in repeat mode.
+ *
+ * Returns < -1 error codes when stat record is used. These
+ * result in the stat information being displayed, but writing
+ * to the file fails and is non fatal.
+ */
static int run_perf_stat(int argc, const char **argv, int run_idx)
{
int ret;
@@ -1024,16 +1158,6 @@ static void sig_atexit(void)
kill(getpid(), signr);
}
-void perf_stat__set_big_num(int set)
-{
- stat_config.big_num = (set != 0);
-}
-
-void perf_stat__set_no_csv_summary(int set)
-{
- stat_config.no_csv_summary = (set != 0);
-}
-
static int stat__set_big_num(const struct option *opt __maybe_unused,
const char *s __maybe_unused, int unset)
{
@@ -1116,7 +1240,7 @@ static int parse_cache_level(const struct option *opt,
int unset __maybe_unused)
{
int level;
- u32 *aggr_mode = (u32 *)opt->value;
+ struct opt_aggr_mode *opt_aggr_mode = (struct opt_aggr_mode *)opt->value;
u32 *aggr_level = (u32 *)opt->data;
/*
@@ -1155,152 +1279,11 @@ static int parse_cache_level(const struct option *opt,
return -EINVAL;
}
out:
- *aggr_mode = AGGR_CACHE;
+ opt_aggr_mode->cache = true;
*aggr_level = level;
return 0;
}
-static struct option stat_options[] = {
- OPT_BOOLEAN('T', "transaction", &transaction_run,
- "hardware transaction statistics"),
- OPT_CALLBACK('e', "event", &parse_events_option_args, "event",
- "event selector. use 'perf list' to list available events",
- parse_events_option),
- OPT_CALLBACK(0, "filter", &evsel_list, "filter",
- "event filter", parse_filter),
- OPT_BOOLEAN('i', "no-inherit", &stat_config.no_inherit,
- "child tasks do not inherit counters"),
- OPT_STRING('p', "pid", &target.pid, "pid",
- "stat events on existing process id"),
- OPT_STRING('t', "tid", &target.tid, "tid",
- "stat events on existing thread id"),
-#ifdef HAVE_BPF_SKEL
- OPT_STRING('b', "bpf-prog", &target.bpf_str, "bpf-prog-id",
- "stat events on existing bpf program id"),
- OPT_BOOLEAN(0, "bpf-counters", &target.use_bpf,
- "use bpf program to count events"),
- OPT_STRING(0, "bpf-attr-map", &target.attr_map, "attr-map-path",
- "path to perf_event_attr map"),
-#endif
- OPT_BOOLEAN('a', "all-cpus", &target.system_wide,
- "system-wide collection from all CPUs"),
- OPT_BOOLEAN(0, "scale", &stat_config.scale,
- "Use --no-scale to disable counter scaling for multiplexing"),
- OPT_INCR('v', "verbose", &verbose,
- "be more verbose (show counter open errors, etc)"),
- OPT_INTEGER('r', "repeat", &stat_config.run_count,
- "repeat command and print average + stddev (max: 100, forever: 0)"),
- OPT_BOOLEAN(0, "table", &stat_config.walltime_run_table,
- "display details about each run (only with -r option)"),
- OPT_BOOLEAN('n', "null", &stat_config.null_run,
- "null run - dont start any counters"),
- OPT_INCR('d', "detailed", &detailed_run,
- "detailed run - start a lot of events"),
- OPT_BOOLEAN('S', "sync", &sync_run,
- "call sync() before starting a run"),
- OPT_CALLBACK_NOOPT('B', "big-num", NULL, NULL,
- "print large numbers with thousands\' separators",
- stat__set_big_num),
- OPT_STRING('C', "cpu", &target.cpu_list, "cpu",
- "list of cpus to monitor in system-wide"),
- OPT_SET_UINT('A', "no-aggr", &stat_config.aggr_mode,
- "disable CPU count aggregation", AGGR_NONE),
- OPT_BOOLEAN(0, "no-merge", &stat_config.no_merge, "Do not merge identical named events"),
- OPT_BOOLEAN(0, "hybrid-merge", &stat_config.hybrid_merge,
- "Merge identical named hybrid events"),
- OPT_STRING('x', "field-separator", &stat_config.csv_sep, "separator",
- "print counts with custom separator"),
- OPT_BOOLEAN('j', "json-output", &stat_config.json_output,
- "print counts in JSON format"),
- OPT_CALLBACK('G', "cgroup", &evsel_list, "name",
- "monitor event in cgroup name only", parse_stat_cgroups),
- OPT_STRING(0, "for-each-cgroup", &stat_config.cgroup_list, "name",
- "expand events for each cgroup"),
- OPT_STRING('o', "output", &output_name, "file", "output file name"),
- OPT_BOOLEAN(0, "append", &append_file, "append to the output file"),
- OPT_INTEGER(0, "log-fd", &output_fd,
- "log output to fd, instead of stderr"),
- OPT_STRING(0, "pre", &pre_cmd, "command",
- "command to run prior to the measured command"),
- OPT_STRING(0, "post", &post_cmd, "command",
- "command to run after to the measured command"),
- OPT_UINTEGER('I', "interval-print", &stat_config.interval,
- "print counts at regular interval in ms "
- "(overhead is possible for values <= 100ms)"),
- OPT_INTEGER(0, "interval-count", &stat_config.times,
- "print counts for fixed number of times"),
- OPT_BOOLEAN(0, "interval-clear", &stat_config.interval_clear,
- "clear screen in between new interval"),
- OPT_UINTEGER(0, "timeout", &stat_config.timeout,
- "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_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),
- 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,
- "aggregate counts per thread", AGGR_THREAD),
