Blame view
tools/perf/builtin-stat.c
64.4 KB
910070454
|
1 |
// SPDX-License-Identifier: GPL-2.0-only |
ddcacfa0f
|
2 |
/* |
bf9e18763
|
3 4 5 6 7 8 |
* builtin-stat.c * * Builtin stat command: Give a precise performance counters summary * overview about any workload, CPU or specific PID. * * Sample output: |
ddcacfa0f
|
9 |
|
2cba3ffb9
|
10 |
$ perf stat ./hackbench 10 |
ddcacfa0f
|
11 |
|
2cba3ffb9
|
12 |
Time: 0.118 |
ddcacfa0f
|
13 |
|
2cba3ffb9
|
14 |
Performance counter stats for './hackbench 10': |
ddcacfa0f
|
15 |
|
2cba3ffb9
|
16 17 18 19 20 21 22 23 24 25 26 27 28 |
1708.761321 task-clock # 11.037 CPUs utilized 41,190 context-switches # 0.024 M/sec 6,735 CPU-migrations # 0.004 M/sec 17,318 page-faults # 0.010 M/sec 5,205,202,243 cycles # 3.046 GHz 3,856,436,920 stalled-cycles-frontend # 74.09% frontend cycles idle 1,600,790,871 stalled-cycles-backend # 30.75% backend cycles idle 2,603,501,247 instructions # 0.50 insns per cycle # 1.48 stalled cycles per insn 484,357,498 branches # 283.455 M/sec 6,388,934 branch-misses # 1.32% of all branches 0.154822978 seconds time elapsed |
ddcacfa0f
|
29 |
|
5242519b0
|
30 |
* |
2cba3ffb9
|
31 |
* Copyright (C) 2008-2011, Red Hat Inc, Ingo Molnar <mingo@redhat.com> |
5242519b0
|
32 33 34 35 36 37 38 39 |
* * Improvements and fixes by: * * Arjan van de Ven <arjan@linux.intel.com> * Yanmin Zhang <yanmin.zhang@intel.com> * Wu Fengguang <fengguang.wu@intel.com> * Mike Galbraith <efault@gmx.de> * Paul Mackerras <paulus@samba.org> |
6e750a8fc
|
40 |
* Jaswinder Singh Rajput <jaswinder@kernel.org> |
ddcacfa0f
|
41 |
*/ |
16f762a2a
|
42 |
#include "builtin.h" |
c1a604dff
|
43 |
#include "perf.h" |
f14d57078
|
44 |
#include "util/cgroup.h" |
4b6ab94ea
|
45 |
#include <subcmd/parse-options.h> |
5242519b0
|
46 |
#include "util/parse-events.h" |
4cabc3d1c
|
47 |
#include "util/pmu.h" |
8f28827a1
|
48 |
#include "util/event.h" |
361c99a66
|
49 |
#include "util/evlist.h" |
69aad6f1e
|
50 |
#include "util/evsel.h" |
8f28827a1
|
51 |
#include "util/debug.h" |
a5d243d04
|
52 |
#include "util/color.h" |
0007eceac
|
53 |
#include "util/stat.h" |
60666c630
|
54 |
#include "util/header.h" |
a12b51c47
|
55 |
#include "util/cpumap.h" |
fd78260b5
|
56 |
#include "util/thread_map.h" |
d809560b3
|
57 |
#include "util/counts.h" |
687986bbe
|
58 |
#include "util/topdown.h" |
4979d0c7d
|
59 |
#include "util/session.h" |
ba6039b6c
|
60 |
#include "util/tool.h" |
a067558e2
|
61 |
#include "util/string2.h" |
b18f3e365
|
62 |
#include "util/metricgroup.h" |
ea49e01cf
|
63 |
#include "util/synthetic-events.h" |
aeb00b1ae
|
64 |
#include "util/target.h" |
f37110205
|
65 |
#include "util/time-utils.h" |
9660e08ee
|
66 |
#include "util/top.h" |
4804e0111
|
67 |
#include "util/affinity.h" |
709434907
|
68 |
#include "util/pfm.h" |
ba6039b6c
|
69 |
#include "asm/bug.h" |
ddcacfa0f
|
70 |
|
bd48c63eb
|
71 |
#include <linux/time64.h> |
7f7c536f2
|
72 |
#include <linux/zalloc.h> |
44b1e60ab
|
73 |
#include <api/fs/fs.h> |
a43783aee
|
74 |
#include <errno.h> |
9607ad3a6
|
75 |
#include <signal.h> |
1f16c5754
|
76 |
#include <stdlib.h> |
ddcacfa0f
|
77 |
#include <sys/prctl.h> |
fd20e8111
|
78 |
#include <inttypes.h> |
5af52b51f
|
79 |
#include <locale.h> |
e3b03b6c1
|
80 |
#include <math.h> |
7a8ef4c4b
|
81 82 |
#include <sys/types.h> #include <sys/stat.h> |
4208735d8
|
83 |
#include <sys/wait.h> |
7a8ef4c4b
|
84 |
#include <unistd.h> |
0ce2da148
|
85 86 |
#include <sys/time.h> #include <sys/resource.h> |
6ef81c55a
|
87 |
#include <linux/err.h> |
16c8a1093
|
88 |
|
3052ba56b
|
89 |
#include <linux/ctype.h> |
453fa0309
|
90 |
#include <perf/evlist.h> |
3d689ed60
|
91 |
|
d7470b6af
|
92 |
#define DEFAULT_SEPARATOR " " |
daefd0bc0
|
93 |
#define FREEZE_ON_SMI_PATH "devices/cpu/freeze_on_smi" |
d7470b6af
|
94 |
|
d4f63a474
|
95 |
static void print_counters(struct timespec *ts, int argc, const char **argv); |
13370a9b5
|
96 |
|
4cabc3d1c
|
97 |
/* Default events used for perf stat -T */ |
a454742c1
|
98 99 |
static const char *transaction_attrs = { "task-clock," |
4cabc3d1c
|
100 101 102 103 104 105 106 107 108 109 110 |
"{" "instructions," "cycles," "cpu/cycles-t/," "cpu/tx-start/," "cpu/el-start/," "cpu/cycles-ct/" "}" }; /* More limited version when the CPU does not have all events. */ |
a454742c1
|
111 112 |
static const char * transaction_limited_attrs = { "task-clock," |
4cabc3d1c
|
113 114 115 116 117 118 119 |
"{" "instructions," "cycles," "cpu/cycles-t/," "cpu/tx-start/" "}" }; |
44b1e60ab
|
120 121 122 123 124 125 126 127 |
static const char * topdown_attrs[] = { "topdown-total-slots", "topdown-slots-retired", "topdown-recovery-bubbles", "topdown-fetch-bubbles", "topdown-slots-issued", NULL, }; |
55c36a9fc
|
128 129 130 131 132 133 134 135 |
static const char *topdown_metric_attrs[] = { "slots", "topdown-retiring", "topdown-bad-spec", "topdown-fe-bound", "topdown-be-bound", NULL, }; |
daefd0bc0
|
136 137 138 139 140 141 142 |
static const char *smi_cost_attrs = { "{" "msr/aperf/," "msr/smi/," "cycles" "}" }; |
63503dba8
|
143 |
static struct evlist *evsel_list; |
361c99a66
|
144 |
|
602ad878d
|
145 |
static struct target target = { |
77a6f014e
|
146 147 |
.uid = UINT_MAX, }; |
ddcacfa0f
|
148 |
|
c1a1f5d9d
|
149 |
#define METRIC_ONLY_LEN 20 |
d07f0b120
|
150 |
static volatile pid_t child_pid = -1; |
2cba3ffb9
|
151 |
static int detailed_run = 0; |
4cabc3d1c
|
152 |
static bool transaction_run; |
44b1e60ab
|
153 |
static bool topdown_run = false; |
daefd0bc0
|
154 155 |
static bool smi_cost = false; static bool smi_reset = false; |
d7470b6af
|
156 |
static int big_num_opt = -1; |
43bece797
|
157 |
static bool group = false; |
1f16c5754
|
158 159 160 |
static const char *pre_cmd = NULL; static const char *post_cmd = NULL; static bool sync_run = false; |
a7e191c37
|
161 |
static bool forever = false; |
44b1e60ab
|
162 |
static bool force_metric_only = false; |
13370a9b5
|
163 |
static struct timespec ref_time; |
e05473111
|
164 |
static bool append_file; |
db06a269e
|
165 |
static bool interval_count; |
e05473111
|
166 167 |
static const char *output_name; static int output_fd; |
5af52b51f
|
168 |
|
4979d0c7d
|
169 170 |
struct perf_stat { bool record; |
8ceb41d7e
|
171 |
struct perf_data data; |
4979d0c7d
|
172 173 |
struct perf_session *session; u64 bytes_written; |
ba6039b6c
|
174 |
struct perf_tool tool; |
1975d36e1
|
175 |
bool maps_allocated; |
f854839ba
|
176 |
struct perf_cpu_map *cpus; |
9749b90e5
|
177 |
struct perf_thread_map *threads; |
89af4e05c
|
178 |
enum aggr_mode aggr_mode; |
4979d0c7d
|
179 180 181 182 |
}; static struct perf_stat perf_stat; #define STAT_RECORD perf_stat.record |
60666c630
|
183 |
static volatile int done = 0; |
421a50f3f
|
184 |
static struct perf_stat_config stat_config = { |
26893a601
|
185 186 187 188 189 190 |
.aggr_mode = AGGR_GLOBAL, .scale = true, .unit_width = 4, /* strlen("unit") */ .run_count = 1, .metric_only_len = METRIC_ONLY_LEN, .walltime_nsecs_stats = &walltime_nsecs_stats, |
34ff0866d
|
191 |
.big_num = true, |
27e9769aa
|
192 193 |
.ctl_fd = -1, .ctl_fd_ack = -1 |
421a50f3f
|
194 |
}; |
a9a179024
|
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 |
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 (a->core.cpus->nr != b->core.cpus->nr) return false; for (int i = 0; i < a->core.cpus->nr; i++) { if (a->core.cpus->map[i] != b->core.cpus->map[i]) return false; } return true; } static void evlist__check_cpu_maps(struct evlist *evlist) { struct evsel *evsel, *pos, *leader; char buf[1024]; evlist__for_each_entry(evlist, evsel) { leader = evsel->leader; /* Check that leader matches cpus with each member. */ if (leader == evsel) continue; if (cpus_map_matched(leader, evsel)) continue; /* If there's mismatch disable the group and warn user. */ WARN_ONCE(1, "WARNING: grouped events cpus do not match, disabling group: "); evsel__group_desc(leader, buf, sizeof(buf)); pr_warning(" %s ", buf); if (verbose) { cpu_map__snprint(leader->core.cpus, buf, sizeof(buf)); pr_warning(" %s: %s ", leader->name, buf); cpu_map__snprint(evsel->core.cpus, buf, sizeof(buf)); pr_warning(" %s: %s ", evsel->name, buf); } for_each_group_evsel(pos, leader) { pos->leader = pos; pos->core.nr_members = 0; } evsel->leader->core.nr_members = 0; } } |
13370a9b5
|
251 252 253 254 255 |
static inline void diff_timespec(struct timespec *r, struct timespec *a, struct timespec *b) { r->tv_sec = a->tv_sec - b->tv_sec; if (a->tv_nsec < b->tv_nsec) { |
310ebb936
|
256 |
r->tv_nsec = a->tv_nsec + NSEC_PER_SEC - b->tv_nsec; |
13370a9b5
|
257 258 259 260 261 |
r->tv_sec--; } else { r->tv_nsec = a->tv_nsec - b->tv_nsec ; } } |
254ecbc74
|
262 263 |
static void perf_stat__reset_stats(void) { |
56739444d
|
264 |
int i; |
254ecbc74
|
265 |
perf_evlist__reset_stats(evsel_list); |
f87027b96
|
266 |
perf_stat__reset_shadow_stats(); |
56739444d
|
267 268 269 |
for (i = 0; i < stat_config.stats_num; i++) perf_stat__reset_shadow_per_stat(&stat_config.stats[i]); |
1eda3b214
|
270 |
} |
8b99b1a4e
|
271 272 273 274 |
static int process_synthesized_event(struct perf_tool *tool __maybe_unused, union perf_event *event, struct perf_sample *sample __maybe_unused, struct machine *machine __maybe_unused) |
4979d0c7d
|
275 |
{ |
8ceb41d7e
|
276 |
if (perf_data__write(&perf_stat.data, event, event->header.size) < 0) { |
4979d0c7d
|
277 278 279 280 |
pr_err("failed to write perf data, error: %m "); return -1; } |
8b99b1a4e
|
281 |
perf_stat.bytes_written += event->header.size; |
4979d0c7d
|
282 283 |
return 0; } |
1975d36e1
|
284 |
static int write_stat_round_event(u64 tm, u64 type) |
7aad0c32b
|
285 |
{ |
1975d36e1
|
286 |
return perf_event__synthesize_stat_round(NULL, tm, type, |
7aad0c32b
|
287 288 289 290 291 292 |
process_synthesized_event, NULL); } #define WRITE_STAT_ROUND_EVENT(time, interval) \ write_stat_round_event(time, PERF_STAT_ROUND_TYPE__ ## interval) |
8cd36f3ef
|
293 |
#define SID(e, x, y) xyarray__entry(e->core.sample_id, x, y) |
