xref: /openbmc/linux/tools/perf/builtin-kwork.c (revision 12109610)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * builtin-kwork.c
4  *
5  * Copyright (c) 2022  Huawei Inc,  Yang Jihong <yangjihong1@huawei.com>
6  */
7 
8 #include "builtin.h"
9 
10 #include "util/data.h"
11 #include "util/evlist.h"
12 #include "util/evsel.h"
13 #include "util/header.h"
14 #include "util/kwork.h"
15 #include "util/debug.h"
16 #include "util/session.h"
17 #include "util/symbol.h"
18 #include "util/thread.h"
19 #include "util/string2.h"
20 #include "util/callchain.h"
21 #include "util/evsel_fprintf.h"
22 #include "util/util.h"
23 
24 #include <subcmd/pager.h>
25 #include <subcmd/parse-options.h>
26 #include <traceevent/event-parse.h>
27 
28 #include <errno.h>
29 #include <inttypes.h>
30 #include <signal.h>
31 #include <linux/err.h>
32 #include <linux/time64.h>
33 #include <linux/zalloc.h>
34 
35 /*
36  * report header elements width
37  */
38 #define PRINT_CPU_WIDTH 4
39 #define PRINT_COUNT_WIDTH 9
40 #define PRINT_RUNTIME_WIDTH 10
41 #define PRINT_LATENCY_WIDTH 10
42 #define PRINT_TIMESTAMP_WIDTH 17
43 #define PRINT_KWORK_NAME_WIDTH 30
44 #define RPINT_DECIMAL_WIDTH 3
45 #define PRINT_BRACKETPAIR_WIDTH 2
46 #define PRINT_TIME_UNIT_SEC_WIDTH 2
47 #define PRINT_TIME_UNIT_MESC_WIDTH 3
48 #define PRINT_RUNTIME_HEADER_WIDTH (PRINT_RUNTIME_WIDTH + PRINT_TIME_UNIT_MESC_WIDTH)
49 #define PRINT_LATENCY_HEADER_WIDTH (PRINT_LATENCY_WIDTH + PRINT_TIME_UNIT_MESC_WIDTH)
50 #define PRINT_TIMEHIST_CPU_WIDTH (PRINT_CPU_WIDTH + PRINT_BRACKETPAIR_WIDTH)
51 #define PRINT_TIMESTAMP_HEADER_WIDTH (PRINT_TIMESTAMP_WIDTH + PRINT_TIME_UNIT_SEC_WIDTH)
52 
53 struct sort_dimension {
54 	const char      *name;
55 	int             (*cmp)(struct kwork_work *l, struct kwork_work *r);
56 	struct          list_head list;
57 };
58 
59 static int id_cmp(struct kwork_work *l, struct kwork_work *r)
60 {
61 	if (l->cpu > r->cpu)
62 		return 1;
63 	if (l->cpu < r->cpu)
64 		return -1;
65 
66 	if (l->id > r->id)
67 		return 1;
68 	if (l->id < r->id)
69 		return -1;
70 
71 	return 0;
72 }
73 
74 static int count_cmp(struct kwork_work *l, struct kwork_work *r)
75 {
76 	if (l->nr_atoms > r->nr_atoms)
77 		return 1;
78 	if (l->nr_atoms < r->nr_atoms)
79 		return -1;
80 
81 	return 0;
82 }
83 
84 static int runtime_cmp(struct kwork_work *l, struct kwork_work *r)
85 {
86 	if (l->total_runtime > r->total_runtime)
87 		return 1;
88 	if (l->total_runtime < r->total_runtime)
89 		return -1;
90 
91 	return 0;
92 }
93 
94 static int max_runtime_cmp(struct kwork_work *l, struct kwork_work *r)
95 {
96 	if (l->max_runtime > r->max_runtime)
97 		return 1;
98 	if (l->max_runtime < r->max_runtime)
99 		return -1;
100 
101 	return 0;
102 }
103 
104 static int avg_latency_cmp(struct kwork_work *l, struct kwork_work *r)
105 {
106 	u64 avgl, avgr;
107 
108 	if (!r->nr_atoms)
109 		return 1;
110 	if (!l->nr_atoms)
111 		return -1;
112 
113 	avgl = l->total_latency / l->nr_atoms;
114 	avgr = r->total_latency / r->nr_atoms;
115 
116 	if (avgl > avgr)
117 		return 1;
118 	if (avgl < avgr)
119 		return -1;
120 
121 	return 0;
122 }
123 
124 static int max_latency_cmp(struct kwork_work *l, struct kwork_work *r)
125 {
126 	if (l->max_latency > r->max_latency)
127 		return 1;
128 	if (l->max_latency < r->max_latency)
129 		return -1;
130 
131 	return 0;
132 }
133 
134 static int sort_dimension__add(struct perf_kwork *kwork __maybe_unused,
135 			       const char *tok, struct list_head *list)
136 {
137 	size_t i;
138 	static struct sort_dimension max_sort_dimension = {
139 		.name = "max",
140 		.cmp  = max_runtime_cmp,
141 	};
142 	static struct sort_dimension id_sort_dimension = {
143 		.name = "id",
144 		.cmp  = id_cmp,
145 	};
146 	static struct sort_dimension runtime_sort_dimension = {
147 		.name = "runtime",
148 		.cmp  = runtime_cmp,
149 	};
150 	static struct sort_dimension count_sort_dimension = {
151 		.name = "count",
152 		.cmp  = count_cmp,
153 	};
154 	static struct sort_dimension avg_sort_dimension = {
155 		.name = "avg",
156 		.cmp  = avg_latency_cmp,
157 	};
158 	struct sort_dimension *available_sorts[] = {
159 		&id_sort_dimension,
160 		&max_sort_dimension,
161 		&count_sort_dimension,
162 		&runtime_sort_dimension,
163 		&avg_sort_dimension,
164 	};
165 
166 	if (kwork->report == KWORK_REPORT_LATENCY)
167 		max_sort_dimension.cmp = max_latency_cmp;
168 
169 	for (i = 0; i < ARRAY_SIZE(available_sorts); i++) {
170 		if (!strcmp(available_sorts[i]->name, tok)) {
171 			list_add_tail(&available_sorts[i]->list, list);
172 			return 0;
173 		}
174 	}
175 
176 	return -1;
177 }
178 
179 static void setup_sorting(struct perf_kwork *kwork,
180 			  const struct option *options,
181 			  const char * const usage_msg[])
182 {
183 	char *tmp, *tok, *str = strdup(kwork->sort_order);
184 
185 	for (tok = strtok_r(str, ", ", &tmp);
186 	     tok; tok = strtok_r(NULL, ", ", &tmp)) {
187 		if (sort_dimension__add(kwork, tok, &kwork->sort_list) < 0)
188 			usage_with_options_msg(usage_msg, options,
189 					       "Unknown --sort key: `%s'", tok);
190 	}
191 
192 	pr_debug("Sort order: %s\n", kwork->sort_order);
193 	free(str);
194 }
195 
196 static struct kwork_atom *atom_new(struct perf_kwork *kwork,
197 				   struct perf_sample *sample)
198 {
199 	unsigned long i;
200 	struct kwork_atom_page *page;
201 	struct kwork_atom *atom = NULL;
202 
203 	list_for_each_entry(page, &kwork->atom_page_list, list) {
204 		if (!bitmap_full(page->bitmap, NR_ATOM_PER_PAGE)) {
205 			i = find_first_zero_bit(page->bitmap, NR_ATOM_PER_PAGE);
206 			BUG_ON(i >= NR_ATOM_PER_PAGE);
207 			atom = &page->atoms[i];
208 			goto found_atom;
209 		}
210 	}
211 
212 	/*
213 	 * new page
214 	 */
215 	page = zalloc(sizeof(*page));
216 	if (page == NULL) {
217 		pr_err("Failed to zalloc kwork atom page\n");
218 		return NULL;
219 	}
220 
221 	i = 0;
222 	atom = &page->atoms[0];
223 	list_add_tail(&page->list, &kwork->atom_page_list);
224 
225 found_atom:
226 	__set_bit(i, page->bitmap);
227 	atom->time = sample->time;
228 	atom->prev = NULL;
229 	atom->page_addr = page;
230 	atom->bit_inpage = i;
231 	return atom;
232 }
233 
234 static void atom_free(struct kwork_atom *atom)
235 {
236 	if (atom->prev != NULL)
237 		atom_free(atom->prev);
238 
239 	__clear_bit(atom->bit_inpage,
240 		    ((struct kwork_atom_page *)atom->page_addr)->bitmap);
241 }
242 
243 static void atom_del(struct kwork_atom *atom)
244 {
245 	list_del(&atom->list);
246 	atom_free(atom);
247 }
248 
249 static int work_cmp(struct list_head *list,
250 		    struct kwork_work *l, struct kwork_work *r)
251 {
252 	int ret = 0;
253 	struct sort_dimension *sort;
254 
255 	BUG_ON(list_empty(list));
256 
257 	list_for_each_entry(sort, list, list) {
258 		ret = sort->cmp(l, r);
259 		if (ret)
260 			return ret;
261 	}
262 
263 	return ret;
264 }
265 
266 static struct kwork_work *work_search(struct rb_root_cached *root,