- OPT_SET_UINT(0, "per-node", &stat_config.aggr_mode,
- "aggregate counts per numa node", AGGR_NODE),
- OPT_INTEGER('D', "delay", &target.initial_delay,
- "ms to wait before starting measurement after program start (-1: start with events disabled)"),
- OPT_CALLBACK_NOOPT(0, "metric-only", &stat_config.metric_only, NULL,
- "Only print computed metrics. No raw values", enable_metric_only),
- OPT_BOOLEAN(0, "metric-no-group", &stat_config.metric_no_group,
- "don't group metric events, impacts multiplexing"),
- OPT_BOOLEAN(0, "metric-no-merge", &stat_config.metric_no_merge,
- "don't try to share events between metrics in a group"),
- OPT_BOOLEAN(0, "metric-no-threshold", &stat_config.metric_no_threshold,
- "don't try to share events between metrics in a group "),
- OPT_BOOLEAN(0, "topdown", &topdown_run,
- "measure top-down statistics"),
- OPT_UINTEGER(0, "td-level", &stat_config.topdown_level,
- "Set the metrics level for the top-down statistics (0: max level)"),
- OPT_BOOLEAN(0, "smi-cost", &smi_cost,
- "measure SMI cost"),
- OPT_CALLBACK('M', "metrics", &evsel_list, "metric/metric group list",
- "monitor specified metrics or metric groups (separated by ,)",
- append_metric_groups),
- OPT_BOOLEAN_FLAG(0, "all-kernel", &stat_config.all_kernel,
- "Configure all used events to run in kernel space.",
- PARSE_OPT_EXCLUSIVE),
- OPT_BOOLEAN_FLAG(0, "all-user", &stat_config.all_user,
- "Configure all used events to run in user space.",
- PARSE_OPT_EXCLUSIVE),
- OPT_BOOLEAN(0, "percore-show-thread", &stat_config.percore_show_thread,
- "Use with 'percore' event qualifier to show the event "
- "counts of one hardware thread by sum up total hardware "
- "threads of same physical core"),
- OPT_BOOLEAN(0, "summary", &stat_config.summary,
- "print summary for interval mode"),
- OPT_BOOLEAN(0, "no-csv-summary", &stat_config.no_csv_summary,
- "don't print 'summary' for CSV summary output"),
- OPT_BOOLEAN(0, "quiet", &quiet,
- "don't print any output, messages or warnings (useful with record)"),
- OPT_CALLBACK(0, "cputype", &evsel_list, "hybrid cpu type",
- "Only enable events on applying cpu with this type "
- "for hybrid platform (e.g. core or atom)",
- parse_cputype),
-#ifdef HAVE_LIBPFM
- OPT_CALLBACK(0, "pfm-events", &evsel_list, "event",
- "libpfm4 event selector. use 'perf list' to list available events",
- parse_libpfm_events_option),
-#endif
- OPT_CALLBACK(0, "control", &stat_config, "fd:ctl-fd[,ack-fd] or fifo:ctl-fifo[,ack-fifo]",
- "Listen on ctl-fd descriptor for command to control measurement ('enable': enable events, 'disable': disable events).\n"
- "\t\t\t Optionally send control command completion ('ack\\n') to ack-fd descriptor.\n"
- "\t\t\t Alternatively, ctl-fifo / ack-fifo will be opened and used as ctl-fd / ack-fd.",
- parse_control_option),
- OPT_CALLBACK_OPTARG(0, "iostat", &evsel_list, &stat_config, "default",
- "measure I/O performance metrics provided by arch/platform",
- iostat_parse),
- OPT_END()
-};
-
/**
* Calculate the cache instance ID from the map in
* /sys/devices/system/cpu/cpuX/cache/indexY/shared_cpu_list
@@ -1316,10 +1299,9 @@ static int cpu__get_cache_id_from_map(struct perf_cpu cpu, char *map)
* be the first online CPU in the cache domain else use the
* first online CPU of the cache domain as the ID.
*/
- if (perf_cpu_map__empty(cpu_map))
+ id = perf_cpu_map__min(cpu_map).cpu;
+ if (id == -1)
id = cpu.cpu;
- else
- id = perf_cpu_map__cpu(cpu_map, 0).cpu;
/* Free the perf_cpu_map used to find the cache ID */
perf_cpu_map__put(cpu_map);
@@ -1427,6 +1409,7 @@ static struct aggr_cpu_id aggr_cpu_id__cache(struct perf_cpu cpu, void *data)
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",
@@ -1454,6 +1437,12 @@ static struct aggr_cpu_id perf_stat__get_cache_id(struct perf_stat_config *confi
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,
struct perf_cpu cpu)
{
@@ -1484,7 +1473,7 @@ static struct aggr_cpu_id perf_stat__get_aggr(struct perf_stat_config *config,
struct aggr_cpu_id id;
/* per-process mode - should use global aggr mode */
- if (cpu.cpu == -1)
+ if (cpu.cpu == -1 || cpu.cpu >= config->cpus_aggr_map->nr)
return get_id(config, cpu);
if (aggr_cpu_id__is_empty(&config->cpus_aggr_map->map[cpu.cpu]))
@@ -1506,6 +1495,12 @@ static struct aggr_cpu_id perf_stat__get_die_cached(struct perf_stat_config *con
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)
{
@@ -1543,6 +1538,8 @@ static aggr_cpu_id_get_t aggr_mode__get_aggr(enum aggr_mode aggr_mode)
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:
@@ -1568,6 +1565,8 @@ static aggr_get_id_t aggr_mode__get_id(enum aggr_mode aggr_mode)
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:
@@ -1622,33 +1621,20 @@ static int perf_stat_init_aggr_mode(void)
* taking the highest cpu number to be the size of
* the aggregation translate cpumap.