5a6ea81b8
|
294 |
|
ddc6999ea
|
295 296 |
static int evsel__write_stat_event(struct evsel *counter, u32 cpu, u32 thread, struct perf_counts_values *count) |
5a6ea81b8
|
297 298 299 300 301 302 |
{ struct perf_sample_id *sid = SID(counter, cpu, thread); return perf_event__synthesize_stat(NULL, cpu, thread, sid->id, count, process_synthesized_event, NULL); } |
32dcd021d
|
303 |
static int read_single_counter(struct evsel *counter, int cpu, |
f0fbb114e
|
304 305 306 307 308 309 310 311 312 313 |
int thread, struct timespec *rs) { if (counter->tool_event == PERF_TOOL_DURATION_TIME) { u64 val = rs->tv_nsec + rs->tv_sec*1000000000ULL; struct perf_counts_values *count = perf_counts(counter->counts, cpu, thread); count->ena = count->run = val; count->val = val; return 0; } |
ea0896927
|
314 |
return evsel__read_counter(counter, cpu, thread); |
f0fbb114e
|
315 |
} |
f5b4a9c3a
|
316 317 318 319 |
/* * Read out the results of a single counter: * do not aggregate counts across CPUs in system-wide mode */ |
4b49ab708
|
320 |
static int read_counter_cpu(struct evsel *counter, struct timespec *rs, int cpu) |
f5b4a9c3a
|
321 |
{ |
a2f354e3a
|
322 |
int nthreads = perf_thread_map__nr(evsel_list->core.threads); |
4b49ab708
|
323 |
int thread; |
f5b4a9c3a
|
324 |
|
3b4331d9a
|
325 326 |
if (!counter->supported) return -ENOENT; |
648b5af3f
|
327 |
if (counter->core.system_wide) |
9bf1a5291
|
328 329 330 |
nthreads = 1; for (thread = 0; thread < nthreads; thread++) { |
4b49ab708
|
331 332 333 334 335 336 |
struct perf_counts_values *count; count = perf_counts(counter->counts, cpu, thread); /* * The leader's group read loads data into its group members |
ea0896927
|
337 |
* (via evsel__read_counter()) and sets their count->loaded. |
4b49ab708
|
338 339 340 341 342 343 344 345 |
*/ if (!perf_counts__is_loaded(counter->counts, cpu, thread) && read_single_counter(counter, cpu, thread, rs)) { counter->counts->scaled = -1; perf_counts(counter->counts, cpu, thread)->ena = 0; perf_counts(counter->counts, cpu, thread)->run = 0; return -1; } |
5a6ea81b8
|
346 |
|
4b49ab708
|
347 |
perf_counts__set_loaded(counter->counts, cpu, thread, false); |
82bf311e1
|
348 |
|
4b49ab708
|
349 |
if (STAT_RECORD) { |
ddc6999ea
|
350 |
if (evsel__write_stat_event(counter, cpu, thread, count)) { |
4b49ab708
|
351 352 353 |
pr_err("failed to write stat event "); return -1; |
5a6ea81b8
|
354 |
} |
4b49ab708
|
355 |
} |
0b1abbf4a
|
356 |
|
4b49ab708
|
357 358 359 360 |
if (verbose > 1) { fprintf(stat_config.output, "%s: %d: %" PRIu64 " %" PRIu64 " %" PRIu64 " ", |
8ab2e96d8
|
361 |
evsel__name(counter), |
4b49ab708
|
362 363 |
cpu, count->val, count->ena, count->run); |
9bf1a5291
|
364 |
} |
f5b4a9c3a
|
365 |
} |
c52b12ed2
|
366 367 |
return 0; |
2996f5ddb
|
368 |
} |
c7e5b328a
|
369 |
static int read_affinity_counters(struct timespec *rs) |
13370a9b5
|
370 |
{ |
32dcd021d
|
371 |
struct evsel *counter; |
4b49ab708
|
372 373 374 375 |
struct affinity affinity; int i, ncpus, cpu; if (affinity__setup(&affinity) < 0) |
c7e5b328a
|
376 |
return -1; |
4b49ab708
|
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 |
ncpus = perf_cpu_map__nr(evsel_list->core.all_cpus); if (!target__has_cpu(&target) || target__has_per_thread(&target)) ncpus = 1; evlist__for_each_cpu(evsel_list, i, cpu) { if (i >= ncpus) break; affinity__set(&affinity, cpu); evlist__for_each_entry(evsel_list, counter) { if (evsel__cpu_iter_skip(counter, cpu)) continue; if (!counter->err) { counter->err = read_counter_cpu(counter, rs, counter->cpu_iter - 1); } } } affinity__cleanup(&affinity); |
c7e5b328a
|
396 397 398 399 400 401 |
return 0; } static void read_counters(struct timespec *rs) { struct evsel *counter; |
ee6a96143
|
402 |
if (!stat_config.stop_read_counter && (read_affinity_counters(rs) < 0)) |
c7e5b328a
|
403 |
return; |
13370a9b5
|
404 |
|
e5cadb93d
|
405 |
evlist__for_each_entry(evsel_list, counter) { |
4b49ab708
|
406 |
if (counter->err) |
245bad8eb
|
407 408 |
pr_debug("failed to read counter %s ", counter->name); |
4b49ab708
|
409 |
if (counter->err == 0 && perf_stat_process_counter(&stat_config, counter)) |
3b3eb0445
|
410 411 |
pr_warning("failed to process counter %s ", counter->name); |
4b49ab708
|
412 |
counter->err = 0; |
13370a9b5
|
413 |
} |
106a94a0f
|
414 |
} |
72f02a947
|
415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 |
static int runtime_stat_new(struct perf_stat_config *config, int nthreads) { int i; config->stats = calloc(nthreads, sizeof(struct runtime_stat)); if (!config->stats) return -1; config->stats_num = nthreads; for (i = 0; i < nthreads; i++) runtime_stat__init(&config->stats[i]); return 0; } static void runtime_stat_delete(struct perf_stat_config *config) { int i; if (!config->stats) return; for (i = 0; i < config->stats_num; i++) runtime_stat__exit(&config->stats[i]); zfree(&config->stats); } static void runtime_stat_reset(struct perf_stat_config *config) { int i; if (!config->stats) return; for (i = 0; i < config->stats_num; i++) perf_stat__reset_shadow_per_stat(&config->stats[i]); } |
ba411a954
|
454 |
static void process_interval(void) |
106a94a0f
|
455 |
{ |
106a94a0f
|
456 |
struct timespec ts, rs; |
106a94a0f
|
457 |
|
13370a9b5
|
458 459 |
clock_gettime(CLOCK_MONOTONIC, &ts); diff_timespec(&rs, &ts, &ref_time); |
13370a9b5
|
460 |
|
197ba86fd
|
461 |
perf_stat__reset_shadow_per_stat(&rt_stat); |
72f02a947
|
462 |
runtime_stat_reset(&stat_config); |
f0fbb114e
|
463 |
read_counters(&rs); |
7aad0c32b
|
464 |
if (STAT_RECORD) { |
bd48c63eb
|
465 |
if (WRITE_STAT_ROUND_EVENT(rs.tv_sec * NSEC_PER_SEC + rs.tv_nsec, INTERVAL)) |
7aad0c32b
|
466 467 468 |
pr_err("failed to write stat round event "); } |
b90f1333e
|
469 |
init_stats(&walltime_nsecs_stats); |
ea9eb1f45
|
470 |
update_stats(&walltime_nsecs_stats, stat_config.interval * 1000000ULL); |
d4f63a474
|
471 |
print_counters(&rs, 0, NULL); |
13370a9b5
|
472 |
} |
dece3a4d3
|
473 474 475 476 477 478 479 480 481 |
static bool handle_interval(unsigned int interval, int *times) { if (interval) { process_interval(); if (interval_count && !(--(*times))) return true; } return false; } |
67ccdecd0
|
482 |
static void enable_counters(void) |
411916880
|
483 |
{ |
2162b9c6b
|
484 485 486 487 488 489 490 |
if (stat_config.initial_delay < 0) { pr_info(EVLIST_DISABLED_MSG); return; } if (stat_config.initial_delay > 0) { pr_info(EVLIST_DISABLED_MSG); |
728c0ee0a
|
491 |
usleep(stat_config.initial_delay * USEC_PER_MSEC); |
2162b9c6b
|
492 |
} |
67ccdecd0
|
493 494 495 496 497 498 |
/* * We need to enable counters only if: * - we don't have tracee (attaching to task or cpu) * - we have initial delay configured */ |
2162b9c6b
|
499 |
if (!target__none(&target) || stat_config.initial_delay) { |
1c87f1654
|
500 |
evlist__enable(evsel_list); |
2162b9c6b
|
501 502 503 |
if (stat_config.initial_delay > 0) pr_info(EVLIST_ENABLED_MSG); } |
411916880
|
504 |
} |
3df33eff2
|
505 506 507 508 509 510 511 512 |
static void disable_counters(void) { /* * If we don't have tracee (attaching to task or cpu), counters may * still be running. To get accurate group ratios, we must stop groups * from counting before reading their constituent counters. */ if (!target__none(&target)) |
e74676deb
|
513 |
evlist__disable(evsel_list); |
3df33eff2
|
514 |
} |
f33cbe72e
|
515 |
static volatile int workload_exec_errno; |
6af206fd9
|
516 517 518 519 520 521 |
/* * perf_evlist__prepare_workload will send a SIGUSR1 * if the fork fails, since we asked by setting its * want_signal to true. */ |
f33cbe72e
|
522 523 |
static void workload_exec_failed_signal(int signo __maybe_unused, siginfo_t *info, void *ucontext __maybe_unused) |
6af206fd9
|
524 |
{ |
f33cbe72e
|
525 |
workload_exec_errno = info->si_value.sival_int; |
6af206fd9
|
526 |
} |
ddc6999ea
|
527 |
static bool evsel__should_store_id(struct evsel *counter) |
82bf311e1
|
528 |
{ |
1fc632cef
|
529 |
return STAT_RECORD || counter->core.attr.read_format & PERF_FORMAT_ID; |
82bf311e1
|
530 |
} |
cbb5df7e9
|
531 |
static bool is_target_alive(struct target *_target, |
9749b90e5
|
532 |
struct perf_thread_map *threads) |
cbb5df7e9
|
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 |
{ struct stat st; int i; if (!target__has_task(_target)) return true; for (i = 0; i < threads->nr; i++) { char path[PATH_MAX]; scnprintf(path, PATH_MAX, "%s/%d", procfs__mountpoint(), threads->map[i].pid); if (!stat(path, &st)) return true; } return false; } |
bee328cb7
|
552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 |
static void process_evlist(struct evlist *evlist, unsigned int interval) { enum evlist_ctl_cmd cmd = EVLIST_CTL_CMD_UNSUPPORTED; if (evlist__ctlfd_process(evlist, &cmd) > 0) { switch (cmd) { case EVLIST_CTL_CMD_ENABLE: pr_info(EVLIST_ENABLED_MSG); if (interval) process_interval(); break; case EVLIST_CTL_CMD_DISABLE: if (interval) process_interval(); pr_info(EVLIST_DISABLED_MSG); break; |
d20aff151
|
568 |
case EVLIST_CTL_CMD_SNAPSHOT: |
bee328cb7
|
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 |
case EVLIST_CTL_CMD_ACK: case EVLIST_CTL_CMD_UNSUPPORTED: default: break; } } } static void compute_tts(struct timespec *time_start, struct timespec *time_stop, int *time_to_sleep) { int tts = *time_to_sleep; struct timespec time_diff; diff_timespec(&time_diff, time_stop, time_start); tts -= time_diff.tv_sec * MSEC_PER_SEC + time_diff.tv_nsec / NSEC_PER_MSEC; if (tts < 0) tts = 0; *time_to_sleep = tts; } static int dispatch_events(bool forks, int timeout, int interval, int *times) |
987b82381
|
595 596 |
{ int child_exited = 0, status = 0; |
bee328cb7
|
597 598 599 600 601 602 603 604 605 606 607 |
int time_to_sleep, sleep_time; struct timespec time_start, time_stop; if (interval) sleep_time = interval; else if (timeout) sleep_time = timeout; else sleep_time = 1000; time_to_sleep = sleep_time; |
987b82381
|
608 609 610 611 612 613 614 615 616 |