267 				      struct kwork_work *key,
268 				      struct list_head *sort_list)
269 {
270 	int cmp;
271 	struct kwork_work *work;
272 	struct rb_node *node = root->rb_root.rb_node;
273 
274 	while (node) {
275 		work = container_of(node, struct kwork_work, node);
276 		cmp = work_cmp(sort_list, key, work);
277 		if (cmp > 0)
278 			node = node->rb_left;
279 		else if (cmp < 0)
280 			node = node->rb_right;
281 		else {
282 			if (work->name == NULL)
283 				work->name = key->name;
284 			return work;
285 		}
286 	}
287 	return NULL;
288 }
289 
290 static void work_insert(struct rb_root_cached *root,
291 			struct kwork_work *key, struct list_head *sort_list)
292 {
293 	int cmp;
294 	bool leftmost = true;
295 	struct kwork_work *cur;
296 	struct rb_node **new = &(root->rb_root.rb_node), *parent = NULL;
297 
298 	while (*new) {
299 		cur = container_of(*new, struct kwork_work, node);
300 		parent = *new;
301 		cmp = work_cmp(sort_list, key, cur);
302 
303 		if (cmp > 0)
304 			new = &((*new)->rb_left);
305 		else {
306 			new = &((*new)->rb_right);
307 			leftmost = false;
308 		}
309 	}
310 
311 	rb_link_node(&key->node, parent, new);
312 	rb_insert_color_cached(&key->node, root, leftmost);
313 }
314 
315 static struct kwork_work *work_new(struct kwork_work *key)
316 {
317 	int i;
318 	struct kwork_work *work = zalloc(sizeof(*work));
319 
320 	if (work == NULL) {
321 		pr_err("Failed to zalloc kwork work\n");
322 		return NULL;
323 	}
324 
325 	for (i = 0; i < KWORK_TRACE_MAX; i++)
326 		INIT_LIST_HEAD(&work->atom_list[i]);
327 
328 	work->id = key->id;
329 	work->cpu = key->cpu;
330 	work->name = key->name;
331 	work->class = key->class;
332 	return work;
333 }
334 
335 static struct kwork_work *work_findnew(struct rb_root_cached *root,
336 				       struct kwork_work *key,
337 				       struct list_head *sort_list)
338 {
339 	struct kwork_work *work = work_search(root, key, sort_list);
340 
341 	if (work != NULL)
342 		return work;
343 
344 	work = work_new(key);
345 	if (work)
346 		work_insert(root, work, sort_list);
347 
348 	return work;
349 }
350 
351 static void profile_update_timespan(struct perf_kwork *kwork,
352 				    struct perf_sample *sample)
353 {
354 	if (!kwork->summary)
355 		return;
356 
357 	if ((kwork->timestart == 0) || (kwork->timestart > sample->time))
358 		kwork->timestart = sample->time;
359 
360 	if (kwork->timeend < sample->time)
361 		kwork->timeend = sample->time;
362 }
363 
364 static bool profile_event_match(struct perf_kwork *kwork,
365 				struct kwork_work *work,
366 				struct perf_sample *sample)
367 {
368 	int cpu = work->cpu;
369 	u64 time = sample->time;
370 	struct perf_time_interval *ptime = &kwork->ptime;
371 
372 	if ((kwork->cpu_list != NULL) && !test_bit(cpu, kwork->cpu_bitmap))
373 		return false;
374 
375 	if (((ptime->start != 0) && (ptime->start > time)) ||
376 	    ((ptime->end != 0) && (ptime->end < time)))
377 		return false;
378 
379 	if ((kwork->profile_name != NULL) &&
380 	    (work->name != NULL) &&
381 	    (strcmp(work->name, kwork->profile_name) != 0))
382 		return false;
383 
384 	profile_update_timespan(kwork, sample);
385 	return true;
386 }
387 
388 static int work_push_atom(struct perf_kwork *kwork,
389 			  struct kwork_class *class,
390 			  enum kwork_trace_type src_type,
391 			  enum kwork_trace_type dst_type,
392 			  struct evsel *evsel,
393 			  struct perf_sample *sample,
394 			  struct machine *machine,
395 			  struct kwork_work **ret_work)
396 {
397 	struct kwork_atom *atom, *dst_atom;
398 	struct kwork_work *work, key;
399 
400 	BUG_ON(class->work_init == NULL);
401 	class->work_init(class, &key, evsel, sample, machine);
402 
403 	atom = atom_new(kwork, sample);
404 	if (atom == NULL)
405 		return -1;
406 
407 	work = work_findnew(&class->work_root, &key, &kwork->cmp_id);
408 	if (work == NULL) {
409 		free(atom);
410 		return -1;
411 	}
412 
413 	if (!profile_event_match(kwork, work, sample))
414 		return 0;
415 
416 	if (dst_type < KWORK_TRACE_MAX) {
417 		dst_atom = list_last_entry_or_null(&work->atom_list[dst_type],
418 						   struct kwork_atom, list);
419 		if (dst_atom != NULL) {
420 			atom->prev = dst_atom;
421 			list_del(&dst_atom->list);
422 		}
423 	}
424 
425 	if (ret_work != NULL)
426 		*ret_work = work;
427 
428 	list_add_tail(&atom->list, &work->atom_list[src_type]);
429 
430 	return 0;
431 }
432 
433 static struct kwork_atom *work_pop_atom(struct perf_kwork *kwork,
434 					struct kwork_class *class,
435 					enum kwork_trace_type src_type,
436 					enum kwork_trace_type dst_type,
437 					struct evsel *evsel,
438 					struct perf_sample *sample,
439 					struct machine *machine,
440 					struct kwork_work **ret_work)
441 {
442 	struct kwork_atom *atom, *src_atom;
443 	struct kwork_work *work, key;
444 
445 	BUG_ON(class->work_init == NULL);
446 	class->work_init(class, &key, evsel, sample, machine);
447 
448 	work = work_findnew(&class->work_root, &key, &kwork->cmp_id);
449 	if (ret_work != NULL)
450 		*ret_work = work;
451 
452 	if (work == NULL)
453 		return NULL;
454 
455 	if (!profile_event_match(kwork, work, sample))
456 		return NULL;
457 
458 	atom = list_last_entry_or_null(&work->atom_list[dst_type],
459 				       struct kwork_atom, list);
460 	if (atom != NULL)
461 		return atom;
462 
463 	src_atom = atom_new(kwork, sample);
464 	if (src_atom != NULL)
465 		list_add_tail(&src_atom->list, &work->atom_list[src_type]);
466 	else {
467 		if (ret_work != NULL)
468 			*ret_work = NULL;
469 	}
470 
471 	return NULL;
472 }
473 
474 static void report_update_exit_event(struct kwork_work *work,
475 				     struct kwork_atom *atom,
476 				     struct perf_sample *sample)
477 {
478 	u64 delta;
479 	u64 exit_time = sample->time;
480 	u64 entry_time = atom->time;
481 
482 	if ((entry_time != 0) && (exit_time >= entry_time)) {
483 		delta = exit_time - entry_time;
484 		if ((delta > work->max_runtime) ||
485 		    (work->max_runtime == 0)) {
486 			work->max_runtime = delta;
487 			work->max_runtime_start = entry_time;
488 			work->max_runtime_end = exit_time;
489 		}
490 		work->total_runtime += delta;
491 		work->nr_atoms++;
492 	}
493 }
494 
495 static int report_entry_event(struct perf_kwork *kwork,
496 			      struct kwork_class *class,
497 			      struct evsel *evsel,
498 			      struct perf_sample *sample,
499 			      struct machine *machine)
500 {
501 	return work_push_atom(kwork, class, KWORK_TRACE_ENTRY,
502 			      KWORK_TRACE_MAX, evsel, sample,
503 			      machine, NULL);
504 }
505 
506 static int report_exit_event(struct perf_kwork *kwork,
507 			     struct kwork_class *class,
508 			     struct evsel *evsel,
509 			     struct perf_sample *sample,
510 			     struct machine *machine)
511 {
512 	struct kwork_atom *atom = NULL;
513 	struct kwork_work *work = NULL;
514 
515 	atom = work_pop_atom(kwork, class, KWORK_TRACE_EXIT,
516 			     KWORK_TRACE_ENTRY, evsel, sample,