*/
- if (evsel_list->core.user_requested_cpus)
- nr = perf_cpu_map__max(evsel_list->core.user_requested_cpus).cpu;
- else
- nr = 0;
- stat_config.cpus_aggr_map = cpu_aggr_map__empty_new(nr + 1);
+ nr = perf_cpu_map__max(evsel_list->core.all_cpus).cpu + 1;
+ stat_config.cpus_aggr_map = cpu_aggr_map__empty_new(nr);
return stat_config.cpus_aggr_map ? 0 : -ENOMEM;
}
static void cpu_aggr_map__delete(struct cpu_aggr_map *map)
{
- if (map) {
- WARN_ONCE(refcount_read(&map->refcnt) != 0,
- "cpu_aggr_map refcnt unbalanced\n");
- free(map);
- }
-}
-
-static void cpu_aggr_map__put(struct cpu_aggr_map *map)
-{
- if (map && refcount_dec_and_test(&map->refcnt))
- cpu_aggr_map__delete(map);
+ free(map);
}
static void perf_stat__exit_aggr_mode(void)
{
- cpu_aggr_map__put(stat_config.aggr_map);
- cpu_aggr_map__put(stat_config.cpus_aggr_map);
+ cpu_aggr_map__delete(stat_config.aggr_map);
+ cpu_aggr_map__delete(stat_config.cpus_aggr_map);
stat_config.aggr_map = NULL;
stat_config.cpus_aggr_map = NULL;
}
@@ -1736,6 +1722,21 @@ static struct aggr_cpu_id perf_env__get_cache_aggr_by_cpu(struct perf_cpu cpu,
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;
@@ -1743,12 +1744,12 @@ static struct aggr_cpu_id perf_env__get_core_aggr_by_cpu(struct perf_cpu cpu, vo
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;
}
@@ -1796,42 +1797,48 @@ static struct aggr_cpu_id perf_env__get_global_aggr_by_cpu(struct perf_cpu cpu _
static struct aggr_cpu_id perf_stat__get_socket_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_socket_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_socket_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static struct aggr_cpu_id perf_stat__get_die_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_die_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_die_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
+}
+
+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_session__env(perf_stat.session));
}
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_cache_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_cache_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static struct aggr_cpu_id perf_stat__get_core_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_core_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_core_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static struct aggr_cpu_id perf_stat__get_cpu_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_cpu_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_cpu_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static struct aggr_cpu_id perf_stat__get_node_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_node_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_node_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static struct aggr_cpu_id perf_stat__get_global_file(struct perf_stat_config *config __maybe_unused,
struct perf_cpu cpu)
{
- return perf_env__get_global_aggr_by_cpu(cpu, &perf_stat.session->header.env);
+ return perf_env__get_global_aggr_by_cpu(cpu, perf_session__env(perf_stat.session));
}
static aggr_cpu_id_get_t aggr_mode__get_aggr_file(enum aggr_mode aggr_mode)
@@ -1841,6 +1848,8 @@ static aggr_cpu_id_get_t aggr_mode__get_aggr_file(enum aggr_mode aggr_mode)
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:
@@ -1866,6 +1875,8 @@ static aggr_get_id_t aggr_mode__get_id_file(enum aggr_mode aggr_mode)
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:
@@ -1886,7 +1897,7 @@ static aggr_get_id_t aggr_mode__get_id_file(enum aggr_mode aggr_mode)
static int perf_stat_init_aggr_mode_file(struct perf_stat *st)
{
- struct perf_env *env = &st->session->header.env;
+ struct perf_env *env = perf_session__env(st->session);
aggr_cpu_id_get_t get_id = aggr_mode__get_aggr_file(stat_config.aggr_mode);
bool needs_sort = stat_config.aggr_mode != AGGR_NONE;
@@ -1919,131 +1930,55 @@ static int perf_stat_init_aggr_mode_file(struct perf_stat *st)
return 0;
}
-/*
- * Add default attributes, if there were no attributes specified or
- * if -d/--detailed, -d -d or -d -d -d is used:
- */
-static int add_default_attributes(void)
+static int default_evlist_evsel_cmp(void *priv __maybe_unused,
+ const struct list_head *l,
+ const struct list_head *r)
{
- struct perf_event_attr default_attrs0[] = {
-
- { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK },
- { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES },
- { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CPU_MIGRATIONS },
- { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_PAGE_FAULTS },
+ const struct perf_evsel *lhs_core = container_of(l, struct perf_evsel, node);
+ const struct evsel *lhs = container_of(lhs_core, struct evsel, core);
+ const struct perf_evsel *rhs_core = container_of(r, struct perf_evsel, node);
+ const struct evsel *rhs = container_of(rhs_core, struct evsel, core);
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES },
-};
- struct perf_event_attr frontend_attrs[] = {
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_FRONTEND },
-};
- struct perf_event_attr backend_attrs[] = {
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_BACKEND },
-};
- struct perf_event_attr default_attrs1[] = {
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_INSTRUCTIONS },
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_INSTRUCTIONS },
- { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_MISSES },
+ if (evsel__leader(lhs) == evsel__leader(rhs)) {
+ /* Within the same group, respect the original order. */
+ return lhs_core->idx - rhs_core->idx;
+ }
-};
+ /* Sort default metrics evsels first, and default show events before those. */
+ if (lhs->default_metricgroup != rhs->default_metricgroup)
+ return lhs->default_metricgroup ? -1 : 1;
-/*
- * Detailed stats (-d), covering the L1 and last level data caches:
- */
- struct perf_event_attr detailed_attrs[] = {
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1D << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1D << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_LL << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_LL << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
-};
+ if (lhs->default_show_events != rhs->default_show_events)
+ return lhs->default_show_events ? -1 : 1;
-/*
- * Very detailed stats (-d -d), covering the instruction cache and the TLB caches:
- */
- struct perf_event_attr very_detailed_attrs[] = {
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1I << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1I << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_DTLB << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_DTLB << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_ITLB << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_ITLB << 0 |
- (PERF_COUNT_HW_CACHE_OP_READ << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
+ /* Sort by PMU type (prefers legacy types first). */
+ if (lhs->pmu != rhs->pmu)
+ return lhs->pmu->type - rhs->pmu->type;
-};
+ /* Sort by name. */
+ return strcmp(evsel__name((struct evsel *)lhs), evsel__name((struct evsel *)rhs));
+}
/*
- * Very, very detailed stats (-d -d -d), adding prefetch events:
+ * Add default events, if there were no attributes specified or
+ * if -d/--detailed, -d -d or -d -d -d is used:
*/
- struct perf_event_attr very_very_detailed_attrs[] = {
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1D << 0 |
- (PERF_COUNT_HW_CACHE_OP_PREFETCH << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) },
-
- { .type = PERF_TYPE_HW_CACHE,
- .config =
- PERF_COUNT_HW_CACHE_L1D << 0 |
- (PERF_COUNT_HW_CACHE_OP_PREFETCH << 8) |
- (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) },
-};
-
- struct perf_event_attr default_null_attrs[] = {};
+static int add_default_events(void)
+{
const char *pmu = parse_events_option_args.pmu_filter ?: "all";
+ struct parse_events_error err;
+ struct evlist *evlist = evlist__new();
+ struct evsel *evsel;
+ int ret = 0;
+
+ if (!evlist)
+ return -ENOMEM;
+
+ parse_events_error__init(&err);
/* Set attrs if no event is selected and !null_run: */
if (stat_config.null_run)
- return 0;
+ goto out;
if (transaction_run) {
/* Handle -T as -M transaction. Once platform specific metrics
@@ -2051,17 +1986,19 @@ static int add_default_attributes(void)
* will use this approach. To determine transaction support
* on an architecture test for such a metric name.