while (!done) { if (forks) child_exited = waitpid(child_pid, &status, WNOHANG); else child_exited = !is_target_alive(&target, evsel_list->core.threads) ? 1 : 0; if (child_exited) break; |
bee328cb7
|
617 618 619 620 621 622 623 624 625 626 |
clock_gettime(CLOCK_MONOTONIC, &time_start); if (!(evlist__poll(evsel_list, time_to_sleep) > 0)) { /* poll timeout or EINTR */ if (timeout || handle_interval(interval, times)) break; time_to_sleep = sleep_time; } else { /* fd revent */ process_evlist(evsel_list, interval); clock_gettime(CLOCK_MONOTONIC, &time_stop); compute_tts(&time_start, &time_stop, &time_to_sleep); } |
987b82381
|
627 628 629 630 |
} return status; } |
e0e6a6ca3
|
631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 |
enum counter_recovery { COUNTER_SKIP, COUNTER_RETRY, COUNTER_FATAL, }; static enum counter_recovery stat_handle_error(struct evsel *counter) { char msg[BUFSIZ]; /* * 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. ", |
8ab2e96d8
|
650 |
evsel__name(counter)); |
e0e6a6ca3
|
651 |
counter->supported = false; |
4804e0111
|
652 653 654 655 656 |
/* * 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; |
e0e6a6ca3
|
657 658 659 660 |
if ((counter->leader != counter) || !(counter->leader->core.nr_members > 1)) return COUNTER_SKIP; |
ae4308927
|
661 |
} else if (evsel__fallback(counter, errno, msg, sizeof(msg))) { |
e0e6a6ca3
|
662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 |
if (verbose > 0) ui__warning("%s ", msg); return COUNTER_RETRY; } else if (target__has_per_thread(&target) && evsel_list->core.threads && evsel_list->core.threads->err_thread != -1) { /* * For global --per-thread case, skip current * error thread. */ if (!thread_map__remove(evsel_list->core.threads, evsel_list->core.threads->err_thread)) { evsel_list->core.threads->err_thread = -1; return COUNTER_RETRY; } } |
2bb72dbb8
|
679 |
evsel__open_strerror(counter, &target, errno, msg, sizeof(msg)); |
e0e6a6ca3
|
680 681 682 683 684 685 686 |
ui__error("%s ", msg); if (child_pid != -1) kill(child_pid, SIGTERM); return COUNTER_FATAL; } |
e55c14af4
|
687 |
static int __run_perf_stat(int argc, const char **argv, int run_idx) |
42202dd56
|
688 |
{ |
ec0d3d1fd
|
689 |
int interval = stat_config.interval; |
db06a269e
|
690 |
int times = stat_config.times; |
f1f8ad52f
|
691 |
int timeout = stat_config.timeout; |
d6195a6a2
|
692 |
char msg[BUFSIZ]; |
42202dd56
|
693 |
unsigned long long t0, t1; |
32dcd021d
|
694 |
struct evsel *counter; |
410136f5d
|
695 |
size_t l; |
42202dd56
|
696 |
int status = 0; |
6be2850ef
|
697 |
const bool forks = (argc > 0); |
8ceb41d7e
|
698 |
bool is_pipe = STAT_RECORD ? perf_stat.data.is_pipe : false; |
4804e0111
|
699 700 701 |
struct affinity affinity; int i, cpu; bool second_pass = false; |
42202dd56
|
702 |
|
60666c630
|
703 |
if (forks) { |
664c98d4e
|
704 |
if (perf_evlist__prepare_workload(evsel_list, &target, argv, is_pipe, |
735f7e0bb
|
705 |
workload_exec_failed_signal) < 0) { |
acf289227
|
706 707 |
perror("failed to prepare workload"); return -1; |
60666c630
|
708 |
} |
d20a47e70
|
709 |
child_pid = evsel_list->workload.pid; |
051ae7f73
|
710 |
} |
6a4bb04ca
|
711 |
if (group) |
63dab225f
|
712 |
perf_evlist__set_leader(evsel_list); |
6a4bb04ca
|
713 |
|
4804e0111
|
714 715 716 717 718 719 720 721 722 723 724 |
if (affinity__setup(&affinity) < 0) return -1; evlist__for_each_cpu (evsel_list, i, cpu) { affinity__set(&affinity, cpu); evlist__for_each_entry(evsel_list, counter) { if (evsel__cpu_iter_skip(counter, cpu)) continue; if (counter->reset_group || counter->errored) continue; |
42ef8a78c
|
725 |
try_again: |
4804e0111
|
726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 |
if (create_perf_stat_counter(counter, &stat_config, &target, counter->cpu_iter - 1) < 0) { /* * Weak group failed. We cannot just undo this here * because earlier CPUs might be in group mode, and the kernel * doesn't support mixing group and non group reads. Defer * it to later. * Don't close here because we're in the wrong affinity. */ if ((errno == EINVAL || errno == EBADF) && counter->leader != counter && counter->weak_group) { perf_evlist__reset_weak_group(evsel_list, counter, false); assert(counter->reset_group); 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; } |
5a5dfe4b8
|
755 |
} |
4804e0111
|
756 757 758 |
counter->supported = true; } } |
5a5dfe4b8
|
759 |
|
4804e0111
|
760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 |
if (second_pass) { /* * Now redo all the weak group after closing them, * and also close errored counters. */ evlist__for_each_cpu(evsel_list, i, cpu) { affinity__set(&affinity, cpu); /* First close errored or weak retry */ evlist__for_each_entry(evsel_list, counter) { if (!counter->reset_group && !counter->errored) continue; if (evsel__cpu_iter_skip_no_inc(counter, cpu)) continue; perf_evsel__close_cpu(&counter->core, counter->cpu_iter); |
ab6c79b81
|
775 |
} |
4804e0111
|
776 777 778 779 780 781 782 783 784 |
/* Now reopen weak */ evlist__for_each_entry(evsel_list, counter) { if (!counter->reset_group && !counter->errored) continue; if (evsel__cpu_iter_skip(counter, cpu)) continue; if (!counter->reset_group) continue; try_again_reset: |
8ab2e96d8
|
785 786 |
pr_debug2("reopening weak %s ", evsel__name(counter)); |
4804e0111
|
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 |
if (create_perf_stat_counter(counter, &stat_config, &target, counter->cpu_iter - 1) < 0) { switch (stat_handle_error(counter)) { case COUNTER_FATAL: return -1; case COUNTER_RETRY: goto try_again_reset; case COUNTER_SKIP: continue; default: break; } } counter->supported = true; } } } affinity__cleanup(&affinity); evlist__for_each_entry(evsel_list, counter) { if (!counter->supported) { perf_evsel__free_fd(&counter->core); continue; |
48290609c
|
811 |
} |
410136f5d
|
812 813 |
l = strlen(counter->unit); |
df4f7b4d4
|
814 815 |
if (l > stat_config.unit_width) stat_config.unit_width = l; |
2af4646d1
|
816 |
|
ddc6999ea
|
817 |
if (evsel__should_store_id(counter) && |
343977534
|
818 |
evsel__store_ids(counter, evsel_list)) |
2af4646d1
|
819 |
return -1; |
084ab9f86
|
820 |
} |
42202dd56
|
821 |
|
23d4aad48
|
822 |
if (perf_evlist__apply_filters(evsel_list, &counter)) { |
62d94b00f
|
823 824 |
pr_err("failed to set filter \"%s\" on event %s with %d (%s) ", |
8ab2e96d8
|
825 |
counter->filter, evsel__name(counter), errno, |
c8b5f2c96
|
826 |
str_error_r(errno, msg, sizeof(msg))); |
cfd748ae0
|
827 828 |
return -1; } |
4979d0c7d
|
829 |
if (STAT_RECORD) { |
8ceb41d7e
|
830 |
int err, fd = perf_data__fd(&perf_stat.data); |
4979d0c7d
|
831 |
|
664c98d4e
|
832 |
if (is_pipe) { |
8ceb41d7e
|
833 |
err = perf_header__write_pipe(perf_data__fd(&perf_stat.data)); |
664c98d4e
|
834 835 836 837 |
} else { err = perf_session__write_header(perf_stat.session, evsel_list, fd, false); } |
4979d0c7d
|
838 839 |
if (err < 0) return err; |
8b99b1a4e
|
840 |
|
b251892d6
|
841 842 |
err = perf_event__synthesize_stat_events(&stat_config, NULL, evsel_list, process_synthesized_event, is_pipe); |
8b99b1a4e
|
843 844 |
if (err < 0) return err; |
4979d0c7d
|
845 |
} |
42202dd56
|
846 847 848 849 |
/* * Enable counters and exec the command: */ t0 = rdclock(); |
13370a9b5
|
850 |
clock_gettime(CLOCK_MONOTONIC, &ref_time); |
42202dd56
|
851 |
|
60666c630
|
852 |
if (forks) { |
acf289227
|
853 |
perf_evlist__start_workload(evsel_list); |
67ccdecd0
|
854 |
enable_counters(); |
acf289227
|
855 |
|
27e9769aa
|
856 |
if (interval || timeout || evlist__ctlfd_initialized(evsel_list)) |
bee328cb7
|
857 |
status = dispatch_events(forks, timeout, interval, ×); |
cfbd41b78
|
858 859 860 |
if (child_pid != -1) { if (timeout) kill(child_pid, SIGTERM); |
8a99255a5
|
861 |
wait4(child_pid, &status, 0, &stat_config.ru_data); |
cfbd41b78
|
862 |
} |
6af206fd9
|
863 |
|
f33cbe72e
|
864 |
if (workload_exec_errno) { |
c8b5f2c96
|
865 |
const char *emsg = str_error_r(workload_exec_errno, msg, sizeof(msg)); |
f33cbe72e
|
866 867 |
pr_err("Workload failed: %s ", emsg); |
6af206fd9
|
868 |
return -1; |
f33cbe72e
|
869 |
} |
6af206fd9
|
870 |
|
33e49ea70
|
871 872 |
if (WIFSIGNALED(status)) psignal(WTERMSIG(status), argv[0]); |
60666c630
|
873 |
} else { |
67ccdecd0
|
874 |
enable_counters(); |
bee328cb7
|
875 |
status = dispatch_events(forks, timeout, interval, ×); |
60666c630
|
876 |
} |
42202dd56
|
877 |
|
3df33eff2
|
878 |
disable_counters(); |
42202dd56
|
879 |
t1 = rdclock(); |
54ac0b1bd
|
880 881 |
if (stat_config.walltime_run_table) stat_config.walltime_run[run_idx] = t1 - t0; |
e55c14af4
|
882 |
|
ee6a96143
|
883 |
if (interval && stat_config.summary) { |
c7e5b328a
|
884 |
stat_config.interval = 0; |
ee6a96143
|
885 |
stat_config.stop_read_counter = true; |
c7e5b328a
|
886 887 888 889 890 891 892 893 894 895 896 897 |
init_stats(&walltime_nsecs_stats); update_stats(&walltime_nsecs_stats, t1 - t0); if (stat_config.aggr_mode == AGGR_GLOBAL) perf_evlist__save_aggr_prev_raw_counts(evsel_list); perf_evlist__copy_prev_raw_counts(evsel_list); perf_evlist__reset_prev_raw_counts(evsel_list); runtime_stat_reset(&stat_config); perf_stat__reset_shadow_per_stat(&rt_stat); } else update_stats(&walltime_nsecs_stats, t1 - t0); |
42202dd56
|
898 |
|
3df33eff2
|
899 900 901 902 903 904 |
/* * Closing a group leader splits the group, and as we only disable * group leaders, results in remaining events becoming enabled. To * avoid arbitrary skew, we must read all counters before closing any * group leaders. */ |
f0fbb114e
|
905 |
read_counters(&(struct timespec) { .tv_nsec = t1-t0 }); |
08ef3af15
|
906 907 908 909 910 911 |
/* * We need to keep evsel_list alive, because it's processed * later the evsel_list will be closed after. */ if (!STAT_RECORD) |
750b4edeb
|
912 |
evlist__close(evsel_list); |
c52b12ed2
|
913 |
|
42202dd56
|
914 915 |
return WEXITSTATUS(status); } |
e55c14af4
|
916 |
static int run_perf_stat(int argc, const char **argv, int run_idx) |
1f16c5754
|