517 			     machine, &work);
518 	if (work == NULL)
519 		return -1;
520 
521 	if (atom != NULL) {
522 		report_update_exit_event(work, atom, sample);
523 		atom_del(atom);
524 	}
525 
526 	return 0;
527 }
528 
529 static void latency_update_entry_event(struct kwork_work *work,
530 				       struct kwork_atom *atom,
531 				       struct perf_sample *sample)
532 {
533 	u64 delta;
534 	u64 entry_time = sample->time;
535 	u64 raise_time = atom->time;
536 
537 	if ((raise_time != 0) && (entry_time >= raise_time)) {
538 		delta = entry_time - raise_time;
539 		if ((delta > work->max_latency) ||
540 		    (work->max_latency == 0)) {
541 			work->max_latency = delta;
542 			work->max_latency_start = raise_time;
543 			work->max_latency_end = entry_time;
544 		}
545 		work->total_latency += delta;
546 		work->nr_atoms++;
547 	}
548 }
549 
550 static int latency_raise_event(struct perf_kwork *kwork,
551 			       struct kwork_class *class,
552 			       struct evsel *evsel,
553 			       struct perf_sample *sample,
554 			       struct machine *machine)
555 {
556 	return work_push_atom(kwork, class, KWORK_TRACE_RAISE,
557 			      KWORK_TRACE_MAX, evsel, sample,
558 			      machine, NULL);
559 }
560 
561 static int latency_entry_event(struct perf_kwork *kwork,
562 			       struct kwork_class *class,
563 			       struct evsel *evsel,
564 			       struct perf_sample *sample,
565 			       struct machine *machine)
566 {
567 	struct kwork_atom *atom = NULL;
568 	struct kwork_work *work = NULL;
569 
570 	atom = work_pop_atom(kwork, class, KWORK_TRACE_ENTRY,
571 			     KWORK_TRACE_RAISE, evsel, sample,
572 			     machine, &work);
573 	if (work == NULL)
574 		return -1;
575 
576 	if (atom != NULL) {
577 		latency_update_entry_event(work, atom, sample);
578 		atom_del(atom);
579 	}
580 
581 	return 0;
582 }
583 
584 static void timehist_save_callchain(struct perf_kwork *kwork,
585 				    struct perf_sample *sample,
586 				    struct evsel *evsel,
587 				    struct machine *machine)
588 {
589 	struct symbol *sym;
590 	struct thread *thread;
591 	struct callchain_cursor_node *node;
592 	struct callchain_cursor *cursor = &callchain_cursor;
593 
594 	if (!kwork->show_callchain || sample->callchain == NULL)
595 		return;
596 
597 	/* want main thread for process - has maps */
598 	thread = machine__findnew_thread(machine, sample->pid, sample->pid);
599 	if (thread == NULL) {
600 		pr_debug("Failed to get thread for pid %d\n", sample->pid);
601 		return;
602 	}
603 
604 	if (thread__resolve_callchain(thread, cursor, evsel, sample,
605 				      NULL, NULL, kwork->max_stack + 2) != 0) {
606 		pr_debug("Failed to resolve callchain, skipping\n");
607 		goto out_put;
608 	}
609 
610 	callchain_cursor_commit(cursor);
611 
612 	while (true) {
613 		node = callchain_cursor_current(cursor);
614 		if (node == NULL)
615 			break;
616 
617 		sym = node->ms.sym;
618 		if (sym) {
619 			if (!strcmp(sym->name, "__softirqentry_text_start") ||
620 			    !strcmp(sym->name, "__do_softirq"))
621 				sym->ignore = 1;
622 		}
623 
624 		callchain_cursor_advance(cursor);
625 	}
626 
627 out_put:
628 	thread__put(thread);
629 }
630 
631 static void timehist_print_event(struct perf_kwork *kwork,
632 				 struct kwork_work *work,
633 				 struct kwork_atom *atom,
634 				 struct perf_sample *sample,
635 				 struct addr_location *al)
636 {
637 	char entrytime[32], exittime[32];
638 	char kwork_name[PRINT_KWORK_NAME_WIDTH];
639 
640 	/*
641 	 * runtime start
642 	 */
643 	timestamp__scnprintf_usec(atom->time,
644 				  entrytime, sizeof(entrytime));
645 	printf(" %*s ", PRINT_TIMESTAMP_WIDTH, entrytime);
646 
647 	/*
648 	 * runtime end
649 	 */
650 	timestamp__scnprintf_usec(sample->time,
651 				  exittime, sizeof(exittime));
652 	printf(" %*s ", PRINT_TIMESTAMP_WIDTH, exittime);
653 
654 	/*
655 	 * cpu
656 	 */
657 	printf(" [%0*d] ", PRINT_CPU_WIDTH, work->cpu);
658 
659 	/*
660 	 * kwork name
661 	 */
662 	if (work->class && work->class->work_name) {
663 		work->class->work_name(work, kwork_name,
664 				       PRINT_KWORK_NAME_WIDTH);
665 		printf(" %-*s ", PRINT_KWORK_NAME_WIDTH, kwork_name);
666 	} else
667 		printf(" %-*s ", PRINT_KWORK_NAME_WIDTH, "");
668 
669 	/*
670 	 *runtime
671 	 */
672 	printf(" %*.*f ",
673 	       PRINT_RUNTIME_WIDTH, RPINT_DECIMAL_WIDTH,
674 	       (double)(sample->time - atom->time) / NSEC_PER_MSEC);
675 
676 	/*
677 	 * delaytime
678 	 */
679 	if (atom->prev != NULL)
680 		printf(" %*.*f ", PRINT_LATENCY_WIDTH, RPINT_DECIMAL_WIDTH,
681 		       (double)(atom->time - atom->prev->time) / NSEC_PER_MSEC);
682 	else
683 		printf(" %*s ", PRINT_LATENCY_WIDTH, " ");
684 
685 	/*
686 	 * callchain
687 	 */
688 	if (kwork->show_callchain) {
689 		printf(" ");
690 		sample__fprintf_sym(sample, al, 0,
691 				    EVSEL__PRINT_SYM | EVSEL__PRINT_ONELINE |
692 				    EVSEL__PRINT_CALLCHAIN_ARROW |
693 				    EVSEL__PRINT_SKIP_IGNORED,
694 				    &callchain_cursor, symbol_conf.bt_stop_list,
695 				    stdout);
696 	}
697 
698 	printf("\n");
699 }
700 
701 static int timehist_raise_event(struct perf_kwork *kwork,
702 				struct kwork_class *class,
703 				struct evsel *evsel,
704 				struct perf_sample *sample,
705 				struct machine *machine)
706 {
707 	return work_push_atom(kwork, class, KWORK_TRACE_RAISE,
708 			      KWORK_TRACE_MAX, evsel, sample,
709 			      machine, NULL);
710 }
711 
712 static int timehist_entry_event(struct perf_kwork *kwork,
713 				struct kwork_class *class,
714 				struct evsel *evsel,
715 				struct perf_sample *sample,
716 				struct machine *machine)
717 {
718 	int ret;
719 	struct kwork_work *work = NULL;
720 
721 	ret = work_push_atom(kwork, class, KWORK_TRACE_ENTRY,
722 			     KWORK_TRACE_RAISE, evsel, sample,
723 			     machine, &work);
724 	if (ret)
725 		return ret;
726 
727 	if (work != NULL)
728 		timehist_save_callchain(kwork, sample, evsel, machine);
729 
730 	return 0;
731 }
732 
733 static int timehist_exit_event(struct perf_kwork *kwork,
734 			       struct kwork_class *class,
735 			       struct evsel *evsel,
736 			       struct perf_sample *sample,
737 			       struct machine *machine)
738 {
739 	struct kwork_atom *atom = NULL;
740 	struct kwork_work *work = NULL;
741 	struct addr_location al;
742 
743 	if (machine__resolve(machine, &al, sample) < 0) {
744 		pr_debug("Problem processing event, skipping it\n");
745 		return -1;
746 	}
747 
748 	atom = work_pop_atom(kwork, class, KWORK_TRACE_EXIT,
749 			     KWORK_TRACE_ENTRY, evsel, sample,
750 			     machine, &work);
751 	if (work == NULL)
752 		return -1;
753 
754 	if (atom != NULL) {
755 		work->nr_atoms++;
756 		timehist_print_event(kwork, work, atom, sample, &al);
757 		atom_del(atom);
758 	}
759 
760 	return 0;
761 }
762 
763 static struct kwork_class kwork_irq;
764 static int process_irq_handler_entry_event(struct perf_tool *tool,
765 					   struct evsel *evsel,
766 					   struct perf_sample *sample,