*/
- if (!metricgroup__has_metric(pmu, "transaction")) {
+ if (!metricgroup__has_metric_or_groups(pmu, "transaction")) {
pr_err("Missing transaction metrics\n");
- return -1;
+ ret = -1;
+ goto out;
}
- return metricgroup__parse_groups(evsel_list, pmu, "transaction",
+ ret = metricgroup__parse_groups(evlist, pmu, "transaction",
stat_config.metric_no_group,
stat_config.metric_no_merge,
stat_config.metric_no_threshold,
stat_config.user_requested_cpu_list,
stat_config.system_wide,
- &stat_config.metric_events);
+ stat_config.hardware_aware_grouping);
+ goto out;
}
if (smi_cost) {
@@ -2069,32 +2006,36 @@ static int add_default_attributes(void)
if (sysfs__read_int(FREEZE_ON_SMI_PATH, &smi) < 0) {
pr_err("freeze_on_smi is not supported.\n");
- return -1;
+ ret = -1;
+ goto out;
}
if (!smi) {
if (sysfs__write_int(FREEZE_ON_SMI_PATH, 1) < 0) {
- fprintf(stderr, "Failed to set freeze_on_smi.\n");
- return -1;
+ pr_err("Failed to set freeze_on_smi.\n");
+ ret = -1;
+ goto out;
}
smi_reset = true;
}
- if (!metricgroup__has_metric(pmu, "smi")) {
+ if (!metricgroup__has_metric_or_groups(pmu, "smi")) {
pr_err("Missing smi metrics\n");
- return -1;
+ ret = -1;
+ goto out;
}
if (!force_metric_only)
stat_config.metric_only = true;
- return metricgroup__parse_groups(evsel_list, pmu, "smi",
+ ret = metricgroup__parse_groups(evlist, pmu, "smi",
stat_config.metric_no_group,
stat_config.metric_no_merge,
stat_config.metric_no_threshold,
stat_config.user_requested_cpu_list,
stat_config.system_wide,
- &stat_config.metric_events);
+ stat_config.hardware_aware_grouping);
+ goto out;
}
if (topdown_run) {
@@ -2107,105 +2048,104 @@ static int add_default_attributes(void)
if (!max_level) {
pr_err("Topdown requested but the topdown metric groups aren't present.\n"
"(See perf list the metric groups have names like TopdownL1)\n");
- return -1;
+ ret = -1;
+ goto out;
}
if (stat_config.topdown_level > max_level) {
pr_err("Invalid top-down metrics level. The max level is %u.\n", max_level);
- return -1;
- } else if (!stat_config.topdown_level)
+ ret = -1;
+ goto out;
+ } else if (!stat_config.topdown_level) {
stat_config.topdown_level = 1;
-
+ }
if (!stat_config.interval && !stat_config.metric_only) {
fprintf(stat_config.output,
"Topdown accuracy may decrease when measuring long periods.\n"
"Please print the result regularly, e.g. -I1000\n");
}
str[8] = stat_config.topdown_level + '0';
- if (metricgroup__parse_groups(evsel_list,
+ if (metricgroup__parse_groups(evlist,
pmu, str,
/*metric_no_group=*/false,
/*metric_no_merge=*/false,
/*metric_no_threshold=*/true,
stat_config.user_requested_cpu_list,
stat_config.system_wide,
- &stat_config.metric_events) < 0)
- return -1;
+ stat_config.hardware_aware_grouping) < 0) {
+ ret = -1;
+ goto out;
+ }
}
if (!stat_config.topdown_level)
stat_config.topdown_level = 1;
- if (!evsel_list->core.nr_entries) {
- /* No events so add defaults. */
- if (target__has_cpu(&target))
- default_attrs0[0].config = PERF_COUNT_SW_CPU_CLOCK;
-
- if (evlist__add_default_attrs(evsel_list, default_attrs0) < 0)
- return -1;
- if (perf_pmus__have_event("cpu", "stalled-cycles-frontend")) {
- if (evlist__add_default_attrs(evsel_list, frontend_attrs) < 0)
- return -1;
- }
- if (perf_pmus__have_event("cpu", "stalled-cycles-backend")) {
- if (evlist__add_default_attrs(evsel_list, backend_attrs) < 0)
- return -1;
- }
- if (evlist__add_default_attrs(evsel_list, default_attrs1) < 0)
- return -1;
+ if (!evlist->core.nr_entries && !evsel_list->core.nr_entries) {
/*
- * Add TopdownL1 metrics if they exist. To minimize
- * multiplexing, don't request threshold computation.
+ * Add Default metrics. To minimize multiplexing, don't request
+ * threshold computation, but it will be computed if the events
+ * are present.