917 918 919 920 921 922 923 924 925 926 927 |
{ int ret; if (pre_cmd) { ret = system(pre_cmd); if (ret) return ret; } if (sync_run) sync(); |
e55c14af4
|
928 |
ret = __run_perf_stat(argc, argv, run_idx); |
1f16c5754
|
929 930 931 932 933 934 935 936 937 938 939 |
if (ret) return ret; if (post_cmd) { ret = system(post_cmd); if (ret) return ret; } return ret; } |
a5a9eac1a
|
940 941 |
static void print_counters(struct timespec *ts, int argc, const char **argv) { |
0174820a8
|
942 943 944 |
/* Do not print anything if we record to the pipe. */ if (STAT_RECORD && perf_stat.data.is_pipe) return; |
c512e0eae
|
945 |
perf_evlist__print_counters(evsel_list, &stat_config, &target, |
b64df7f33
|
946 |
ts, argc, argv); |
a5a9eac1a
|
947 |
} |
f7b7c26e0
|
948 |
static volatile int signr = -1; |
5242519b0
|
949 |
static void skip_signal(int signo) |
ddcacfa0f
|
950 |
{ |
ec0d3d1fd
|
951 |
if ((child_pid == -1) || stat_config.interval) |
60666c630
|
952 |
done = 1; |
f7b7c26e0
|
953 |
signr = signo; |
d07f0b120
|
954 955 956 957 958 959 960 |
/* * render child_pid harmless * won't send SIGTERM to a random * process in case of race condition * and fast PID recycling */ child_pid = -1; |
f7b7c26e0
|
961 962 963 964 |
} static void sig_atexit(void) { |
d07f0b120
|
965 966 967 968 969 970 971 972 973 974 975 |
sigset_t set, oset; /* * avoid race condition with SIGCHLD handler * in skip_signal() which is modifying child_pid * goal is to avoid send SIGTERM to a random * process */ sigemptyset(&set); sigaddset(&set, SIGCHLD); sigprocmask(SIG_BLOCK, &set, &oset); |
933da83aa
|
976 977 |
if (child_pid != -1) kill(child_pid, SIGTERM); |
d07f0b120
|
978 |
sigprocmask(SIG_SETMASK, &oset, NULL); |
f7b7c26e0
|
979 980 981 982 983 |
if (signr == -1) return; signal(signr, SIG_DFL); kill(getpid(), signr); |
5242519b0
|
984 |
} |
d778a778a
|
985 986 987 988 |
void perf_stat__set_big_num(int set) { stat_config.big_num = (set != 0); } |
1d037ca16
|
989 990 |
static int stat__set_big_num(const struct option *opt __maybe_unused, const char *s __maybe_unused, int unset) |
d7470b6af
|
991 992 |
{ big_num_opt = unset ? 0 : 1; |
d778a778a
|
993 |
perf_stat__set_big_num(!unset); |
d7470b6af
|
994 995 |
return 0; } |
44b1e60ab
|
996 997 998 999 |
static int enable_metric_only(const struct option *opt __maybe_unused, const char *s __maybe_unused, int unset) { force_metric_only = true; |
0ce5aa026
|
1000 |
stat_config.metric_only = !unset; |
44b1e60ab
|
1001 1002 |
return 0; } |
b18f3e365
|
1003 1004 1005 1006 |
static int parse_metric_groups(const struct option *opt, const char *str, int unset __maybe_unused) { |
05530a792
|
1007 1008 1009 1010 |
return metricgroup__parse_groups(opt, str, stat_config.metric_no_group, stat_config.metric_no_merge, &stat_config.metric_events); |
b18f3e365
|
1011 |
} |
27e9769aa
|
1012 1013 1014 1015 |
static int parse_control_option(const struct option *opt, const char *str, int unset __maybe_unused) { |
9864a66de
|
1016 |
struct perf_stat_config *config = opt->value; |
27e9769aa
|
1017 |
|
a8fcbd269
|
1018 1019 |
return evlist__parse_control(str, &config->ctl_fd, &config->ctl_fd_ack, &config->ctl_fd_close); } |
d1c5a0e86
|
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 |
static int parse_stat_cgroups(const struct option *opt, const char *str, int unset) { if (stat_config.cgroup_list) { pr_err("--cgroup and --for-each-cgroup cannot be used together "); return -1; } return parse_cgroups(opt, str, unset); } |
51433ead1
|
1031 |
static struct option stat_options[] = { |
e05473111
|
1032 1033 1034 1035 1036 1037 1038 |
OPT_BOOLEAN('T', "transaction", &transaction_run, "hardware transaction statistics"), OPT_CALLBACK('e', "event", &evsel_list, "event", "event selector. use 'perf list' to list available events", parse_events_option), OPT_CALLBACK(0, "filter", &evsel_list, "filter", "event filter", parse_filter), |
5698f26b4
|
1039 |
OPT_BOOLEAN('i', "no-inherit", &stat_config.no_inherit, |
e05473111
|
1040 1041 1042 1043 1044 1045 1046 1047 1048 |
"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"), OPT_BOOLEAN('a', "all-cpus", &target.system_wide, "system-wide collection from all CPUs"), OPT_BOOLEAN('g', "group", &group, "put the counters into a counter group"), |
75998bb26
|
1049 1050 |
OPT_BOOLEAN(0, "scale", &stat_config.scale, "Use --no-scale to disable counter scaling for multiplexing"), |
e05473111
|
1051 1052 |
OPT_INCR('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), |
d97ae04b3
|
1053 |
OPT_INTEGER('r', "repeat", &stat_config.run_count, |
e05473111
|
1054 |
"repeat command and print average + stddev (max: 100, forever: 0)"), |
54ac0b1bd
|
1055 |
OPT_BOOLEAN(0, "table", &stat_config.walltime_run_table, |
e55c14af4
|
1056 |
"display details about each run (only with -r option)"), |
aea0dca16
|
1057 |
OPT_BOOLEAN('n', "null", &stat_config.null_run, |
e05473111
|
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 |
"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), |
fdee335b0
|
1070 |
OPT_BOOLEAN(0, "no-merge", &stat_config.no_merge, "Do not merge identical named events"), |
fa7070a38
|
1071 |
OPT_STRING('x', "field-separator", &stat_config.csv_sep, "separator", |
e05473111
|
1072 1073 |
"print counts with custom separator"), OPT_CALLBACK('G', "cgroup", &evsel_list, "name", |
d1c5a0e86
|
1074 1075 1076 |
"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"), |
e05473111
|
1077 1078 1079 1080 1081 1082 1083 1084 1085 |
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, |
9dc9a95f0
|
1086 1087 |
"print counts at regular interval in ms " "(overhead is possible for values <= 100ms)"), |
db06a269e
|
1088 1089 |
OPT_INTEGER(0, "interval-count", &stat_config.times, "print counts for fixed number of times"), |
132c6ba3c
|
1090 |
OPT_BOOLEAN(0, "interval-clear", &stat_config.interval_clear, |
9660e08ee
|
1091 |
"clear screen in between new interval"), |
f1f8ad52f
|
1092 1093 |
OPT_UINTEGER(0, "timeout", &stat_config.timeout, "stop workload and print counts after a timeout period in ms (>= 10ms)"), |
e05473111
|
1094 1095 |
OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode, "aggregate counts per processor socket", AGGR_SOCKET), |
db5742b68
|
1096 1097 |
OPT_SET_UINT(0, "per-die", &stat_config.aggr_mode, "aggregate counts per processor die", AGGR_DIE), |
e05473111
|
1098 1099 1100 1101 |
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), |
86895b480
|
1102 1103 |
OPT_SET_UINT(0, "per-node", &stat_config.aggr_mode, "aggregate counts per numa node", AGGR_NODE), |
2162b9c6b
|
1104 1105 |
OPT_INTEGER('D', "delay", &stat_config.initial_delay, "ms to wait before starting measurement after program start (-1: start with events disabled)"), |
0ce5aa026
|
1106 |
OPT_CALLBACK_NOOPT(0, "metric-only", &stat_config.metric_only, NULL, |
44b1e60ab
|
1107 |
"Only print computed metrics. No raw values", enable_metric_only), |
05530a792
|
1108 1109 1110 1111 |
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"), |
44b1e60ab
|
1112 1113 |
OPT_BOOLEAN(0, "topdown", &topdown_run, "measure topdown level 1 statistics"), |
daefd0bc0
|
1114 1115 |
OPT_BOOLEAN(0, "smi-cost", &smi_cost, "measure SMI cost"), |
b18f3e365
|
1116 1117 1118 |
OPT_CALLBACK('M', "metrics", &evsel_list, "metric/metric group list", "monitor specified metrics or metric groups (separated by ,)", parse_metric_groups), |
dd071024b
|
1119 1120 1121 1122 1123 1124 |
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), |
1af62ce61
|
1125 1126 1127 1128 |
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"), |
ee6a96143
|
1129 1130 |
OPT_BOOLEAN(0, "summary", &stat_config.summary, "print summary for interval mode"), |
709434907
|
1131 1132 1133 1134 1135 |
#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 |
a8fcbd269
|
1136 |
OPT_CALLBACK(0, "control", &stat_config, "fd:ctl-fd[,ack-fd] or fifo:ctl-fifo[,ack-fifo]", |
27e9769aa
|
1137 1138 |
"Listen on ctl-fd descriptor for command to control measurement ('enable': enable events, 'disable': disable events). " |
a8fcbd269
|
1139 1140 1141 1142 |
"\t\t\t Optionally send control command completion ('ack\ ') to ack-fd descriptor. " "\t\t\t Alternatively, ctl-fifo / ack-fifo will be opened and used as ctl-fd / ack-fd.", |
27e9769aa
|
1143 |
parse_control_option), |
e05473111
|
1144 1145 |
OPT_END() }; |
6f6b6594b
|
1146 |
static int perf_stat__get_socket(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1147 |
struct perf_cpu_map *map, int cpu) |
1fe7a3002
|
1148 1149 1150 |
{ return cpu_map__get_socket(map, cpu, NULL); } |
db5742b68
|
1151 |
static int perf_stat__get_die(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1152 |
struct perf_cpu_map *map, int cpu) |
db5742b68
|
1153 1154 1155 |
{ return cpu_map__get_die(map, cpu, NULL); } |
6f6b6594b
|
1156 |
static int perf_stat__get_core(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1157 |
struct perf_cpu_map *map, int cpu) |
1fe7a3002
|
1158 1159 1160 |
{ return cpu_map__get_core(map, cpu, NULL); } |
86895b480
|
1161 1162 1163 1164 1165 |
static int perf_stat__get_node(struct perf_stat_config *config __maybe_unused, struct perf_cpu_map *map, int cpu) { return cpu_map__get_node(map, cpu, NULL); } |
6f6b6594b
|
1166 |
static int perf_stat__get_aggr(struct perf_stat_config *config, |
f854839ba
|
1167 |
aggr_get_id_t get_id, struct perf_cpu_map *map, int idx) |
1e5a29318
|
1168 1169 1170 1171 1172 1173 1174 |
{ int cpu; if (idx >= map->nr) return -1; cpu = map->map[idx]; |
6f6b6594b
|
1175 1176 |
if (config->cpus_aggr_map->map[cpu] == -1) config->cpus_aggr_map->map[cpu] = get_id(config, map, idx); |
1e5a29318
|
1177 |
|
6f6b6594b
|
1178 |
return config->cpus_aggr_map->map[cpu]; |
1e5a29318
|
1179 |
} |
6f6b6594b
|
1180 |
static int perf_stat__get_socket_cached(struct perf_stat_config *config, |
f854839ba
|
1181 |
struct perf_cpu_map *map, int idx) |
1e5a29318
|
1182 |
{ |
6f6b6594b
|
1183 |
return perf_stat__get_aggr(config, perf_stat__get_socket, map, idx); |