767 					   struct machine *machine)
768 {
769 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
770 
771 	if (kwork->tp_handler->entry_event)
772 		return kwork->tp_handler->entry_event(kwork, &kwork_irq,
773 						      evsel, sample, machine);
774 	return 0;
775 }
776 
777 static int process_irq_handler_exit_event(struct perf_tool *tool,
778 					  struct evsel *evsel,
779 					  struct perf_sample *sample,
780 					  struct machine *machine)
781 {
782 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
783 
784 	if (kwork->tp_handler->exit_event)
785 		return kwork->tp_handler->exit_event(kwork, &kwork_irq,
786 						     evsel, sample, machine);
787 	return 0;
788 }
789 
790 const struct evsel_str_handler irq_tp_handlers[] = {
791 	{ "irq:irq_handler_entry", process_irq_handler_entry_event, },
792 	{ "irq:irq_handler_exit",  process_irq_handler_exit_event,  },
793 };
794 
795 static int irq_class_init(struct kwork_class *class,
796 			  struct perf_session *session)
797 {
798 	if (perf_session__set_tracepoints_handlers(session, irq_tp_handlers)) {
799 		pr_err("Failed to set irq tracepoints handlers\n");
800 		return -1;
801 	}
802 
803 	class->work_root = RB_ROOT_CACHED;
804 	return 0;
805 }
806 
807 static void irq_work_init(struct kwork_class *class,
808 			  struct kwork_work *work,
809 			  struct evsel *evsel,
810 			  struct perf_sample *sample,
811 			  struct machine *machine __maybe_unused)
812 {
813 	work->class = class;
814 	work->cpu = sample->cpu;
815 	work->id = evsel__intval(evsel, sample, "irq");
816 	work->name = evsel__strval(evsel, sample, "name");
817 }
818 
819 static void irq_work_name(struct kwork_work *work, char *buf, int len)
820 {
821 	snprintf(buf, len, "%s:%" PRIu64 "", work->name, work->id);
822 }
823 
824 static struct kwork_class kwork_irq = {
825 	.name           = "irq",
826 	.type           = KWORK_CLASS_IRQ,
827 	.nr_tracepoints = 2,
828 	.tp_handlers    = irq_tp_handlers,
829 	.class_init     = irq_class_init,
830 	.work_init      = irq_work_init,
831 	.work_name      = irq_work_name,
832 };
833 
834 static struct kwork_class kwork_softirq;
835 static int process_softirq_raise_event(struct perf_tool *tool,
836 				       struct evsel *evsel,
837 				       struct perf_sample *sample,
838 				       struct machine *machine)
839 {
840 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
841 
842 	if (kwork->tp_handler->raise_event)
843 		return kwork->tp_handler->raise_event(kwork, &kwork_softirq,
844 						      evsel, sample, machine);
845 
846 	return 0;
847 }
848 
849 static int process_softirq_entry_event(struct perf_tool *tool,
850 				       struct evsel *evsel,
851 				       struct perf_sample *sample,
852 				       struct machine *machine)
853 {
854 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
855 
856 	if (kwork->tp_handler->entry_event)
857 		return kwork->tp_handler->entry_event(kwork, &kwork_softirq,
858 						      evsel, sample, machine);
859 
860 	return 0;
861 }
862 
863 static int process_softirq_exit_event(struct perf_tool *tool,
864 				      struct evsel *evsel,
865 				      struct perf_sample *sample,
866 				      struct machine *machine)
867 {
868 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
869 
870 	if (kwork->tp_handler->exit_event)
871 		return kwork->tp_handler->exit_event(kwork, &kwork_softirq,
872 						     evsel, sample, machine);
873 
874 	return 0;
875 }
876 
877 const struct evsel_str_handler softirq_tp_handlers[] = {
878 	{ "irq:softirq_raise", process_softirq_raise_event, },
879 	{ "irq:softirq_entry", process_softirq_entry_event, },
880 	{ "irq:softirq_exit",  process_softirq_exit_event,  },
881 };
882 
883 static int softirq_class_init(struct kwork_class *class,
884 			      struct perf_session *session)
885 {
886 	if (perf_session__set_tracepoints_handlers(session,
887 						   softirq_tp_handlers)) {
888 		pr_err("Failed to set softirq tracepoints handlers\n");
889 		return -1;
890 	}
891 
892 	class->work_root = RB_ROOT_CACHED;
893 	return 0;
894 }
895 
896 static char *evsel__softirq_name(struct evsel *evsel, u64 num)
897 {
898 	char *name = NULL;
899 	bool found = false;
900 	struct tep_print_flag_sym *sym = NULL;
901 	struct tep_print_arg *args = evsel->tp_format->print_fmt.args;
902 
903 	if ((args == NULL) || (args->next == NULL))
904 		return NULL;
905 
906 	/* skip softirq field: "REC->vec" */
907 	for (sym = args->next->symbol.symbols; sym != NULL; sym = sym->next) {
908 		if ((eval_flag(sym->value) == (unsigned long long)num) &&
909 		    (strlen(sym->str) != 0)) {
910 			found = true;
911 			break;
912 		}
913 	}
914 
915 	if (!found)
916 		return NULL;
917 
918 	name = strdup(sym->str);
919 	if (name == NULL) {
920 		pr_err("Failed to copy symbol name\n");
921 		return NULL;
922 	}
923 	return name;
924 }
925 
926 static void softirq_work_init(struct kwork_class *class,
927 			      struct kwork_work *work,
928 			      struct evsel *evsel,
929 			      struct perf_sample *sample,
930 			      struct machine *machine __maybe_unused)
931 {
932 	u64 num = evsel__intval(evsel, sample, "vec");
933 
934 	work->id = num;
935 	work->class = class;
936 	work->cpu = sample->cpu;
937 	work->name = evsel__softirq_name(evsel, num);
938 }
939 
940 static void softirq_work_name(struct kwork_work *work, char *buf, int len)
941 {
942 	snprintf(buf, len, "(s)%s:%" PRIu64 "", work->name, work->id);
943 }
944 
945 static struct kwork_class kwork_softirq = {
946 	.name           = "softirq",
947 	.type           = KWORK_CLASS_SOFTIRQ,
948 	.nr_tracepoints = 3,
949 	.tp_handlers    = softirq_tp_handlers,
950 	.class_init     = softirq_class_init,
951 	.work_init      = softirq_work_init,
952 	.work_name      = softirq_work_name,
953 };
954 
955 static struct kwork_class kwork_workqueue;
956 static int process_workqueue_activate_work_event(struct perf_tool *tool,
957 						 struct evsel *evsel,
958 						 struct perf_sample *sample,
959 						 struct machine *machine)
960 {
961 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
962 
963 	if (kwork->tp_handler->raise_event)
964 		return kwork->tp_handler->raise_event(kwork, &kwork_workqueue,
965 						    evsel, sample, machine);
966 
967 	return 0;
968 }
969 
970 static int process_workqueue_execute_start_event(struct perf_tool *tool,
971 						 struct evsel *evsel,
972 						 struct perf_sample *sample,
973 						 struct machine *machine)
974 {
975 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
976 
977 	if (kwork->tp_handler->entry_event)
978 		return kwork->tp_handler->entry_event(kwork, &kwork_workqueue,
979 						    evsel, sample, machine);
980 
981 	return 0;
982 }
983 
984 static int process_workqueue_execute_end_event(struct perf_tool *tool,
985 					       struct evsel *evsel,
986 					       struct perf_sample *sample,
987 					       struct machine *machine)
988 {