*/
- if (metricgroup__has_metric(pmu, "Default")) {
- struct evlist *metric_evlist = evlist__new();
- struct evsel *metric_evsel;
+ const char *default_metricgroup_names[] = {
+ "Default", "Default2", "Default3", "Default4",
+ };
- if (!metric_evlist)
- return -1;
+ for (size_t i = 0; i < ARRAY_SIZE(default_metricgroup_names); i++) {
+ struct evlist *metric_evlist;
+
+ if (!metricgroup__has_metric_or_groups(pmu, default_metricgroup_names[i]))
+ continue;
+
+ if ((int)i > detailed_run)
+ break;
- if (metricgroup__parse_groups(metric_evlist, pmu, "Default",
+ metric_evlist = evlist__new();
+ if (!metric_evlist) {
+ ret = -ENOMEM;
+ break;
+ }
+ if (metricgroup__parse_groups(metric_evlist, pmu, default_metricgroup_names[i],
/*metric_no_group=*/false,
/*metric_no_merge=*/false,
/*metric_no_threshold=*/true,
stat_config.user_requested_cpu_list,
stat_config.system_wide,
- &stat_config.metric_events) < 0)
- return -1;
-
- evlist__for_each_entry(metric_evlist, metric_evsel) {
- metric_evsel->skippable = true;
- metric_evsel->default_metricgroup = true;
+ stat_config.hardware_aware_grouping) < 0) {
+ evlist__delete(metric_evlist);
+ ret = -1;
+ break;
}
- evlist__splice_list_tail(evsel_list, &metric_evlist->core.entries);
+
+ evlist__for_each_entry(metric_evlist, evsel)
+ evsel->default_metricgroup = true;
+
+ evlist__splice_list_tail(evlist, &metric_evlist->core.entries);
+ metricgroup__copy_metric_events(evlist, /*cgrp=*/NULL,
+ &evlist->metric_events,
+ &metric_evlist->metric_events);
evlist__delete(metric_evlist);
}
+ list_sort(/*priv=*/NULL, &evlist->core.entries, default_evlist_evsel_cmp);
- /* Platform specific attrs */
- if (evlist__add_default_attrs(evsel_list, default_null_attrs) < 0)
- return -1;
}
-
- /* Detailed events get appended to the event list: */
-
- if (detailed_run < 1)
- return 0;
-
- /* Append detailed run extra attributes: */
- if (evlist__add_default_attrs(evsel_list, detailed_attrs) < 0)
- return -1;
-
- if (detailed_run < 2)
- return 0;
-
- /* Append very detailed run extra attributes: */
- if (evlist__add_default_attrs(evsel_list, very_detailed_attrs) < 0)
- return -1;
-
- if (detailed_run < 3)
- return 0;
-
- /* Append very, very detailed run extra attributes: */
- return evlist__add_default_attrs(evsel_list, very_very_detailed_attrs);
+out:
+ if (!ret) {
+ evlist__for_each_entry(evlist, evsel) {
+ /*
+ * Make at least one event non-skippable so fatal errors are visible.
+ * 'cycles' always used to be default and non-skippable, so use that.
+ */
+ if (!evsel__match(evsel, HARDWARE, HW_CPU_CYCLES))
+ evsel->skippable = true;
+ }
+ }
+ parse_events_error__exit(&err);
+ evlist__splice_list_tail(evsel_list, &evlist->core.entries);
+ metricgroup__copy_metric_events(evsel_list, /*cgrp=*/NULL,
+ &evsel_list->metric_events,
+ &evlist->metric_events);
+ evlist__delete(evlist);
+ return ret;
}
static const char * const stat_record_usage[] = {
@@ -2227,13 +2167,15 @@ static void init_features(struct perf_session *session)
perf_header__clear_feat(&session->header, HEADER_AUXTRACE);
}
-static int __cmd_record(int argc, const char **argv)
+static int __cmd_record(const struct option stat_options[], struct opt_aggr_mode *opt_mode,
+ int argc, const char **argv)
{
struct perf_session *session;
struct perf_data *data = &perf_stat.data;
argc = parse_options(argc, argv, stat_options, stat_record_usage,
PARSE_OPT_STOP_AT_NON_OPTION);
+ stat_config.aggr_mode = opt_aggr_mode_to_aggr_mode(opt_mode);
if (output_name)
data->path = output_name;
@@ -2257,18 +2199,20 @@ static int __cmd_record(int argc, const char **argv)
return argc;
}
-static int process_stat_round_event(struct perf_session *session,
+static int process_stat_round_event(const struct perf_tool *tool __maybe_unused,
+ struct perf_session *session,
union perf_event *event)
{
struct perf_record_stat_round *stat_round = &event->stat_round;
struct timespec tsh, *ts = NULL;
- const char **argv = session->header.env.cmdline_argv;
- int argc = session->header.env.nr_cmdline;
+ struct perf_env *env = perf_session__env(session);
+ const char **argv = env->cmdline_argv;
+ int argc = env->nr_cmdline;
process_counters();
if (stat_round->type == PERF_STAT_ROUND_TYPE__FINAL)
- update_stats(&walltime_nsecs_stats, stat_round->time);
+ update_stats(stat_config.walltime_nsecs_stats, stat_round->time);
if (stat_config.interval && stat_round->time) {
tsh.tv_sec = stat_round->time / NSEC_PER_SEC;
@@ -2281,15 +2225,15 @@ static int process_stat_round_event(struct perf_session *session,
}
static
-int process_stat_config_event(struct perf_session *session,
+int process_stat_config_event(const struct perf_tool *tool,
+ struct perf_session *session,
union perf_event *event)
{
- struct perf_tool *tool = session->tool;
struct perf_stat *st = container_of(tool, struct perf_stat, tool);
perf_event__read_stat_config(&stat_config, &event->stat_config);
- if (perf_cpu_map__empty(st->cpus)) {
+ if (perf_cpu_map__is_empty(st->cpus)) {
if (st->aggr_mode != AGGR_UNSET)
pr_warning("warning: processing task data, aggregation mode not set\n");
} else if (st->aggr_mode != AGGR_UNSET) {
@@ -2330,10 +2274,10 @@ static int set_maps(struct perf_stat *st)
}
static
-int process_thread_map_event(struct perf_session *session,
+int process_thread_map_event(const struct perf_tool *tool,
+ struct perf_session *session __maybe_unused,
union perf_event *event)
{
- struct perf_tool *tool = session->tool;
struct perf_stat *st = container_of(tool, struct perf_stat, tool);
if (st->threads) {