1e5a29318
|
1184 |
} |
db5742b68
|
1185 |
static int perf_stat__get_die_cached(struct perf_stat_config *config, |
f854839ba
|
1186 |
struct perf_cpu_map *map, int idx) |
db5742b68
|
1187 1188 1189 |
{ return perf_stat__get_aggr(config, perf_stat__get_die, map, idx); } |
6f6b6594b
|
1190 |
static int perf_stat__get_core_cached(struct perf_stat_config *config, |
f854839ba
|
1191 |
struct perf_cpu_map *map, int idx) |
1e5a29318
|
1192 |
{ |
6f6b6594b
|
1193 |
return perf_stat__get_aggr(config, perf_stat__get_core, map, idx); |
1e5a29318
|
1194 |
} |
86895b480
|
1195 1196 1197 1198 1199 |
static int perf_stat__get_node_cached(struct perf_stat_config *config, struct perf_cpu_map *map, int idx) { return perf_stat__get_aggr(config, perf_stat__get_node, map, idx); } |
4fc4d8dfa
|
1200 1201 |
static bool term_percore_set(void) { |
32dcd021d
|
1202 |
struct evsel *counter; |
4fc4d8dfa
|
1203 1204 1205 1206 1207 1208 1209 1210 |
evlist__for_each_entry(evsel_list, counter) { if (counter->percore) return true; } return false; } |
86ee6e18f
|
1211 1212 |
static int perf_stat_init_aggr_mode(void) { |
1e5a29318
|
1213 |
int nr; |
421a50f3f
|
1214 |
switch (stat_config.aggr_mode) { |
86ee6e18f
|
1215 |
case AGGR_SOCKET: |
f72f901d9
|
1216 |
if (cpu_map__build_socket_map(evsel_list->core.cpus, &stat_config.aggr_map)) { |
86ee6e18f
|
1217 1218 1219 |
perror("cannot build socket map"); return -1; } |
6f6b6594b
|
1220 |
stat_config.aggr_get_id = perf_stat__get_socket_cached; |
86ee6e18f
|
1221 |
break; |
db5742b68
|
1222 |
case AGGR_DIE: |
f72f901d9
|
1223 |
if (cpu_map__build_die_map(evsel_list->core.cpus, &stat_config.aggr_map)) { |
db5742b68
|
1224 1225 1226 1227 1228 |
perror("cannot build die map"); return -1; } stat_config.aggr_get_id = perf_stat__get_die_cached; break; |
12c08a9f5
|
1229 |
case AGGR_CORE: |
f72f901d9
|
1230 |
if (cpu_map__build_core_map(evsel_list->core.cpus, &stat_config.aggr_map)) { |
12c08a9f5
|
1231 1232 1233 |
perror("cannot build core map"); return -1; } |
6f6b6594b
|
1234 |
stat_config.aggr_get_id = perf_stat__get_core_cached; |
12c08a9f5
|
1235 |
break; |
86895b480
|
1236 1237 1238 1239 1240 1241 1242 |
case AGGR_NODE: if (cpu_map__build_node_map(evsel_list->core.cpus, &stat_config.aggr_map)) { perror("cannot build core map"); return -1; } stat_config.aggr_get_id = perf_stat__get_node_cached; break; |
86ee6e18f
|
1243 |
case AGGR_NONE: |
4fc4d8dfa
|
1244 |
if (term_percore_set()) { |
f72f901d9
|
1245 |
if (cpu_map__build_core_map(evsel_list->core.cpus, |
4fc4d8dfa
|
1246 1247 1248 1249 1250 1251 1252 |
&stat_config.aggr_map)) { perror("cannot build core map"); return -1; } stat_config.aggr_get_id = perf_stat__get_core_cached; } break; |
86ee6e18f
|
1253 |
case AGGR_GLOBAL: |
32b8af82e
|
1254 |
case AGGR_THREAD: |
208df99ed
|
1255 |
case AGGR_UNSET: |
86ee6e18f
|
1256 1257 1258 |
default: break; } |
1e5a29318
|
1259 1260 1261 1262 1263 1264 |
/* * The evsel_list->cpus is the base we operate on, * taking the highest cpu number to be the size of * the aggregation translate cpumap. */ |
4256d4349
|
1265 |
nr = perf_cpu_map__max(evsel_list->core.cpus); |
315c0a1f0
|
1266 |
stat_config.cpus_aggr_map = perf_cpu_map__empty_new(nr + 1); |
6f6b6594b
|
1267 |
return stat_config.cpus_aggr_map ? 0 : -ENOMEM; |
86ee6e18f
|
1268 |
} |
544c2ae7b
|
1269 1270 |
static void perf_stat__exit_aggr_mode(void) { |
38f01d8da
|
1271 1272 |
perf_cpu_map__put(stat_config.aggr_map); perf_cpu_map__put(stat_config.cpus_aggr_map); |
6f6b6594b
|
1273 1274 |
stat_config.aggr_map = NULL; stat_config.cpus_aggr_map = NULL; |
544c2ae7b
|
1275 |
} |
f854839ba
|
1276 |
static inline int perf_env__get_cpu(struct perf_env *env, struct perf_cpu_map *map, int idx) |
68d702f7a
|
1277 1278 1279 1280 1281 1282 1283 |
{ int cpu; if (idx > map->nr) return -1; cpu = map->map[idx]; |
da8a58b56
|
1284 |
if (cpu >= env->nr_cpus_avail) |
68d702f7a
|
1285 1286 1287 1288 |
return -1; return cpu; } |
f854839ba
|
1289 |
static int perf_env__get_socket(struct perf_cpu_map *map, int idx, void *data) |
68d702f7a
|
1290 1291 1292 1293 1294 1295 |
{ struct perf_env *env = data; int cpu = perf_env__get_cpu(env, map, idx); return cpu == -1 ? -1 : env->cpu[cpu].socket_id; } |
f854839ba
|
1296 |
static int perf_env__get_die(struct perf_cpu_map *map, int idx, void *data) |
db5742b68
|
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 |
{ struct perf_env *env = data; int die_id = -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. ")) return -1; if (WARN_ONCE(env->cpu[cpu].die_id >> 8, "The die id number is too big. ")) return -1; die_id = (env->cpu[cpu].socket_id << 8) | (env->cpu[cpu].die_id & 0xff); } return die_id; } |
f854839ba
|
1321 |
static int perf_env__get_core(struct perf_cpu_map *map, int idx, void *data) |
68d702f7a
|
1322 1323 1324 1325 1326 |
{ struct perf_env *env = data; int core = -1, cpu = perf_env__get_cpu(env, map, idx); if (cpu != -1) { |
68d702f7a
|
1327 |
/* |
db5742b68
|
1328 1329 1330 |
* Encode socket in bit range 31:24 * encode die id in bit range 23:16 * core_id is relative to socket and die, |
68d702f7a
|
1331 |
* we need a global id. So we combine |
db5742b68
|
1332 |
* socket + die id + core id |
68d702f7a
|
1333 |
*/ |
db5742b68
|
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 |
if (WARN_ONCE(env->cpu[cpu].socket_id >> 8, "The socket id number is too big. ")) return -1; if (WARN_ONCE(env->cpu[cpu].die_id >> 8, "The die id number is too big. ")) return -1; if (WARN_ONCE(env->cpu[cpu].core_id >> 16, "The core id number is too big. ")) return -1; core = (env->cpu[cpu].socket_id << 24) | (env->cpu[cpu].die_id << 16) | (env->cpu[cpu].core_id & 0xffff); |
68d702f7a
|
1349 1350 1351 1352 |
} return core; } |
86895b480
|
1353 1354 1355 1356 1357 1358 |
static int perf_env__get_node(struct perf_cpu_map *map, int idx, void *data) { int cpu = perf_env__get_cpu(data, map, idx); return perf_env__numa_node(data, cpu); } |
f854839ba
|
1359 1360 |
static int perf_env__build_socket_map(struct perf_env *env, struct perf_cpu_map *cpus, struct perf_cpu_map **sockp) |
68d702f7a
|
1361 1362 1363 |
{ return cpu_map__build_map(cpus, sockp, perf_env__get_socket, env); } |
f854839ba
|
1364 1365 |
static int perf_env__build_die_map(struct perf_env *env, struct perf_cpu_map *cpus, struct perf_cpu_map **diep) |
db5742b68
|
1366 1367 1368 |
{ return cpu_map__build_map(cpus, diep, perf_env__get_die, env); } |
f854839ba
|
1369 1370 |
static int perf_env__build_core_map(struct perf_env *env, struct perf_cpu_map *cpus, struct perf_cpu_map **corep) |
68d702f7a
|
1371 1372 1373 |
{ return cpu_map__build_map(cpus, corep, perf_env__get_core, env); } |
86895b480
|
1374 1375 1376 1377 1378 |
static int perf_env__build_node_map(struct perf_env *env, struct perf_cpu_map *cpus, struct perf_cpu_map **nodep) { return cpu_map__build_map(cpus, nodep, perf_env__get_node, env); } |
6f6b6594b
|
1379 |
static int perf_stat__get_socket_file(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1380 |
struct perf_cpu_map *map, int idx) |
68d702f7a
|
1381 1382 1383 |
{ return perf_env__get_socket(map, idx, &perf_stat.session->header.env); } |
db5742b68
|
1384 |
static int perf_stat__get_die_file(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1385 |
struct perf_cpu_map *map, int idx) |
db5742b68
|
1386 1387 1388 |
{ return perf_env__get_die(map, idx, &perf_stat.session->header.env); } |
68d702f7a
|
1389 |
|
6f6b6594b
|
1390 |
static int perf_stat__get_core_file(struct perf_stat_config *config __maybe_unused, |
f854839ba
|
1391 |
struct perf_cpu_map *map, int idx) |
68d702f7a
|
1392 1393 1394 |
{ return perf_env__get_core(map, idx, &perf_stat.session->header.env); } |
86895b480
|
1395 1396 1397 1398 1399 |
static int perf_stat__get_node_file(struct perf_stat_config *config __maybe_unused, struct perf_cpu_map *map, int idx) { return perf_env__get_node(map, idx, &perf_stat.session->header.env); } |
68d702f7a
|
1400 1401 1402 1403 1404 1405 |
static int perf_stat_init_aggr_mode_file(struct perf_stat *st) { struct perf_env *env = &st->session->header.env; switch (stat_config.aggr_mode) { case AGGR_SOCKET: |
f72f901d9
|
1406 |
if (perf_env__build_socket_map(env, evsel_list->core.cpus, &stat_config.aggr_map)) { |
68d702f7a
|
1407 1408 1409 |
perror("cannot build socket map"); return -1; } |
6f6b6594b
|
1410 |
stat_config.aggr_get_id = perf_stat__get_socket_file; |
68d702f7a
|
1411 |
break; |
db5742b68
|
1412 |
case AGGR_DIE: |
f72f901d9
|
1413 |
if (perf_env__build_die_map(env, evsel_list->core.cpus, &stat_config.aggr_map)) { |
db5742b68
|
1414 1415 1416 1417 1418 |
perror("cannot build die map"); return -1; } stat_config.aggr_get_id = perf_stat__get_die_file; break; |
68d702f7a
|
1419 |
case AGGR_CORE: |
f72f901d9
|
1420 |
if (perf_env__build_core_map(env, evsel_list->core.cpus, &stat_config.aggr_map)) { |
68d702f7a
|
1421 1422 1423 |
perror("cannot build core map"); return -1; } |
6f6b6594b
|
1424 |
stat_config.aggr_get_id = perf_stat__get_core_file; |
68d702f7a
|
1425 |
break; |
86895b480
|
1426 1427 1428 1429 1430 1431 1432 |
case AGGR_NODE: if (perf_env__build_node_map(env, evsel_list->core.cpus, &stat_config.aggr_map)) { perror("cannot build core map"); return -1; } stat_config.aggr_get_id = perf_stat__get_node_file; break; |
68d702f7a
|
1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 |
case AGGR_NONE: case AGGR_GLOBAL: case AGGR_THREAD: case AGGR_UNSET: default: break; } return 0; } |
2cba3ffb9
|
1443 1444 1445 1446 1447 1448 |
/* * 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) { |
44b1e60ab
|
1449 |
int err; |
9dec4473a
|
1450 |
struct perf_event_attr default_attrs0[] = { |
b070a547f
|
1451 1452 1453 1454 1455 1456 1457 |
{ .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 }, { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES }, |
9dec4473a
|
1458 1459 |
}; struct perf_event_attr frontend_attrs[] = { |
b070a547f
|
1460 |