989 	struct perf_kwork *kwork = container_of(tool, struct perf_kwork, tool);
990 
991 	if (kwork->tp_handler->exit_event)
992 		return kwork->tp_handler->exit_event(kwork, &kwork_workqueue,
993 						   evsel, sample, machine);
994 
995 	return 0;
996 }
997 
998 const struct evsel_str_handler workqueue_tp_handlers[] = {
999 	{ "workqueue:workqueue_activate_work", process_workqueue_activate_work_event, },
1000 	{ "workqueue:workqueue_execute_start", process_workqueue_execute_start_event, },
1001 	{ "workqueue:workqueue_execute_end",   process_workqueue_execute_end_event,   },
1002 };
1003 
1004 static int workqueue_class_init(struct kwork_class *class,
1005 				struct perf_session *session)
1006 {
1007 	if (perf_session__set_tracepoints_handlers(session,
1008 						   workqueue_tp_handlers)) {
1009 		pr_err("Failed to set workqueue tracepoints handlers\n");
1010 		return -1;
1011 	}
1012 
1013 	class->work_root = RB_ROOT_CACHED;
1014 	return 0;
1015 }
1016 
1017 static void workqueue_work_init(struct kwork_class *class,
1018 				struct kwork_work *work,
1019 				struct evsel *evsel,
1020 				struct perf_sample *sample,
1021 				struct machine *machine)
1022 {
1023 	char *modp = NULL;
1024 	unsigned long long function_addr = evsel__intval(evsel,
1025 							 sample, "function");
1026 
1027 	work->class = class;
1028 	work->cpu = sample->cpu;
1029 	work->id = evsel__intval(evsel, sample, "work");
1030 	work->name = function_addr == 0 ? NULL :
1031 		machine__resolve_kernel_addr(machine, &function_addr, &modp);
1032 }
1033 
1034 static void workqueue_work_name(struct kwork_work *work, char *buf, int len)
1035 {
1036 	if (work->name != NULL)
1037 		snprintf(buf, len, "(w)%s", work->name);
1038 	else
1039 		snprintf(buf, len, "(w)0x%" PRIx64, work->id);
1040 }
1041 
1042 static struct kwork_class kwork_workqueue = {
1043 	.name           = "workqueue",
1044 	.type           = KWORK_CLASS_WORKQUEUE,
1045 	.nr_tracepoints = 3,
1046 	.tp_handlers    = workqueue_tp_handlers,
1047 	.class_init     = workqueue_class_init,
1048 	.work_init      = workqueue_work_init,
1049 	.work_name      = workqueue_work_name,
1050 };
1051 
1052 static struct kwork_class *kwork_class_supported_list[KWORK_CLASS_MAX] = {
1053 	[KWORK_CLASS_IRQ]       = &kwork_irq,
1054 	[KWORK_CLASS_SOFTIRQ]   = &kwork_softirq,
1055 	[KWORK_CLASS_WORKQUEUE] = &kwork_workqueue,
1056 };
1057 
1058 static void print_separator(int len)
1059 {
1060 	printf(" %.*s\n", len, graph_dotted_line);
1061 }
1062 
1063 static int report_print_work(struct perf_kwork *kwork, struct kwork_work *work)
1064 {
1065 	int ret = 0;
1066 	char kwork_name[PRINT_KWORK_NAME_WIDTH];
1067 	char max_runtime_start[32], max_runtime_end[32];
1068 	char max_latency_start[32], max_latency_end[32];
1069 
1070 	printf(" ");
1071 
1072 	/*
1073 	 * kwork name
1074 	 */
1075 	if (work->class && work->class->work_name) {
1076 		work->class->work_name(work, kwork_name,
1077 				       PRINT_KWORK_NAME_WIDTH);
1078 		ret += printf(" %-*s |", PRINT_KWORK_NAME_WIDTH, kwork_name);
1079 	} else {
1080 		ret += printf(" %-*s |", PRINT_KWORK_NAME_WIDTH, "");
1081 	}
1082 
1083 	/*
1084 	 * cpu
1085 	 */
1086 	ret += printf(" %0*d |", PRINT_CPU_WIDTH, work->cpu);
1087 
1088 	/*
1089 	 * total runtime
1090 	 */
1091 	if (kwork->report == KWORK_REPORT_RUNTIME) {
1092 		ret += printf(" %*.*f ms |",
1093 			      PRINT_RUNTIME_WIDTH, RPINT_DECIMAL_WIDTH,
1094 			      (double)work->total_runtime / NSEC_PER_MSEC);
1095 	} else if (kwork->report == KWORK_REPORT_LATENCY) { // avg delay
1096 		ret += printf(" %*.*f ms |",
1097 			      PRINT_LATENCY_WIDTH, RPINT_DECIMAL_WIDTH,
1098 			      (double)work->total_latency /
1099 			      work->nr_atoms / NSEC_PER_MSEC);
1100 	}
1101 
1102 	/*
1103 	 * count
1104 	 */
1105 	ret += printf(" %*" PRIu64 " |", PRINT_COUNT_WIDTH, work->nr_atoms);
1106 
1107 	/*
1108 	 * max runtime, max runtime start, max runtime end
1109 	 */
1110 	if (kwork->report == KWORK_REPORT_RUNTIME) {
1111 		timestamp__scnprintf_usec(work->max_runtime_start,
1112 					  max_runtime_start,
1113 					  sizeof(max_runtime_start));
1114 		timestamp__scnprintf_usec(work->max_runtime_end,
1115 					  max_runtime_end,
1116 					  sizeof(max_runtime_end));
1117 		ret += printf(" %*.*f ms | %*s s | %*s s |",
1118 			      PRINT_RUNTIME_WIDTH, RPINT_DECIMAL_WIDTH,
1119 			      (double)work->max_runtime / NSEC_PER_MSEC,
1120 			      PRINT_TIMESTAMP_WIDTH, max_runtime_start,
1121 			      PRINT_TIMESTAMP_WIDTH, max_runtime_end);
1122 	}
1123 	/*
1124 	 * max delay, max delay start, max delay end
1125 	 */
1126 	else if (kwork->report == KWORK_REPORT_LATENCY) {
1127 		timestamp__scnprintf_usec(work->max_latency_start,
1128 					  max_latency_start,
1129 					  sizeof(max_latency_start));
1130 		timestamp__scnprintf_usec(work->max_latency_end,
1131 					  max_latency_end,
1132 					  sizeof(max_latency_end));
1133 		ret += printf(" %*.*f ms | %*s s | %*s s |",
1134 			      PRINT_LATENCY_WIDTH, RPINT_DECIMAL_WIDTH,
1135 			      (double)work->max_latency / NSEC_PER_MSEC,
1136 			      PRINT_TIMESTAMP_WIDTH, max_latency_start,
1137 			      PRINT_TIMESTAMP_WIDTH, max_latency_end);
1138 	}
1139 
1140 	printf("\n");
1141 	return ret;
1142 }
1143 
1144 static int report_print_header(struct perf_kwork *kwork)
1145 {
1146 	int ret;
1147 
1148 	printf("\n ");
1149 	ret = printf(" %-*s | %-*s |",
1150 		     PRINT_KWORK_NAME_WIDTH, "Kwork Name",
1151 		     PRINT_CPU_WIDTH, "Cpu");
1152 
1153 	if (kwork->report == KWORK_REPORT_RUNTIME) {
1154 		ret += printf(" %-*s |",
1155 			      PRINT_RUNTIME_HEADER_WIDTH, "Total Runtime");
1156 	} else if (kwork->report == KWORK_REPORT_LATENCY) {
1157 		ret += printf(" %-*s |",
1158 			      PRINT_LATENCY_HEADER_WIDTH, "Avg delay");
1159 	}
1160 
1161 	ret += printf(" %-*s |", PRINT_COUNT_WIDTH, "Count");
1162 
1163 	if (kwork->report == KWORK_REPORT_RUNTIME) {
1164 		ret += printf(" %-*s | %-*s | %-*s |",
1165 			      PRINT_RUNTIME_HEADER_WIDTH, "Max runtime",
1166 			      PRINT_TIMESTAMP_HEADER_WIDTH, "Max runtime start",
1167 			      PRINT_TIMESTAMP_HEADER_WIDTH, "Max runtime end");
1168 	} else if (kwork->report == KWORK_REPORT_LATENCY) {
1169 		ret += printf(" %-*s | %-*s | %-*s |",
1170 			      PRINT_LATENCY_HEADER_WIDTH, "Max delay",
1171 			      PRINT_TIMESTAMP_HEADER_WIDTH, "Max delay start",
1172 			      PRINT_TIMESTAMP_HEADER_WIDTH, "Max delay end");
1173 	}
1174 
1175 	printf("\n");
1176 	print_separator(ret);
1177 	return ret;
1178 }
1179 
1180 static void timehist_print_header(void)
1181 {
1182 	/*
1183 	 * header row
1184 	 */
1185 	printf(" %-*s  %-*s  %-*s  %-*s  %-*s  %-*s\n",
1186 	       PRINT_TIMESTAMP_WIDTH, "Runtime start",
1187 	       PRINT_TIMESTAMP_WIDTH, "Runtime end",
1188 	       PRINT_TIMEHIST_CPU_WIDTH, "Cpu",
1189 	       PRINT_KWORK_NAME_WIDTH, "Kwork name",
1190 	       PRINT_RUNTIME_WIDTH, "Runtime",
1191 	       PRINT_RUNTIME_WIDTH, "Delaytime");