@@ -2349,10 +2293,10 @@ int process_thread_map_event(struct perf_session *session,
}
static
-int process_cpu_map_event(struct perf_session *session,
+int process_cpu_map_event(const struct perf_tool *tool,
+ struct perf_session *session __maybe_unused,
union perf_event *event)
{
- struct perf_tool *tool = session->tool;
struct perf_stat *st = container_of(tool, struct perf_stat, tool);
struct perf_cpu_map *cpus;
@@ -2375,15 +2319,6 @@ static const char * const stat_report_usage[] = {
};
static struct perf_stat perf_stat = {
- .tool = {
- .attr = perf_event__process_attr,
- .event_update = perf_event__process_event_update,
- .thread_map = process_thread_map_event,
- .cpu_map = process_cpu_map_event,
- .stat_config = process_stat_config_event,
- .stat = perf_event__process_stat_event,
- .stat_round = process_stat_round_event,
- },
.aggr_mode = AGGR_UNSET,
.aggr_level = 0,
};
@@ -2397,6 +2332,8 @@ static int __cmd_report(int argc, const char **argv)
"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)",
@@ -2424,6 +2361,15 @@ static int __cmd_report(int argc, const char **argv)
perf_stat.data.path = input_name;
perf_stat.data.mode = PERF_DATA_MODE_READ;
+ perf_tool__init(&perf_stat.tool, /*ordered_events=*/false);
+ perf_stat.tool.attr = perf_event__process_attr;
+ perf_stat.tool.event_update = perf_event__process_event_update;
+ perf_stat.tool.thread_map = process_thread_map_event;
+ perf_stat.tool.cpu_map = process_cpu_map_event;
+ perf_stat.tool.stat_config = process_stat_config_event;
+ perf_stat.tool.stat = perf_event__process_stat_event;
+ perf_stat.tool.stat_round = process_stat_round_event;
+
session = perf_session__new(&perf_stat.data, &perf_stat.tool);
if (IS_ERR(session))
return PTR_ERR(session);
@@ -2472,8 +2418,182 @@ static void setup_system_wide(int forks)
}
}
+#ifdef HAVE_ARCH_X86_64_SUPPORT
+static int parse_tpebs_mode(const struct option *opt, const char *str,
+ int unset __maybe_unused)
+{
+ enum tpebs_mode *mode = opt->value;
+
+ if (!strcasecmp("mean", str)) {
+ *mode = TPEBS_MODE__MEAN;
+ return 0;
+ }
+ if (!strcasecmp("min", str)) {
+ *mode = TPEBS_MODE__MIN;
+ return 0;
+ }
+ if (!strcasecmp("max", str)) {
+ *mode = TPEBS_MODE__MAX;
+ return 0;
+ }
+ if (!strcasecmp("last", str)) {
+ *mode = TPEBS_MODE__LAST;
+ return 0;
+ }
+ return -1;
+}
+#endif // HAVE_ARCH_X86_64_SUPPORT
+
int cmd_stat(int argc, const char **argv)
{
+ struct opt_aggr_mode opt_mode = {};
+ struct option stat_options[] = {
+ OPT_BOOLEAN('T', "transaction", &transaction_run,
+ "hardware transaction statistics"),
+ OPT_CALLBACK('e', "event", &parse_events_option_args, "event",
+ "event selector. use 'perf list' to list available events",
+ parse_events_option),
+ OPT_CALLBACK(0, "filter", &evsel_list, "filter",
+ "event filter", parse_filter),
+ OPT_BOOLEAN('i', "no-inherit", &stat_config.no_inherit,
+ "child tasks do not inherit counters"),
+ OPT_STRING('p', "pid", &target.pid, "pid",
+ "stat events on existing process id"),
+ OPT_STRING('t', "tid", &target.tid, "tid",
+ "stat events on existing thread id"),
+#ifdef HAVE_BPF_SKEL
+ OPT_STRING('b', "bpf-prog", &target.bpf_str, "bpf-prog-id",
+ "stat events on existing bpf program id"),
+ OPT_BOOLEAN(0, "bpf-counters", &target.use_bpf,
+ "use bpf program to count events"),
+ OPT_STRING(0, "bpf-attr-map", &target.attr_map, "attr-map-path",
+ "path to perf_event_attr map"),
+#endif
+ OPT_BOOLEAN('a', "all-cpus", &target.system_wide,
+ "system-wide collection from all CPUs"),
+ OPT_BOOLEAN(0, "scale", &stat_config.scale,
+ "Use --no-scale to disable counter scaling for multiplexing"),
+ OPT_INCR('v', "verbose", &verbose,
+ "be more verbose (show counter open errors, etc)"),
+ OPT_INTEGER('r', "repeat", &stat_config.run_count,
+ "repeat command and print average + stddev (max: 100, forever: 0)"),
+ OPT_BOOLEAN(0, "table", &stat_config.walltime_run_table,
+ "display details about each run (only with -r option)"),
+ OPT_BOOLEAN('n', "null", &stat_config.null_run,
+ "null run - dont start any counters"),
+ OPT_INCR('d', "detailed", &detailed_run,
+ "detailed run - start a lot of events"),
+ OPT_BOOLEAN('S', "sync", &sync_run,
+ "call sync() before starting a run"),
+ OPT_CALLBACK_NOOPT('B', "big-num", NULL, NULL,
+ "print large numbers with thousands\' separators",
+ stat__set_big_num),
+ OPT_STRING('C', "cpu", &target.cpu_list, "cpu",
+ "list of cpus to monitor in system-wide"),
+ OPT_BOOLEAN('A', "no-aggr", &opt_mode.no_aggr,
+ "disable aggregation across CPUs or PMUs"),
+ OPT_BOOLEAN(0, "no-merge", &opt_mode.no_aggr,
+ "disable aggregation the same as -A or -no-aggr"),
+ OPT_BOOLEAN(0, "hybrid-merge", &stat_config.hybrid_merge,
+ "Merge identical named hybrid events"),
+ OPT_STRING('x', "field-separator", &stat_config.csv_sep, "separator",
+ "print counts with custom separator"),
+ OPT_BOOLEAN('j', "json-output", &stat_config.json_output,
+ "print counts in JSON format"),
+ OPT_CALLBACK('G', "cgroup", &evsel_list, "name",
+ "monitor event in cgroup name only", parse_stat_cgroups),
+ OPT_STRING(0, "for-each-cgroup", &stat_config.cgroup_list, "name",
+ "expand events for each cgroup"),
+ OPT_STRING('o', "output", &output_name, "file", "output file name"),
+ OPT_BOOLEAN(0, "append", &append_file, "append to the output file"),
+ OPT_INTEGER(0, "log-fd", &output_fd,
+ "log output to fd, instead of stderr"),
+ OPT_STRING(0, "pre", &pre_cmd, "command",
+ "command to run prior to the measured command"),
+ OPT_STRING(0, "post", &post_cmd, "command",
+ "command to run after to the measured command"),
+ OPT_UINTEGER('I', "interval-print", &stat_config.interval,