{ .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, |
9dec4473a
|
1461 1462 |
}; struct perf_event_attr backend_attrs[] = { |
b070a547f
|
1463 |
{ .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, |
9dec4473a
|
1464 1465 |
}; struct perf_event_attr default_attrs1[] = { |
b070a547f
|
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 |
{ .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 }, }; /* * 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) }, }; /* * 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) }, }; /* * Very, very detailed stats (-d -d -d), adding prefetch events: */ 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) }, }; |
a5cfa6217
|
1562 |
struct parse_events_error errinfo; |
b070a547f
|
1563 |
|
2cba3ffb9
|
1564 |
/* Set attrs if no event is selected and !null_run: */ |
aea0dca16
|
1565 |
if (stat_config.null_run) |
2cba3ffb9
|
1566 |
return 0; |
a910e4666
|
1567 |
bzero(&errinfo, sizeof(errinfo)); |
4cabc3d1c
|
1568 |
if (transaction_run) { |
742d92ff2
|
1569 1570 1571 1572 1573 1574 1575 1576 1577 |
/* Handle -T as -M transaction. Once platform specific metrics * support has been added to the json files, all archictures * will use this approach. To determine transaction support * on an architecture test for such a metric name. */ if (metricgroup__has_metric("transaction")) { struct option opt = { .value = &evsel_list }; return metricgroup__parse_groups(&opt, "transaction", |
05530a792
|
1578 1579 |
stat_config.metric_no_group, stat_config.metric_no_merge, |
d0192fdba
|
1580 |
&stat_config.metric_events); |
742d92ff2
|
1581 |
} |
4cabc3d1c
|
1582 1583 |
if (pmu_have_event("cpu", "cycles-ct") && pmu_have_event("cpu", "el-start")) |
fca32340a
|
1584 1585 |
err = parse_events(evsel_list, transaction_attrs, &errinfo); |
4cabc3d1c
|
1586 |
else |
fca32340a
|
1587 1588 1589 |
err = parse_events(evsel_list, transaction_limited_attrs, &errinfo); |
a454742c1
|
1590 |
if (err) { |
4cabc3d1c
|
1591 1592 |
fprintf(stderr, "Cannot set up transaction events "); |
a5cfa6217
|
1593 |
parse_events_print_error(&errinfo, transaction_attrs); |
4cabc3d1c
|
1594 1595 1596 1597 |
return -1; } return 0; } |
daefd0bc0
|
1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 |
if (smi_cost) { int smi; if (sysfs__read_int(FREEZE_ON_SMI_PATH, &smi) < 0) { fprintf(stderr, "freeze_on_smi is not supported. "); return -1; } if (!smi) { if (sysfs__write_int(FREEZE_ON_SMI_PATH, 1) < 0) { fprintf(stderr, "Failed to set freeze_on_smi. "); return -1; } smi_reset = true; } if (pmu_have_event("msr", "aperf") && pmu_have_event("msr", "smi")) { if (!force_metric_only) |
0ce5aa026
|
1619 |
stat_config.metric_only = true; |
a5cfa6217
|
1620 |
err = parse_events(evsel_list, smi_cost_attrs, &errinfo); |
daefd0bc0
|
1621 1622 1623 1624 |
} else { fprintf(stderr, "To measure SMI cost, it needs " "msr/aperf/, msr/smi/ and cpu/cycles/ support "); |
a5cfa6217
|
1625 |
parse_events_print_error(&errinfo, smi_cost_attrs); |
daefd0bc0
|
1626 1627 1628 |
return -1; } if (err) { |
a910e4666
|
1629 |
parse_events_print_error(&errinfo, smi_cost_attrs); |
daefd0bc0
|
1630 1631 1632 1633 1634 1635 |
fprintf(stderr, "Cannot set up SMI cost events "); return -1; } return 0; } |
44b1e60ab
|
1636 1637 1638 |
if (topdown_run) { char *str = NULL; bool warn = false; |
55c36a9fc
|
1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 |
if (!force_metric_only) stat_config.metric_only = true; if (topdown_filter_events(topdown_metric_attrs, &str, 1) < 0) { pr_err("Out of memory "); return -1; } if (topdown_metric_attrs[0] && str) { if (!stat_config.interval && !stat_config.metric_only) { fprintf(stat_config.output, "Topdown accuracy may decrease when measuring long periods. " "Please print the result regularly, e.g. -I1000 "); } goto setup_metrics; } zfree(&str); |
44b1e60ab
|
1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 |
if (stat_config.aggr_mode != AGGR_GLOBAL && stat_config.aggr_mode != AGGR_CORE) { pr_err("top down event configuration requires --per-core mode "); return -1; } stat_config.aggr_mode = AGGR_CORE; if (nr_cgroups || !target__has_cpu(&target)) { pr_err("top down event configuration requires system-wide mode (-a) "); return -1; } |
44b1e60ab
|
1671 1672 1673 1674 1675 1676 1677 1678 1679 |
if (topdown_filter_events(topdown_attrs, &str, arch_topdown_check_group(&warn)) < 0) { pr_err("Out of memory "); return -1; } if (topdown_attrs[0] && str) { if (warn) arch_topdown_group_warn(); |
55c36a9fc
|
1680 |
setup_metrics: |
a5cfa6217
|
1681 |
err = parse_events(evsel_list, str, &errinfo); |
44b1e60ab
|
1682 1683 1684 1685 1686 |
if (err) { fprintf(stderr, "Cannot set up top down events %s: %d ", str, err); |
a5cfa6217
|
1687 |
parse_events_print_error(&errinfo, str); |
c74b05030
|
1688 |
free(str); |
44b1e60ab
|
1689 1690 1691 1692 1693 1694 1695 1696 1697 |
return -1; } } else { fprintf(stderr, "System does not support topdown "); return -1; } free(str); } |
6484d2f9d
|
1698 |
if (!evsel_list->core.nr_entries) { |
a1f3d5676
|
1699 1700 |
if (target__has_cpu(&target)) default_attrs0[0].config = PERF_COUNT_SW_CPU_CLOCK; |
e251abee8
|
1701 |
if (evlist__add_default_attrs(evsel_list, default_attrs0) < 0) |
9dec4473a
|
1702 1703 |
return -1; if (pmu_have_event("cpu", "stalled-cycles-frontend")) { |
e251abee8
|
1704 |
if (evlist__add_default_attrs(evsel_list, frontend_attrs) < 0) |
9dec4473a
|
1705 1706 1707 |
return -1; } if (pmu_have_event("cpu", "stalled-cycles-backend")) { |
e251abee8
|
1708 |
if (evlist__add_default_attrs(evsel_list, backend_attrs) < 0) |
9dec4473a
|
1709 1710 |
return -1; } |
e251abee8
|
1711 |
if (evlist__add_default_attrs(evsel_list, default_attrs1) < 0) |
50d08e47b
|
1712 |
return -1; |
2cba3ffb9
|
1713 1714 1715 1716 1717 1718 1719 1720 |
} /* Detailed events get appended to the event list: */ if (detailed_run < 1) return 0; /* Append detailed run extra attributes: */ |
e251abee8
|
1721 |
if (evlist__add_default_attrs(evsel_list, detailed_attrs) < 0) |
50d08e47b
|
1722 |
return -1; |
2cba3ffb9
|
1723 1724 1725 1726 1727 |
if (detailed_run < 2) return 0; /* Append very detailed run extra attributes: */ |
e251abee8
|
1728 |
if (evlist__add_default_attrs(evsel_list, very_detailed_attrs) < 0) |
50d08e47b
|
1729 |
return -1; |
2cba3ffb9
|
1730 1731 1732 1733 1734 |
if (detailed_run < 3) return 0; /* Append very, very detailed run extra attributes: */ |
e251abee8
|
1735 |
return evlist__add_default_attrs(evsel_list, very_very_detailed_attrs); |
2cba3ffb9
|
1736 |
} |
8a59f3ccb
|
1737 |
static const char * const stat_record_usage[] = { |
4979d0c7d
|
1738 1739 1740 |
"perf stat record [<options>]", NULL, }; |
3ba78bd00
|
1741 1742 1743 1744 1745 1746 |
static void init_features(struct perf_session *session) { int feat; for (feat = HEADER_FIRST_FEATURE; feat < HEADER_LAST_FEATURE; feat++) perf_header__set_feat(&session->header, feat); |
8002a63f9
|
1747 |
perf_header__clear_feat(&session->header, HEADER_DIR_FORMAT); |
3ba78bd00
|
1748 1749 1750 1751 1752 |
perf_header__clear_feat(&session->header, HEADER_BUILD_ID); perf_header__clear_feat(&session->header, HEADER_TRACING_DATA); perf_header__clear_feat(&session->header, HEADER_BRANCH_STACK); perf_header__clear_feat(&session->header, HEADER_AUXTRACE); } |
4979d0c7d
|
1753 1754 1755 |
static int __cmd_record(int argc, const char **argv) { struct perf_session *session; |
8ceb41d7e
|
1756 |
struct perf_data *data = &perf_stat.data; |
4979d0c7d
|
1757 |
|
8a59f3ccb
|
1758 |
argc = parse_options(argc, argv, stat_options, stat_record_usage, |
4979d0c7d
|
1759 1760 1761 |
PARSE_OPT_STOP_AT_NON_OPTION); if (output_name) |
2d4f27999
|
1762 |
data->path = output_name; |
4979d0c7d
|
1763 |
|
d97ae04b3
|
1764 |
if (stat_config.run_count != 1 || forever) { |
e9d6db8e8
|
1765 1766 1767 1768 |
pr_err("Cannot use -r option with perf stat record. "); return -1; } |
8ceb41d7e
|
1769 |
session = perf_session__new(data, false, NULL); |
6ef81c55a
|
1770 1771 1772 1773 |
if (IS_ERR(session)) { pr_err("Perf session creation failed "); return PTR_ERR(session); |
4979d0c7d
|
1774 |
} |
3ba78bd00
|
1775 |
init_features(session); |
4979d0c7d
|
1776 1777 1778 1779 1780 |
session->evlist = evsel_list; perf_stat.session = session; perf_stat.record = true; return argc; } |
89f1688a5
|
1781 1782 |
static int process_stat_round_event(struct perf_session *session, union perf_event *event) |
a56f9390a
|
1783 |
{ |
72932371e
|
1784 |
struct perf_record_stat_round *stat_round = &event->stat_round; |
32dcd021d
|
1785 |
struct evsel *counter; |
a56f9390a
|
1786 1787 1788 |
struct timespec tsh, *ts = NULL; const char **argv = session->header.env.cmdline_argv; int argc = session->header.env.nr_cmdline; |
e5cadb93d
|
1789 |
evlist__for_each_entry(evsel_list, counter) |
a56f9390a
|
1790 |
perf_stat_process_counter(&stat_config, counter); |
e3b03b6c1
|
1791 1792 |
if (stat_round->type == PERF_STAT_ROUND_TYPE__FINAL) update_stats(&walltime_nsecs_stats, stat_round->time); |
a56f9390a
|
1793 |
|
e3b03b6c1
|
1794 |
if (stat_config.interval && stat_round->time) { |
bd48c63eb
|
1795 1796 |
tsh.tv_sec = stat_round->time / NSEC_PER_SEC; tsh.tv_nsec = stat_round->time % NSEC_PER_SEC; |
a56f9390a
|
1797 1798 1799 1800 1801 1802 |
ts = &tsh; } print_counters(ts, argc, argv); return 0; } |
62ba18ba9
|
1803 |
static |
89f1688a5
|
1804 1805 |
int process_stat_config_event(struct perf_session *session, union perf_event *event) |
62ba18ba9
|
1806 |
{ |
89f1688a5
|
1807 |
struct perf_tool *tool = session->tool; |
68d702f7a
|
1808 |
struct perf_stat *st = container_of(tool, struct perf_stat, tool); |
62ba18ba9
|
1809 |
perf_event__read_stat_config(&stat_config, &event->stat_config); |
68d702f7a
|
1810 |
|
315c0a1f0
|
1811 |
if (perf_cpu_map__empty(st->cpus)) { |
89af4e05c
|
1812 1813 1814 1815 1816 1817 1818 1819 |
if (st->aggr_mode != AGGR_UNSET) pr_warning("warning: processing task data, aggregation mode not set "); return 0; } if (st->aggr_mode != AGGR_UNSET) stat_config.aggr_mode = st->aggr_mode; |