1192 
1193 	/*
1194 	 * units row
1195 	 */
1196 	printf(" %-*s  %-*s  %-*s  %-*s  %-*s  %-*s\n",
1197 	       PRINT_TIMESTAMP_WIDTH, "",
1198 	       PRINT_TIMESTAMP_WIDTH, "",
1199 	       PRINT_TIMEHIST_CPU_WIDTH, "",
1200 	       PRINT_KWORK_NAME_WIDTH, "(TYPE)NAME:NUM",
1201 	       PRINT_RUNTIME_WIDTH, "(msec)",
1202 	       PRINT_RUNTIME_WIDTH, "(msec)");
1203 
1204 	/*
1205 	 * separator
1206 	 */
1207 	printf(" %.*s  %.*s  %.*s  %.*s  %.*s  %.*s\n",
1208 	       PRINT_TIMESTAMP_WIDTH, graph_dotted_line,
1209 	       PRINT_TIMESTAMP_WIDTH, graph_dotted_line,
1210 	       PRINT_TIMEHIST_CPU_WIDTH, graph_dotted_line,
1211 	       PRINT_KWORK_NAME_WIDTH, graph_dotted_line,
1212 	       PRINT_RUNTIME_WIDTH, graph_dotted_line,
1213 	       PRINT_RUNTIME_WIDTH, graph_dotted_line);
1214 }
1215 
1216 static void print_summary(struct perf_kwork *kwork)
1217 {
1218 	u64 time = kwork->timeend - kwork->timestart;
1219 
1220 	printf("  Total count            : %9" PRIu64 "\n", kwork->all_count);
1221 	printf("  Total runtime   (msec) : %9.3f (%.3f%% load average)\n",
1222 	       (double)kwork->all_runtime / NSEC_PER_MSEC,
1223 	       time == 0 ? 0 : (double)kwork->all_runtime / time);
1224 	printf("  Total time span (msec) : %9.3f\n",
1225 	       (double)time / NSEC_PER_MSEC);
1226 }
1227 
1228 static unsigned long long nr_list_entry(struct list_head *head)
1229 {
1230 	struct list_head *pos;
1231 	unsigned long long n = 0;
1232 
1233 	list_for_each(pos, head)
1234 		n++;
1235 
1236 	return n;
1237 }
1238 
1239 static void print_skipped_events(struct perf_kwork *kwork)
1240 {
1241 	int i;
1242 	const char *const kwork_event_str[] = {
1243 		[KWORK_TRACE_RAISE] = "raise",
1244 		[KWORK_TRACE_ENTRY] = "entry",
1245 		[KWORK_TRACE_EXIT]  = "exit",
1246 	};
1247 
1248 	if ((kwork->nr_skipped_events[KWORK_TRACE_MAX] != 0) &&
1249 	    (kwork->nr_events != 0)) {
1250 		printf("  INFO: %.3f%% skipped events (%" PRIu64 " including ",
1251 		       (double)kwork->nr_skipped_events[KWORK_TRACE_MAX] /
1252 		       (double)kwork->nr_events * 100.0,
1253 		       kwork->nr_skipped_events[KWORK_TRACE_MAX]);
1254 
1255 		for (i = 0; i < KWORK_TRACE_MAX; i++) {
1256 			printf("%" PRIu64 " %s%s",
1257 			       kwork->nr_skipped_events[i],
1258 			       kwork_event_str[i],
1259 			       (i == KWORK_TRACE_MAX - 1) ? ")\n" : ", ");
1260 		}
1261 	}
1262 
1263 	if (verbose > 0)
1264 		printf("  INFO: use %lld atom pages\n",
1265 		       nr_list_entry(&kwork->atom_page_list));
1266 }
1267 
1268 static void print_bad_events(struct perf_kwork *kwork)
1269 {
1270 	if ((kwork->nr_lost_events != 0) && (kwork->nr_events != 0)) {
1271 		printf("  INFO: %.3f%% lost events (%ld out of %ld, in %ld chunks)\n",
1272 		       (double)kwork->nr_lost_events /
1273 		       (double)kwork->nr_events * 100.0,
1274 		       kwork->nr_lost_events, kwork->nr_events,
1275 		       kwork->nr_lost_chunks);
1276 	}
1277 }
1278 
1279 static void work_sort(struct perf_kwork *kwork, struct kwork_class *class)
1280 {
1281 	struct rb_node *node;
1282 	struct kwork_work *data;
1283 	struct rb_root_cached *root = &class->work_root;
1284 
1285 	pr_debug("Sorting %s ...\n", class->name);
1286 	for (;;) {
1287 		node = rb_first_cached(root);
1288 		if (!node)
1289 			break;
1290 
1291 		rb_erase_cached(node, root);
1292 		data = rb_entry(node, struct kwork_work, node);
1293 		work_insert(&kwork->sorted_work_root,
1294 			       data, &kwork->sort_list);
1295 	}
1296 }
1297 
1298 static void perf_kwork__sort(struct perf_kwork *kwork)
1299 {
1300 	struct kwork_class *class;
1301 
1302 	list_for_each_entry(class, &kwork->class_list, list)
1303 		work_sort(kwork, class);
1304 }
1305 
1306 static int perf_kwork__check_config(struct perf_kwork *kwork,
1307 				    struct perf_session *session)
1308 {
1309 	int ret;
1310 	struct evsel *evsel;
1311 	struct kwork_class *class;
1312 
1313 	static struct trace_kwork_handler report_ops = {
1314 		.entry_event = report_entry_event,
1315 		.exit_event  = report_exit_event,
1316 	};
1317 	static struct trace_kwork_handler latency_ops = {
1318 		.raise_event = latency_raise_event,
1319 		.entry_event = latency_entry_event,
1320 	};
1321 	static struct trace_kwork_handler timehist_ops = {
1322 		.raise_event = timehist_raise_event,
1323 		.entry_event = timehist_entry_event,
1324 		.exit_event  = timehist_exit_event,
1325 	};
1326 
1327 	switch (kwork->report) {
1328 	case KWORK_REPORT_RUNTIME:
1329 		kwork->tp_handler = &report_ops;
1330 		break;
1331 	case KWORK_REPORT_LATENCY:
1332 		kwork->tp_handler = &latency_ops;
1333 		break;
1334 	case KWORK_REPORT_TIMEHIST:
1335 		kwork->tp_handler = &timehist_ops;
1336 		break;
1337 	default:
1338 		pr_debug("Invalid report type %d\n", kwork->report);
1339 		return -1;
1340 	}
1341 
1342 	list_for_each_entry(class, &kwork->class_list, list)
1343 		if ((class->class_init != NULL) &&
1344 		    (class->class_init(class, session) != 0))
1345 			return -1;
1346 
1347 	if (kwork->cpu_list != NULL) {
1348 		ret = perf_session__cpu_bitmap(session,
1349 					       kwork->cpu_list,
1350 					       kwork->cpu_bitmap);
1351 		if (ret < 0) {
1352 			pr_err("Invalid cpu bitmap\n");
1353 			return -1;
1354 		}
1355 	}
1356 
1357 	if (kwork->time_str != NULL) {
1358 		ret = perf_time__parse_str(&kwork->ptime, kwork->time_str);
1359 		if (ret != 0) {
1360 			pr_err("Invalid time span\n");
1361 			return -1;
1362 		}
1363 	}
1364 
1365 	list_for_each_entry(evsel, &session->evlist->core.entries, core.node) {
1366 		if (kwork->show_callchain && !evsel__has_callchain(evsel)) {
1367 			pr_debug("Samples do not have callchains\n");
1368 			kwork->show_callchain = 0;
1369 			symbol_conf.use_callchain = 0;
1370 		}
1371 	}
1372 
1373 	return 0;
1374 }
1375 
1376 static int perf_kwork__read_events(struct perf_kwork *kwork)
1377 {
1378 	int ret = -1;
1379 	struct perf_session *session = NULL;
1380 
1381 	struct perf_data data = {
1382 		.path  = input_name,
1383 		.mode  = PERF_DATA_MODE_READ,
1384 		.force = kwork->force,
1385 	};
1386 
1387 	session = perf_session__new(&data, &kwork->tool);
1388 	if (IS_ERR(session)) {
1389 		pr_debug("Error creating perf session\n");
1390 		return PTR_ERR(session);
1391 	}
1392 
1393 	symbol__init(&session->header.env);
1394 
1395 	if (perf_kwork__check_config(kwork, session) != 0)
1396 		goto out_delete;
1397 
1398 	if (session->tevent.pevent &&
1399 	    tep_set_function_resolver(session->tevent.pevent,
1400 				      machine__resolve_kernel_addr,
1401 				      &session->machines.host) < 0) {
1402 		pr_err("Failed to set libtraceevent function resolver\n");
1403 		goto out_delete;
1404 	}
1405 
1406 	if (kwork->report == KWORK_REPORT_TIMEHIST)
1407 		timehist_print_header();
1408 
1409 	ret = perf_session__process_events(session);
1410 	if (ret) {
1411 		pr_debug("Failed to process events, error %d\n", ret);
1412 		goto out_delete;
1413 	}
1414 