+ "print counts at regular interval in ms "
+ "(overhead is possible for values <= 100ms)"),
+ OPT_INTEGER(0, "interval-count", &stat_config.times,
+ "print counts for fixed number of times"),
+ OPT_BOOLEAN(0, "interval-clear", &stat_config.interval_clear,
+ "clear screen in between new interval"),
+ OPT_UINTEGER(0, "timeout", &stat_config.timeout,
+ "stop workload and print counts after a timeout period in ms (>= 10ms)"),
+ OPT_BOOLEAN(0, "per-socket", &opt_mode.socket,
+ "aggregate counts per processor socket"),
+ OPT_BOOLEAN(0, "per-die", &opt_mode.die, "aggregate counts per processor die"),
+ OPT_BOOLEAN(0, "per-cluster", &opt_mode.cluster,
+ "aggregate counts per processor cluster"),
+ OPT_CALLBACK_OPTARG(0, "per-cache", &opt_mode, &stat_config.aggr_level,
+ "cache level", "aggregate count at this cache level (Default: LLC)",
+ parse_cache_level),
+ OPT_BOOLEAN(0, "per-core", &opt_mode.core,
+ "aggregate counts per physical processor core"),
+ OPT_BOOLEAN(0, "per-thread", &opt_mode.thread, "aggregate counts per thread"),
+ OPT_BOOLEAN(0, "per-node", &opt_mode.node, "aggregate counts per numa node"),
+ OPT_INTEGER('D', "delay", &target.initial_delay,
+ "ms to wait before starting measurement after program start (-1: start with events disabled)"),
+ OPT_CALLBACK_NOOPT(0, "metric-only", &stat_config.metric_only, NULL,
+ "Only print computed metrics. No raw values", enable_metric_only),
+ OPT_BOOLEAN(0, "metric-no-group", &stat_config.metric_no_group,
+ "don't group metric events, impacts multiplexing"),
+ OPT_BOOLEAN(0, "metric-no-merge", &stat_config.metric_no_merge,
+ "don't try to share events between metrics in a group"),
+ OPT_BOOLEAN(0, "metric-no-threshold", &stat_config.metric_no_threshold,
+ "disable adding events for the metric threshold calculation"),
+ OPT_BOOLEAN(0, "topdown", &topdown_run,
+ "measure top-down statistics"),
+#ifdef HAVE_ARCH_X86_64_SUPPORT
+ OPT_BOOLEAN(0, "record-tpebs", &tpebs_recording,
+ "enable recording for tpebs when retire_latency required"),
+ OPT_CALLBACK(0, "tpebs-mode", &tpebs_mode, "tpebs-mode",
+ "Mode of TPEBS recording: mean, min or max",
+ parse_tpebs_mode),
+#endif
+ OPT_UINTEGER(0, "td-level", &stat_config.topdown_level,
+ "Set the metrics level for the top-down statistics (0: max level)"),
+ OPT_BOOLEAN(0, "smi-cost", &smi_cost,
+ "measure SMI cost"),
+ OPT_CALLBACK('M', "metrics", &evsel_list, "metric/metric group list",
+ "monitor specified metrics or metric groups (separated by ,)",
+ append_metric_groups),
+ OPT_BOOLEAN_FLAG(0, "all-kernel", &stat_config.all_kernel,
+ "Configure all used events to run in kernel space.",
+ PARSE_OPT_EXCLUSIVE),
+ OPT_BOOLEAN_FLAG(0, "all-user", &stat_config.all_user,
+ "Configure all used events to run in user space.",
+ PARSE_OPT_EXCLUSIVE),
+ OPT_BOOLEAN(0, "percore-show-thread", &stat_config.percore_show_thread,
+ "Use with 'percore' event qualifier to show the event "
+ "counts of one hardware thread by sum up total hardware "
+ "threads of same physical core"),
+ OPT_BOOLEAN(0, "summary", &stat_config.summary,
+ "print summary for interval mode"),
+ OPT_BOOLEAN(0, "no-csv-summary", &stat_config.no_csv_summary,
+ "don't print 'summary' for CSV summary output"),
+ OPT_BOOLEAN(0, "quiet", &quiet,
+ "don't print any output, messages or warnings (useful with record)"),
+ OPT_CALLBACK(0, "cputype", &evsel_list, "hybrid cpu type",
+ "Only enable events on applying cpu with this type "
+ "for hybrid platform (e.g. core or atom)",
+ parse_cputype),
+#ifdef HAVE_LIBPFM
+ OPT_CALLBACK(0, "pfm-events", &evsel_list, "event",
+ "libpfm4 event selector. use 'perf list' to list available events",
+ parse_libpfm_events_option),
+#endif
+ OPT_CALLBACK(0, "control", &stat_config, "fd:ctl-fd[,ack-fd] or fifo:ctl-fifo[,ack-fifo]",
+ "Listen on ctl-fd descriptor for command to control measurement ('enable': enable events, 'disable': disable events).\n"
+ "\t\t\t Optionally send control command completion ('ack\\n') to ack-fd descriptor.\n"
+ "\t\t\t Alternatively, ctl-fifo / ack-fifo will be opened and used as ctl-fd / ack-fd.",
+ parse_control_option),
+ OPT_CALLBACK_OPTARG(0, "iostat", &evsel_list, &stat_config, "default",
+ "measure I/O performance metrics provided by arch/platform",
+ iostat_parse),
+ OPT_END()
+ };
const char * const stat_usage[] = {
"perf stat [<options>] [<command>]",
NULL
@@ -2484,6 +2604,7 @@ int cmd_stat(int argc, const char **argv)
unsigned int interval, timeout;
const char * const stat_subcommands[] = { "record", "report" };
char errbuf[BUFSIZ];
+ struct evsel *counter;
setlocale(LC_ALL, "");
@@ -2502,6 +2623,8 @@ int cmd_stat(int argc, const char **argv)
(const char **) stat_usage,
PARSE_OPT_STOP_AT_NON_OPTION);
+ stat_config.aggr_mode = opt_aggr_mode_to_aggr_mode(&opt_mode);
+
if (stat_config.csv_sep) {
stat_config.csv_output = true;
if (!strcmp(stat_config.csv_sep, "\\t"))
@@ -2510,7 +2633,7 @@ int cmd_stat(int argc, const char **argv)
stat_config.csv_sep = DEFAULT_SEPARATOR;
if (argc && strlen(argv[0]) > 2 && strstarts("record", argv[0])) {
- argc = __cmd_record(argc, argv);
+ argc = __cmd_record(stat_options, &opt_mode, argc, argv);
if (argc < 0)
return -1;
} else if (argc && strlen(argv[0]) > 2 && strstarts("report", argv[0]))
@@ -2542,6 +2665,14 @@ int cmd_stat(int argc, const char **argv)
goto out;
}
+ if (stat_config.csv_output || (stat_config.metric_only && stat_config.json_output)) {
+ /*
+ * Current CSV and metric-only JSON output doesn't display the
+ * metric threshold so don't compute it.