8ceb41d7e
|
1820 |
if (perf_stat.data.is_pipe) |
68d702f7a
|
1821 1822 1823 |
perf_stat_init_aggr_mode(); else perf_stat_init_aggr_mode_file(st); |
62ba18ba9
|
1824 1825 |
return 0; } |
1975d36e1
|
1826 1827 1828 1829 1830 1831 1832 1833 |
static int set_maps(struct perf_stat *st) { if (!st->cpus || !st->threads) return 0; if (WARN_ONCE(st->maps_allocated, "stats double allocation ")) return -EINVAL; |
453fa0309
|
1834 |
perf_evlist__set_maps(&evsel_list->core, st->cpus, st->threads); |
1975d36e1
|
1835 1836 1837 1838 1839 1840 1841 1842 1843 |
if (perf_evlist__alloc_stats(evsel_list, true)) return -ENOMEM; st->maps_allocated = true; return 0; } static |
89f1688a5
|
1844 1845 |
int process_thread_map_event(struct perf_session *session, union perf_event *event) |
1975d36e1
|
1846 |
{ |
89f1688a5
|
1847 |
struct perf_tool *tool = session->tool; |
1975d36e1
|
1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 |
struct perf_stat *st = container_of(tool, struct perf_stat, tool); if (st->threads) { pr_warning("Extra thread map event, ignoring. "); return 0; } st->threads = thread_map__new_event(&event->thread_map); if (!st->threads) return -ENOMEM; return set_maps(st); } static |
89f1688a5
|
1864 1865 |
int process_cpu_map_event(struct perf_session *session, union perf_event *event) |
1975d36e1
|
1866 |
{ |
89f1688a5
|
1867 |
struct perf_tool *tool = session->tool; |
1975d36e1
|
1868 |
struct perf_stat *st = container_of(tool, struct perf_stat, tool); |
f854839ba
|
1869 |
struct perf_cpu_map *cpus; |
1975d36e1
|
1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 |
if (st->cpus) { pr_warning("Extra cpu map event, ignoring. "); return 0; } cpus = cpu_map__new_data(&event->cpu_map.data); if (!cpus) return -ENOMEM; st->cpus = cpus; return set_maps(st); } |
8a59f3ccb
|
1884 |
static const char * const stat_report_usage[] = { |
ba6039b6c
|
1885 1886 1887 1888 1889 1890 1891 |
"perf stat report [<options>]", NULL, }; static struct perf_stat perf_stat = { .tool = { .attr = perf_event__process_attr, |
fa6ea7817
|
1892 |
.event_update = perf_event__process_event_update, |
1975d36e1
|
1893 1894 |
.thread_map = process_thread_map_event, .cpu_map = process_cpu_map_event, |
62ba18ba9
|
1895 |
.stat_config = process_stat_config_event, |
a56f9390a
|
1896 1897 |
.stat = perf_event__process_stat_event, .stat_round = process_stat_round_event, |
ba6039b6c
|
1898 |
}, |
89af4e05c
|
1899 |
.aggr_mode = AGGR_UNSET, |
ba6039b6c
|
1900 1901 1902 1903 1904 1905 1906 |
}; static int __cmd_report(int argc, const char **argv) { struct perf_session *session; const struct option options[] = { OPT_STRING('i', "input", &input_name, "file", "input file name"), |
89af4e05c
|
1907 1908 |
OPT_SET_UINT(0, "per-socket", &perf_stat.aggr_mode, "aggregate counts per processor socket", AGGR_SOCKET), |
db5742b68
|
1909 1910 |
OPT_SET_UINT(0, "per-die", &perf_stat.aggr_mode, "aggregate counts per processor die", AGGR_DIE), |
89af4e05c
|
1911 1912 |
OPT_SET_UINT(0, "per-core", &perf_stat.aggr_mode, "aggregate counts per physical processor core", AGGR_CORE), |
86895b480
|
1913 1914 |
OPT_SET_UINT(0, "per-node", &perf_stat.aggr_mode, "aggregate counts per numa node", AGGR_NODE), |
89af4e05c
|
1915 1916 |
OPT_SET_UINT('A', "no-aggr", &perf_stat.aggr_mode, "disable CPU count aggregation", AGGR_NONE), |
ba6039b6c
|
1917 1918 1919 1920 |
OPT_END() }; struct stat st; int ret; |
8a59f3ccb
|
1921 |
argc = parse_options(argc, argv, options, stat_report_usage, 0); |
ba6039b6c
|
1922 1923 1924 1925 1926 1927 1928 |
if (!input_name || !strlen(input_name)) { if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode)) input_name = "-"; else input_name = "perf.data"; } |
2d4f27999
|
1929 1930 |
perf_stat.data.path = input_name; perf_stat.data.mode = PERF_DATA_MODE_READ; |
ba6039b6c
|
1931 |
|
8ceb41d7e
|
1932 |
session = perf_session__new(&perf_stat.data, false, &perf_stat.tool); |
6ef81c55a
|
1933 1934 |
if (IS_ERR(session)) return PTR_ERR(session); |
ba6039b6c
|
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 |
perf_stat.session = session; stat_config.output = stderr; evsel_list = session->evlist; ret = perf_session__process_events(session); if (ret) return ret; perf_session__delete(session); return 0; } |
e3ba76dee
|
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 |
static void setup_system_wide(int forks) { /* * Make system wide (-a) the default target if * no target was specified and one of following * conditions is met: * * - there's no workload specified * - there is workload specified but all requested * events are system wide events */ if (!target__none(&target)) return; if (!forks) target.system_wide = true; else { |
32dcd021d
|
1964 |
struct evsel *counter; |
e3ba76dee
|
1965 1966 |
evlist__for_each_entry(evsel_list, counter) { |
002a3d690
|
1967 1968 |
if (!counter->core.system_wide && strcmp(counter->name, "duration_time")) { |
e3ba76dee
|
1969 |
return; |
002a3d690
|
1970 |
} |
e3ba76dee
|
1971 |
} |
6484d2f9d
|
1972 |
if (evsel_list->core.nr_entries) |
e3ba76dee
|
1973 1974 1975 |
target.system_wide = true; } } |
b0ad8ea66
|
1976 |
int cmd_stat(int argc, const char **argv) |
5242519b0
|
1977 |
{ |
b070a547f
|
1978 1979 1980 1981 |
const char * const stat_usage[] = { "perf stat [<options>] [<command>]", NULL }; |
cc03c5429
|
1982 |
int status = -EINVAL, run_idx; |
4aa9015f8
|
1983 |
const char *mode; |
5821522e9
|
1984 |
FILE *output = stderr; |
f1f8ad52f
|
1985 |
unsigned int interval, timeout; |
ba6039b6c
|
1986 |
const char * const stat_subcommands[] = { "record", "report" }; |
42202dd56
|
1987 |
|
5af52b51f
|
1988 |
setlocale(LC_ALL, ""); |
0f98b11c6
|
1989 |
evsel_list = evlist__new(); |
361c99a66
|
1990 1991 |
if (evsel_list == NULL) return -ENOMEM; |
1669e509e
|
1992 |
parse_events__shrink_config_terms(); |
51433ead1
|
1993 1994 1995 1996 1997 |
/* String-parsing callback-based options would segfault when negated */ set_option_flag(stat_options, 'e', "event", PARSE_OPT_NONEG); set_option_flag(stat_options, 'M', "metrics", PARSE_OPT_NONEG); set_option_flag(stat_options, 'G', "cgroup", PARSE_OPT_NONEG); |
4979d0c7d
|
1998 1999 2000 |
argc = parse_options_subcommand(argc, argv, stat_options, stat_subcommands, (const char **) stat_usage, PARSE_OPT_STOP_AT_NON_OPTION); |
37932c188
|
2001 |
perf_stat__collect_metric_expr(evsel_list); |
fb4605ba4
|
2002 |
perf_stat__init_shadow_stats(); |
4979d0c7d
|
2003 |
|
fa7070a38
|
2004 2005 2006 2007 |
if (stat_config.csv_sep) { stat_config.csv_output = true; if (!strcmp(stat_config.csv_sep, "\\t")) stat_config.csv_sep = "\t"; |
6edb78a21
|
2008 |
} else |
fa7070a38
|
2009 |
stat_config.csv_sep = DEFAULT_SEPARATOR; |
6edb78a21
|
2010 |
|
4979d0c7d
|
2011 2012 2013 2014 |
if (argc && !strncmp(argv[0], "rec", 3)) { argc = __cmd_record(argc, argv); if (argc < 0) return -1; |
ba6039b6c
|
2015 2016 |
} else if (argc && !strncmp(argv[0], "rep", 3)) return __cmd_report(argc, argv); |
d7470b6af
|
2017 |
|
ec0d3d1fd
|
2018 |
interval = stat_config.interval; |
f1f8ad52f
|
2019 |
timeout = stat_config.timeout; |
ec0d3d1fd
|
2020 |
|
4979d0c7d
|
2021 2022 2023 2024 |
/* * For record command the -o is already taken care of. */ if (!STAT_RECORD && output_name && strcmp(output_name, "-")) |
4aa9015f8
|
2025 |
output = NULL; |
56f3bae70
|
2026 2027 2028 |
if (output_name && output_fd) { fprintf(stderr, "cannot use both --output and --log-fd "); |
e05473111
|
2029 2030 |
parse_options_usage(stat_usage, stat_options, "o", 1); parse_options_usage(NULL, stat_options, "log-fd", 0); |
cc03c5429
|
2031 |
goto out; |
56f3bae70
|
2032 |
} |
fc3e4d077
|
2033 |
|
0ce5aa026
|
2034 |
if (stat_config.metric_only && stat_config.aggr_mode == AGGR_THREAD) { |
54b509160
|
2035 2036 2037 2038 |
fprintf(stderr, "--metric-only is not supported with --per-thread "); goto out; } |
d97ae04b3
|
2039 |
if (stat_config.metric_only && stat_config.run_count > 1) { |
54b509160
|
2040 2041 2042 2043 |
fprintf(stderr, "--metric-only is not supported with -r "); goto out; } |
54ac0b1bd
|
2044 |
if (stat_config.walltime_run_table && stat_config.run_count <= 1) { |
e55c14af4
|
2045 2046 2047 2048 2049 2050 |
fprintf(stderr, "--table is only supported with -r "); parse_options_usage(stat_usage, stat_options, "r", 1); parse_options_usage(NULL, stat_options, "table", 0); goto out; } |
fc3e4d077
|
2051 2052 2053 |
if (output_fd < 0) { fprintf(stderr, "argument to --log-fd must be a > 0 "); |
e05473111
|
2054 |
parse_options_usage(stat_usage, stat_options, "log-fd", 0); |
cc03c5429
|
2055 |
goto out; |
fc3e4d077
|
2056 |
} |
4aa9015f8
|
2057 2058 2059 2060 2061 2062 2063 |
if (!output) { struct timespec tm; mode = append_file ? "a" : "w"; output = fopen(output_name, mode); if (!output) { perror("failed to create output file"); |
fceda7feb
|
2064 |
return -1; |
4aa9015f8
|
2065 2066 2067 2068 |
} clock_gettime(CLOCK_REALTIME, &tm); fprintf(output, "# started on %s ", ctime(&tm.tv_sec)); |
fc3e4d077
|
2069 |
} else if (output_fd > 0) { |
56f3bae70
|
2070 2071 2072 2073 2074 2075 |
mode = append_file ? "a" : "w"; output = fdopen(output_fd, mode); if (!output) { perror("Failed opening logfd"); return -errno; } |
4aa9015f8
|
2076 |
} |
5821522e9
|
2077 |
stat_config.output = output; |
d7470b6af
|
2078 2079 2080 |
/* * let the spreadsheet do the pretty-printing */ |
fa7070a38
|
2081 |
if (stat_config.csv_output) { |
61a9f3242
|
2082 |
/* User explicitly passed -B? */ |
d7470b6af
|
2083 2084 2085 |
if (big_num_opt == 1) { fprintf(stderr, "-B option not supported with -x "); |
e05473111
|
2086 2087 |
parse_options_usage(stat_usage, stat_options, "B", 1); parse_options_usage(NULL, stat_options, "x", 1); |
cc03c5429
|
2088 |
goto out; |
d7470b6af
|
2089 |
} else /* Nope, so disable big number formatting */ |
34ff0866d
|
2090 |
stat_config.big_num = false; |
d7470b6af
|
2091 |