1415 	kwork->nr_events      = session->evlist->stats.nr_events[0];
1416 	kwork->nr_lost_events = session->evlist->stats.total_lost;
1417 	kwork->nr_lost_chunks = session->evlist->stats.nr_events[PERF_RECORD_LOST];
1418 
1419 out_delete:
1420 	perf_session__delete(session);
1421 	return ret;
1422 }
1423 
1424 static void process_skipped_events(struct perf_kwork *kwork,
1425 				   struct kwork_work *work)
1426 {
1427 	int i;
1428 	unsigned long long count;
1429 
1430 	for (i = 0; i < KWORK_TRACE_MAX; i++) {
1431 		count = nr_list_entry(&work->atom_list[i]);
1432 		kwork->nr_skipped_events[i] += count;
1433 		kwork->nr_skipped_events[KWORK_TRACE_MAX] += count;
1434 	}
1435 }
1436 
1437 struct kwork_work *perf_kwork_add_work(struct perf_kwork *kwork,
1438 				       struct kwork_class *class,
1439 				       struct kwork_work *key)
1440 {
1441 	struct kwork_work *work = NULL;
1442 
1443 	work = work_new(key);
1444 	if (work == NULL)
1445 		return NULL;
1446 
1447 	work_insert(&class->work_root, work, &kwork->cmp_id);
1448 	return work;
1449 }
1450 
1451 static void sig_handler(int sig)
1452 {
1453 	/*
1454 	 * Simply capture termination signal so that
1455 	 * the program can continue after pause returns
1456 	 */
1457 	pr_debug("Captuer signal %d\n", sig);
1458 }
1459 
1460 static int perf_kwork__report_bpf(struct perf_kwork *kwork)
1461 {
1462 	int ret;
1463 
1464 	signal(SIGINT, sig_handler);
1465 	signal(SIGTERM, sig_handler);
1466 
1467 	ret = perf_kwork__trace_prepare_bpf(kwork);
1468 	if (ret)
1469 		return -1;
1470 
1471 	printf("Starting trace, Hit <Ctrl+C> to stop and report\n");
1472 
1473 	perf_kwork__trace_start();
1474 
1475 	/*
1476 	 * a simple pause, wait here for stop signal
1477 	 */
1478 	pause();
1479 
1480 	perf_kwork__trace_finish();
1481 
1482 	perf_kwork__report_read_bpf(kwork);
1483 
1484 	perf_kwork__report_cleanup_bpf();
1485 
1486 	return 0;
1487 }
1488 
1489 static int perf_kwork__report(struct perf_kwork *kwork)
1490 {
1491 	int ret;
1492 	struct rb_node *next;
1493 	struct kwork_work *work;
1494 
1495 	if (kwork->use_bpf)
1496 		ret = perf_kwork__report_bpf(kwork);
1497 	else
1498 		ret = perf_kwork__read_events(kwork);
1499 
1500 	if (ret != 0)
1501 		return -1;
1502 
1503 	perf_kwork__sort(kwork);
1504 
1505 	setup_pager();
1506 
1507 	ret = report_print_header(kwork);
1508 	next = rb_first_cached(&kwork->sorted_work_root);
1509 	while (next) {
1510 		work = rb_entry(next, struct kwork_work, node);
1511 		process_skipped_events(kwork, work);
1512 
1513 		if (work->nr_atoms != 0) {
1514 			report_print_work(kwork, work);
1515 			if (kwork->summary) {
1516 				kwork->all_runtime += work->total_runtime;
1517 				kwork->all_count += work->nr_atoms;
1518 			}
1519 		}
1520 		next = rb_next(next);
1521 	}
1522 	print_separator(ret);
1523 
1524 	if (kwork->summary) {
1525 		print_summary(kwork);
1526 		print_separator(ret);
1527 	}
1528 
1529 	print_bad_events(kwork);
1530 	print_skipped_events(kwork);
1531 	printf("\n");
1532 
1533 	return 0;
1534 }
1535 
1536 typedef int (*tracepoint_handler)(struct perf_tool *tool,
1537 				  struct evsel *evsel,
1538 				  struct perf_sample *sample,
1539 				  struct machine *machine);
1540 
1541 static int perf_kwork__process_tracepoint_sample(struct perf_tool *tool,
1542 						 union perf_event *event __maybe_unused,
1543 						 struct perf_sample *sample,
1544 						 struct evsel *evsel,
1545 						 struct machine *machine)
1546 {
1547 	int err = 0;
1548 
1549 	if (evsel->handler != NULL) {
1550 		tracepoint_handler f = evsel->handler;
1551 
1552 		err = f(tool, evsel, sample, machine);
1553 	}
1554 
1555 	return err;
1556 }
1557 
1558 static int perf_kwork__timehist(struct perf_kwork *kwork)
1559 {
1560 	/*
1561 	 * event handlers for timehist option
1562 	 */
1563 	kwork->tool.comm	 = perf_event__process_comm;
1564 	kwork->tool.exit	 = perf_event__process_exit;
1565 	kwork->tool.fork	 = perf_event__process_fork;
1566 	kwork->tool.attr	 = perf_event__process_attr;
1567 	kwork->tool.tracing_data = perf_event__process_tracing_data;
1568 	kwork->tool.build_id	 = perf_event__process_build_id;
1569 	kwork->tool.ordered_events = true;
1570 	kwork->tool.ordering_requires_timestamps = true;
1571 	symbol_conf.use_callchain = kwork->show_callchain;
1572 
1573 	if (symbol__validate_sym_arguments()) {
1574 		pr_err("Failed to validate sym arguments\n");
1575 		return -1;
1576 	}
1577 
1578 	setup_pager();
1579 
1580 	return perf_kwork__read_events(kwork);
1581 }
1582 
1583 static void setup_event_list(struct perf_kwork *kwork,
1584 			     const struct option *options,
1585 			     const char * const usage_msg[])
1586 {
1587 	int i;
1588 	struct kwork_class *class;
1589 	char *tmp, *tok, *str;
1590 
1591 	if (kwork->event_list_str == NULL)
1592 		goto null_event_list_str;
1593 
1594 	str = strdup(kwork->event_list_str);
1595 	for (tok = strtok_r(str, ", ", &tmp);
1596 	     tok; tok = strtok_r(NULL, ", ", &tmp)) {
1597 		for (i = 0; i < KWORK_CLASS_MAX; i++) {
1598 			class = kwork_class_supported_list[i];
1599 			if (strcmp(tok, class->name) == 0) {
1600 				list_add_tail(&class->list, &kwork->class_list);
1601 				break;
1602 			}
1603 		}
1604 		if (i == KWORK_CLASS_MAX) {
1605 			usage_with_options_msg(usage_msg, options,
1606 					       "Unknown --event key: `%s'", tok);
1607 		}
1608 	}
1609 	free(str);
1610 
1611 null_event_list_str:
1612 	/*
1613 	 * config all kwork events if not specified
1614 	 */
1615 	if (list_empty(&kwork->class_list)) {
1616 		for (i = 0; i < KWORK_CLASS_MAX; i++) {
1617 			list_add_tail(&kwork_class_supported_list[i]->list,
1618 				      &kwork->class_list);
1619 		}
1620 	}
1621 
1622 	pr_debug("Config event list:");
1623 	list_for_each_entry(class, &kwork->class_list, list)
1624 		pr_debug(" %s", class->name);
1625 	pr_debug("\n");
1626 }
1627 
1628 static int perf_kwork__record(struct perf_kwork *kwork,
1629 			      int argc, const char **argv)
1630 {
1631 	const char **rec_argv;
1632 	unsigned int rec_argc, i, j;
1633 	struct kwork_class *class;
1634 
1635 	const char *const record_args[] = {
1636 		"record",
1637 		"-a",
1638 		"-R",
1639 		"-m", "1024",
1640 		"-c", "1",
1641 	};
1642 
1643 	rec_argc = ARRAY_SIZE(record_args) + argc - 1;
1644 
1645 	list_for_each_entry(class, &kwork->class_list, list)
1646 		rec_argc += 2 * class->nr_tracepoints;
1647 
1648 	rec_argv = calloc(rec_argc + 1, sizeof(char *));
1649 	if (rec_argv == NULL)
1650 		return -ENOMEM;
1651 
1652 	for (i = 0; i < ARRAY_SIZE(record_args); i++)
1653 		rec_argv[i] = strdup(record_args[i]);
1654 
1655 	list_for_each_entry(class, &kwork->class_list, list) {
1656 		for (j = 0; j < class->nr_tracepoints; j++) {
1657 			rec_argv[i++] = strdup("-e");
1658 			rec_argv[i++] = strdup(class->tp_handlers[j].name);
1659 		}
1660 	}
1661 
1662 	for (j = 1; j < (unsigned int)argc; j++, i++)
1663 		rec_argv[i] = argv[j];
1664 