+ */
+ stat_config.metric_no_threshold = true;
+ }
+
if (stat_config.walltime_run_table && stat_config.run_count <= 1) {
fprintf(stderr, "--table is only supported with -r\n");
parse_options_usage(stat_usage, stat_options, "r", 1);
@@ -2602,6 +2733,7 @@ int cmd_stat(int argc, const char **argv)
} else if (big_num_opt == 0) /* User passed --no-big-num */
stat_config.big_num = false;
+ target.inherit = !stat_config.no_inherit;
err = target__validate(&target);
if (err) {
target__strerror(&target, err, errbuf, BUFSIZ);
@@ -2695,18 +2827,22 @@ int cmd_stat(int argc, const char **argv)
*/
if (metrics) {
const char *pmu = parse_events_option_args.pmu_filter ?: "all";
+ int ret = metricgroup__parse_groups(evsel_list, pmu, metrics,
+ stat_config.metric_no_group,
+ stat_config.metric_no_merge,
+ stat_config.metric_no_threshold,
+ stat_config.user_requested_cpu_list,
+ stat_config.system_wide,
+ stat_config.hardware_aware_grouping);
- metricgroup__parse_groups(evsel_list, pmu, metrics,
- stat_config.metric_no_group,
- stat_config.metric_no_merge,
- stat_config.metric_no_threshold,
- stat_config.user_requested_cpu_list,
- stat_config.system_wide,
- &stat_config.metric_events);
zfree(&metrics);
+ if (ret) {
+ status = ret;
+ goto out;
+ }
}
- if (add_default_attributes())
+ if (add_default_events())
goto out;
if (stat_config.cgroup_list) {
@@ -2717,16 +2853,34 @@ int cmd_stat(int argc, const char **argv)
goto out;
}
- if (evlist__expand_cgroup(evsel_list, stat_config.cgroup_list,
- &stat_config.metric_events, true) < 0) {
+ if (evlist__expand_cgroup(evsel_list, stat_config.cgroup_list, true) < 0) {
parse_options_usage(stat_usage, stat_options,
"for-each-cgroup", 0);
goto out;
}
}
-
+#ifdef HAVE_BPF_SKEL
+ if (target.use_bpf && nr_cgroups &&
+ (evsel_list->core.nr_entries / nr_cgroups) > BPERF_CGROUP__MAX_EVENTS) {
+ pr_warning("Disabling BPF counters due to more events (%d) than the max (%d)\n",
+ evsel_list->core.nr_entries / nr_cgroups, BPERF_CGROUP__MAX_EVENTS);
+ target.use_bpf = false;
+ }
+#endif // HAVE_BPF_SKEL
evlist__warn_user_requested_cpus(evsel_list, target.cpu_list);
+ evlist__for_each_entry(evsel_list, counter) {
+ /*
+ * Setup BPF counters to require CPUs as any(-1) isn't
+ * supported. evlist__create_maps below will propagate this
+ * information to the evsels. Note, evsel__is_bperf isn't yet
+ * set up, and this change must happen early, so directly use
+ * the bpf_counter variable and target information.
+ */
+ if ((counter->bpf_counter || target.use_bpf) && !target__has_cpu(&target))
+ counter->core.requires_cpu = true;
+ }
+
if (evlist__create_maps(evsel_list, &target) < 0) {
if (target__has_task(&target)) {
pr_err("Problems finding threads of monitor\n");
@@ -2825,7 +2979,10 @@ int cmd_stat(int argc, const char **argv)
evlist__reset_prev_raw_counts(evsel_list);
status = run_perf_stat(argc, argv, run_idx);
- if (forever && status != -1 && !interval) {
+ if (status < 0)
+ break;
+
+ if (forever && !interval) {
print_counters(NULL, argc, argv);
perf_stat__reset_stats();
}
@@ -2863,7 +3020,7 @@ int cmd_stat(int argc, const char **argv)
}
if (!interval) {
- if (WRITE_STAT_ROUND_EVENT(walltime_nsecs_stats.max, FINAL))
+ if (WRITE_STAT_ROUND_EVENT(stat_config.walltime_nsecs_stats->max, FINAL))
pr_err("failed to write stat round event\n");
}
@@ -2890,8 +3047,8 @@ out:
evlist__delete(evsel_list);
- metricgroup__rblist_exit(&stat_config.metric_events);
evlist__close_control(stat_config.ctl_fd, stat_config.ctl_fd_ack, &stat_config.ctl_fd_close);
- return status;
+ /* Only the low byte of status becomes the exit code. */
+ return abs(status);
}
generated by cgit (git 2.34.1) at 2025-12-25 02:56:43 +0000