} else if (big_num_opt == 0) /* User passed --no-big-num */ |
34ff0866d
|
2092 |
stat_config.big_num = false; |
d7470b6af
|
2093 |
|
e3ba76dee
|
2094 |
setup_system_wide(argc); |
ac3063bd4
|
2095 |
|
0ce2da148
|
2096 2097 2098 2099 |
/* * Display user/system times only for single * run and when there's specified tracee. */ |
d97ae04b3
|
2100 |
if ((stat_config.run_count == 1) && target__none(&target)) |
8897a8916
|
2101 |
stat_config.ru_display = true; |
0ce2da148
|
2102 |
|
d97ae04b3
|
2103 |
if (stat_config.run_count < 0) { |
cc03c5429
|
2104 2105 |
pr_err("Run count must be a positive number "); |
e05473111
|
2106 |
parse_options_usage(stat_usage, stat_options, "r", 1); |
cc03c5429
|
2107 |
goto out; |
d97ae04b3
|
2108 |
} else if (stat_config.run_count == 0) { |
a7e191c37
|
2109 |
forever = true; |
d97ae04b3
|
2110 |
stat_config.run_count = 1; |
a7e191c37
|
2111 |
} |
ddcacfa0f
|
2112 |
|
54ac0b1bd
|
2113 2114 2115 |
if (stat_config.walltime_run_table) { stat_config.walltime_run = zalloc(stat_config.run_count * sizeof(stat_config.walltime_run[0])); if (!stat_config.walltime_run) { |
e55c14af4
|
2116 2117 2118 2119 |
pr_err("failed to setup -r option"); goto out; } } |
1d9f8d1b8
|
2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 |
if ((stat_config.aggr_mode == AGGR_THREAD) && !target__has_task(&target)) { if (!target.system_wide || target.cpu_list) { fprintf(stderr, "The --per-thread option is only " "available when monitoring via -p -t -a " "options or only --per-thread. "); parse_options_usage(NULL, stat_options, "p", 1); parse_options_usage(NULL, stat_options, "t", 1); goto out; } |
32b8af82e
|
2131 2132 2133 2134 2135 2136 |
} /* * no_aggr, cgroup are for system-wide only * --per-thread is aggregated per thread, we dont mix it with cpu mode */ |
421a50f3f
|
2137 2138 |
if (((stat_config.aggr_mode != AGGR_GLOBAL && stat_config.aggr_mode != AGGR_THREAD) || nr_cgroups) && |
602ad878d
|
2139 |
!target__has_cpu(&target)) { |
023695d96
|
2140 2141 2142 |
fprintf(stderr, "both cgroup and no-aggregation " "modes only available in system-wide mode "); |
e05473111
|
2143 2144 2145 |
parse_options_usage(stat_usage, stat_options, "G", 1); parse_options_usage(NULL, stat_options, "A", 1); parse_options_usage(NULL, stat_options, "a", 1); |
cc03c5429
|
2146 |
goto out; |
d7e7a451c
|
2147 |
} |
2cba3ffb9
|
2148 2149 |
if (add_default_attributes()) goto out; |
ddcacfa0f
|
2150 |
|
d1c5a0e86
|
2151 2152 2153 2154 2155 2156 2157 2158 |
if (stat_config.cgroup_list) { if (nr_cgroups > 0) { pr_err("--cgroup and --for-each-cgroup cannot be used together "); parse_options_usage(stat_usage, stat_options, "G", 1); parse_options_usage(NULL, stat_options, "for-each-cgroup", 0); goto out; } |
b214ba8c4
|
2159 |
if (evlist__expand_cgroup(evsel_list, stat_config.cgroup_list, |
89fb1ca2a
|
2160 |
&stat_config.metric_events, true) < 0) |
d1c5a0e86
|
2161 2162 |
goto out; } |
602ad878d
|
2163 |
target__validate(&target); |
5c98d466e
|
2164 |
|
1d9f8d1b8
|
2165 2166 |
if ((stat_config.aggr_mode == AGGR_THREAD) && (target.system_wide)) target.per_thread = true; |
77a6f014e
|
2167 |
if (perf_evlist__create_maps(evsel_list, &target) < 0) { |
602ad878d
|
2168 |
if (target__has_task(&target)) { |
77a6f014e
|
2169 2170 |
pr_err("Problems finding threads of monitor "); |
e05473111
|
2171 2172 |
parse_options_usage(stat_usage, stat_options, "p", 1); parse_options_usage(NULL, stat_options, "t", 1); |
602ad878d
|
2173 |
} else if (target__has_cpu(&target)) { |
77a6f014e
|
2174 |
perror("failed to parse CPUs map"); |
e05473111
|
2175 2176 |
parse_options_usage(stat_usage, stat_options, "C", 1); parse_options_usage(NULL, stat_options, "a", 1); |
cc03c5429
|
2177 2178 |
} goto out; |
60d567e2d
|
2179 |
} |
32b8af82e
|
2180 |
|
a9a179024
|
2181 |
evlist__check_cpu_maps(evsel_list); |
32b8af82e
|
2182 2183 2184 2185 |
/* * Initialize thread_map with comm names, * so we could print it out on output. */ |
56739444d
|
2186 |
if (stat_config.aggr_mode == AGGR_THREAD) { |
03617c22e
|
2187 |
thread_map__read_comms(evsel_list->core.threads); |
56739444d
|
2188 2189 |
if (target.system_wide) { if (runtime_stat_new(&stat_config, |
a2f354e3a
|
2190 |
perf_thread_map__nr(evsel_list->core.threads))) { |
56739444d
|
2191 2192 2193 2194 |
goto out; } } } |
32b8af82e
|
2195 |
|
86895b480
|
2196 2197 |
if (stat_config.aggr_mode == AGGR_NODE) cpu__setup_cpunode_map(); |
db06a269e
|
2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 |
if (stat_config.times && interval) interval_count = true; else if (stat_config.times && !interval) { pr_err("interval-count option should be used together with " "interval-print. "); parse_options_usage(stat_usage, stat_options, "interval-count", 0); parse_options_usage(stat_usage, stat_options, "I", 1); goto out; } |
c45c6ea2e
|
2208 |
|
f1f8ad52f
|
2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 |
if (timeout && timeout < 100) { if (timeout < 10) { pr_err("timeout must be >= 10ms. "); parse_options_usage(stat_usage, stat_options, "timeout", 0); goto out; } else pr_warning("timeout < 100ms. " "The overhead percentage could be high in some cases. " "Please proceed with caution. "); } if (timeout && interval) { pr_err("timeout option is not supported with interval-print. "); parse_options_usage(stat_usage, stat_options, "timeout", 0); parse_options_usage(stat_usage, stat_options, "I", 1); goto out; } |
d134ffb91
|
2228 |
if (perf_evlist__alloc_stats(evsel_list, interval)) |
03ad9747c
|
2229 |
goto out; |
d6d901c23
|
2230 |
|
86ee6e18f
|
2231 |
if (perf_stat_init_aggr_mode()) |
03ad9747c
|
2232 |
goto out; |
86ee6e18f
|
2233 |
|
58d7e993b
|
2234 |
/* |
7d9ad16af
|
2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 |
* Set sample_type to PERF_SAMPLE_IDENTIFIER, which should be harmless * while avoiding that older tools show confusing messages. * * However for pipe sessions we need to keep it zero, * because script's perf_evsel__check_attr is triggered * by attr->sample_type != 0, and we can't run it on * stat sessions. */ stat_config.identifier = !(STAT_RECORD && perf_stat.data.is_pipe); /* |
58d7e993b
|
2246 2247 2248 2249 2250 |
* We dont want to block the signals - that would cause * child tasks to inherit that and Ctrl-C would not work. * What we want is for Ctrl-C to work in the exec()-ed * task, but being ignored by perf stat itself: */ |
f7b7c26e0
|
2251 |
atexit(sig_atexit); |
a7e191c37
|
2252 2253 |
if (!forever) signal(SIGINT, skip_signal); |
13370a9b5
|
2254 |
signal(SIGCHLD, skip_signal); |
58d7e993b
|
2255 2256 |
signal(SIGALRM, skip_signal); signal(SIGABRT, skip_signal); |
27e9769aa
|
2257 2258 |
if (evlist__initialize_ctlfd(evsel_list, stat_config.ctl_fd, stat_config.ctl_fd_ack)) goto out; |
42202dd56
|
2259 |
status = 0; |
d97ae04b3
|
2260 2261 |
for (run_idx = 0; forever || run_idx < stat_config.run_count; run_idx++) { if (stat_config.run_count != 1 && verbose > 0) |
4aa9015f8
|
2262 2263 2264 |
fprintf(output, "[ perf stat: executing run #%d ... ] ", run_idx + 1); |
f9cef0a90
|
2265 |
|
b63fd11cc
|
2266 2267 |
if (run_idx != 0) perf_evlist__reset_prev_raw_counts(evsel_list); |
e55c14af4
|
2268 |
status = run_perf_stat(argc, argv, run_idx); |
443f2d5ba
|
2269 |
if (forever && status != -1 && !interval) { |
d4f63a474
|
2270 |
print_counters(NULL, argc, argv); |
254ecbc74
|
2271 |
perf_stat__reset_stats(); |
a7e191c37
|
2272 |
} |
42202dd56
|
2273 |
} |
c7e5b328a
|
2274 |
if (!forever && status != -1 && (!interval || stat_config.summary)) |
d4f63a474
|
2275 |
print_counters(NULL, argc, argv); |
d134ffb91
|
2276 |
|
27e9769aa
|
2277 |
evlist__finalize_ctlfd(evsel_list); |
4979d0c7d
|
2278 2279 2280 2281 2282 2283 2284 2285 |
if (STAT_RECORD) { /* * We synthesize the kernel mmap record just so that older tools * don't emit warnings about not being able to resolve symbols * due to /proc/sys/kernel/kptr_restrict settings and instear provide * a saner message about no samples being in the perf.data file. * * This also serves to suppress a warning about f_header.data.size == 0 |
8b99b1a4e
|
2286 2287 2288 2289 |
* in header.c at the moment 'perf stat record' gets introduced, which * is not really needed once we start adding the stat specific PERF_RECORD_ * records, but the need to suppress the kptr_restrict messages in older * tools remain -acme |
4979d0c7d
|
2290 |
*/ |
8ceb41d7e
|
2291 |
int fd = perf_data__fd(&perf_stat.data); |
4979d0c7d
|
2292 2293 2294 2295 2296 2297 2298 2299 |
int err = perf_event__synthesize_kernel_mmap((void *)&perf_stat, process_synthesized_event, &perf_stat.session->machines.host); if (err) { pr_warning("Couldn't synthesize the kernel mmap record, harmless, " "older tools may produce warnings about this file ."); } |
7aad0c32b
|
2300 2301 2302 2303 2304 |
if (!interval) { if (WRITE_STAT_ROUND_EVENT(walltime_nsecs_stats.max, FINAL)) pr_err("failed to write stat round event "); } |
8ceb41d7e
|
2305 |
if (!perf_stat.data.is_pipe) { |
664c98d4e
|
2306 2307 2308 |
perf_stat.session->header.data_size += perf_stat.bytes_written; perf_session__write_header(perf_stat.session, evsel_list, fd, true); } |
4979d0c7d
|
2309 |
|
750b4edeb
|
2310 |
evlist__close(evsel_list); |
4979d0c7d
|
2311 2312 |
perf_session__delete(perf_stat.session); } |
544c2ae7b
|
2313 |
perf_stat__exit_aggr_mode(); |
d134ffb91
|
2314 |
perf_evlist__free_stats(evsel_list); |
0015e2e10
|
2315 |
out: |
d8f9da240
|
2316 |
zfree(&stat_config.walltime_run); |
e55c14af4
|
2317 |
|
daefd0bc0
|
2318 2319 |
if (smi_cost && smi_reset) sysfs__write_int(FREEZE_ON_SMI_PATH, 0); |
c12995a55
|
2320 |
evlist__delete(evsel_list); |
56739444d
|
2321 |
|
9afe5658a
|
2322 |
metricgroup__rblist_exit(&stat_config.metric_events); |
56739444d
|
2323 |
runtime_stat_delete(&stat_config); |
ee7fe31e6
|
2324 |
evlist__close_control(stat_config.ctl_fd, stat_config.ctl_fd_ack, &stat_config.ctl_fd_close); |
56739444d
|
2325 |
|
42202dd56
|
2326 |
return status; |
ddcacfa0f
|
2327 |
} |