1665 	BUG_ON(i != rec_argc);
1666 
1667 	pr_debug("record comm: ");
1668 	for (j = 0; j < rec_argc; j++)
1669 		pr_debug("%s ", rec_argv[j]);
1670 	pr_debug("\n");
1671 
1672 	return cmd_record(i, rec_argv);
1673 }
1674 
1675 int cmd_kwork(int argc, const char **argv)
1676 {
1677 	static struct perf_kwork kwork = {
1678 		.class_list          = LIST_HEAD_INIT(kwork.class_list),
1679 		.tool = {
1680 			.mmap    = perf_event__process_mmap,
1681 			.mmap2   = perf_event__process_mmap2,
1682 			.sample  = perf_kwork__process_tracepoint_sample,
1683 		},
1684 		.atom_page_list      = LIST_HEAD_INIT(kwork.atom_page_list),
1685 		.sort_list           = LIST_HEAD_INIT(kwork.sort_list),
1686 		.cmp_id              = LIST_HEAD_INIT(kwork.cmp_id),
1687 		.sorted_work_root    = RB_ROOT_CACHED,
1688 		.tp_handler          = NULL,
1689 		.profile_name        = NULL,
1690 		.cpu_list            = NULL,
1691 		.time_str            = NULL,
1692 		.force               = false,
1693 		.event_list_str      = NULL,
1694 		.summary             = false,
1695 		.sort_order          = NULL,
1696 		.show_callchain      = false,
1697 		.max_stack           = 5,
1698 		.timestart           = 0,
1699 		.timeend             = 0,
1700 		.nr_events           = 0,
1701 		.nr_lost_chunks      = 0,
1702 		.nr_lost_events      = 0,
1703 		.all_runtime         = 0,
1704 		.all_count           = 0,
1705 		.nr_skipped_events   = { 0 },
1706 	};
1707 	static const char default_report_sort_order[] = "runtime, max, count";
1708 	static const char default_latency_sort_order[] = "avg, max, count";
1709 	const struct option kwork_options[] = {
1710 	OPT_INCR('v', "verbose", &verbose,
1711 		 "be more verbose (show symbol address, etc)"),
1712 	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
1713 		    "dump raw trace in ASCII"),
1714 	OPT_STRING('k', "kwork", &kwork.event_list_str, "kwork",
1715 		   "list of kwork to profile (irq, softirq, workqueue, etc)"),
1716 	OPT_BOOLEAN('f', "force", &kwork.force, "don't complain, do it"),
1717 	OPT_END()
1718 	};
1719 	const struct option report_options[] = {
1720 	OPT_STRING('s', "sort", &kwork.sort_order, "key[,key2...]",
1721 		   "sort by key(s): runtime, max, count"),
1722 	OPT_STRING('C', "cpu", &kwork.cpu_list, "cpu",
1723 		   "list of cpus to profile"),
1724 	OPT_STRING('n', "name", &kwork.profile_name, "name",
1725 		   "event name to profile"),
1726 	OPT_STRING(0, "time", &kwork.time_str, "str",
1727 		   "Time span for analysis (start,stop)"),
1728 	OPT_STRING('i', "input", &input_name, "file",
1729 		   "input file name"),
1730 	OPT_BOOLEAN('S', "with-summary", &kwork.summary,
1731 		    "Show summary with statistics"),
1732 #ifdef HAVE_BPF_SKEL
1733 	OPT_BOOLEAN('b', "use-bpf", &kwork.use_bpf,
1734 		    "Use BPF to measure kwork runtime"),
1735 #endif
1736 	OPT_PARENT(kwork_options)
1737 	};
1738 	const struct option latency_options[] = {
1739 	OPT_STRING('s', "sort", &kwork.sort_order, "key[,key2...]",
1740 		   "sort by key(s): avg, max, count"),
1741 	OPT_STRING('C', "cpu", &kwork.cpu_list, "cpu",
1742 		   "list of cpus to profile"),
1743 	OPT_STRING('n', "name", &kwork.profile_name, "name",
1744 		   "event name to profile"),
1745 	OPT_STRING(0, "time", &kwork.time_str, "str",
1746 		   "Time span for analysis (start,stop)"),
1747 	OPT_STRING('i', "input", &input_name, "file",
1748 		   "input file name"),
1749 #ifdef HAVE_BPF_SKEL
1750 	OPT_BOOLEAN('b', "use-bpf", &kwork.use_bpf,
1751 		    "Use BPF to measure kwork latency"),
1752 #endif
1753 	OPT_PARENT(kwork_options)
1754 	};
1755 	const struct option timehist_options[] = {
1756 	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
1757 		   "file", "vmlinux pathname"),
1758 	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
1759 		   "file", "kallsyms pathname"),
1760 	OPT_BOOLEAN('g', "call-graph", &kwork.show_callchain,
1761 		    "Display call chains if present"),
1762 	OPT_UINTEGER(0, "max-stack", &kwork.max_stack,
1763 		   "Maximum number of functions to display backtrace."),
1764 	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
1765 		    "Look for files with symbols relative to this directory"),
1766 	OPT_STRING(0, "time", &kwork.time_str, "str",
1767 		   "Time span for analysis (start,stop)"),
1768 	OPT_STRING('C', "cpu", &kwork.cpu_list, "cpu",
1769 		   "list of cpus to profile"),
1770 	OPT_STRING('n', "name", &kwork.profile_name, "name",
1771 		   "event name to profile"),
1772 	OPT_STRING('i', "input", &input_name, "file",
1773 		   "input file name"),
1774 	OPT_PARENT(kwork_options)
1775 	};
1776 	const char *kwork_usage[] = {
1777 		NULL,
1778 		NULL
1779 	};
1780 	const char * const report_usage[] = {
1781 		"perf kwork report [<options>]",
1782 		NULL
1783 	};
1784 	const char * const latency_usage[] = {
1785 		"perf kwork latency [<options>]",
1786 		NULL
1787 	};
1788 	const char * const timehist_usage[] = {
1789 		"perf kwork timehist [<options>]",
1790 		NULL
1791 	};
1792 	const char *const kwork_subcommands[] = {
1793 		"record", "report", "latency", "timehist", NULL
1794 	};
1795 
1796 	argc = parse_options_subcommand(argc, argv, kwork_options,
1797 					kwork_subcommands, kwork_usage,
1798 					PARSE_OPT_STOP_AT_NON_OPTION);
1799 	if (!argc)
1800 		usage_with_options(kwork_usage, kwork_options);
1801 
1802 	setup_event_list(&kwork, kwork_options, kwork_usage);
1803 	sort_dimension__add(&kwork, "id", &kwork.cmp_id);
1804 
1805 	if (strlen(argv[0]) > 2 && strstarts("record", argv[0]))
1806 		return perf_kwork__record(&kwork, argc, argv);
1807 	else if (strlen(argv[0]) > 2 && strstarts("report", argv[0])) {
1808 		kwork.sort_order = default_report_sort_order;
1809 		if (argc > 1) {
1810 			argc = parse_options(argc, argv, report_options, report_usage, 0);
1811 			if (argc)
1812 				usage_with_options(report_usage, report_options);
1813 		}
1814 		kwork.report = KWORK_REPORT_RUNTIME;
1815 		setup_sorting(&kwork, report_options, report_usage);
1816 		return perf_kwork__report(&kwork);
1817 	} else if (strlen(argv[0]) > 2 && strstarts("latency", argv[0])) {
1818 		kwork.sort_order = default_latency_sort_order;
1819 		if (argc > 1) {
1820 			argc = parse_options(argc, argv, latency_options, latency_usage, 0);
1821 			if (argc)
1822 				usage_with_options(latency_usage, latency_options);
1823 		}
1824 		kwork.report = KWORK_REPORT_LATENCY;
1825 		setup_sorting(&kwork, latency_options, latency_usage);
1826 		return perf_kwork__report(&kwork);
1827 	} else if (strlen(argv[0]) > 2 && strstarts("timehist", argv[0])) {
1828 		if (argc > 1) {
1829 			argc = parse_options(argc, argv, timehist_options, timehist_usage, 0);
1830 			if (argc)
1831 				usage_with_options(timehist_usage, timehist_options);
1832 		}
1833 		kwork.report = KWORK_REPORT_TIMEHIST;
1834 		return perf_kwork__timehist(&kwork);
1835 	} else
1836 		usage_with_options(kwork_usage, kwork_options);
1837 
1838 	return 0;
1839 }
1840