]>
Commit | Line | Data |
---|---|---|
9b5e350c HM |
1 | #include "builtin.h" |
2 | #include "perf.h" | |
3 | ||
4 | #include "util/util.h" | |
5 | #include "util/cache.h" | |
6 | #include "util/symbol.h" | |
7 | #include "util/thread.h" | |
8 | #include "util/header.h" | |
9 | ||
10 | #include "util/parse-options.h" | |
11 | #include "util/trace-event.h" | |
12 | ||
13 | #include "util/debug.h" | |
14 | #include "util/session.h" | |
15 | ||
16 | #include <sys/types.h> | |
17 | #include <sys/prctl.h> | |
18 | #include <semaphore.h> | |
19 | #include <pthread.h> | |
20 | #include <math.h> | |
21 | #include <limits.h> | |
22 | ||
23 | #include <linux/list.h> | |
24 | #include <linux/hash.h> | |
25 | ||
e4cef1f6 HM |
26 | static struct perf_session *session; |
27 | ||
9b5e350c HM |
28 | /* based on kernel/lockdep.c */ |
29 | #define LOCKHASH_BITS 12 | |
30 | #define LOCKHASH_SIZE (1UL << LOCKHASH_BITS) | |
31 | ||
32 | static struct list_head lockhash_table[LOCKHASH_SIZE]; | |
33 | ||
34 | #define __lockhashfn(key) hash_long((unsigned long)key, LOCKHASH_BITS) | |
35 | #define lockhashentry(key) (lockhash_table + __lockhashfn((key))) | |
36 | ||
9b5e350c | 37 | struct lock_stat { |
59f411b6 IM |
38 | struct list_head hash_entry; |
39 | struct rb_node rb; /* used for sorting */ | |
9b5e350c | 40 | |
59f411b6 IM |
41 | /* |
42 | * FIXME: raw_field_value() returns unsigned long long, | |
9b5e350c | 43 | * so address of lockdep_map should be dealed as 64bit. |
59f411b6 IM |
44 | * Is there more better solution? |
45 | */ | |
46 | void *addr; /* address of lockdep_map, used as ID */ | |
47 | char *name; /* for strcpy(), we cannot use const */ | |
9b5e350c | 48 | |
59f411b6 | 49 | unsigned int nr_acquire; |
e4cef1f6 | 50 | unsigned int nr_acquired; |
59f411b6 IM |
51 | unsigned int nr_contended; |
52 | unsigned int nr_release; | |
9b5e350c | 53 | |
e4cef1f6 HM |
54 | unsigned int nr_readlock; |
55 | unsigned int nr_trylock; | |
9b5e350c | 56 | /* these times are in nano sec. */ |
59f411b6 IM |
57 | u64 wait_time_total; |
58 | u64 wait_time_min; | |
59 | u64 wait_time_max; | |
e4cef1f6 HM |
60 | |
61 | int discard; /* flag of blacklist */ | |
62 | }; | |
63 | ||
64 | /* | |
65 | * States of lock_seq_stat | |
66 | * | |
67 | * UNINITIALIZED is required for detecting first event of acquire. | |
68 | * As the nature of lock events, there is no guarantee | |
69 | * that the first event for the locks are acquire, | |
70 | * it can be acquired, contended or release. | |
71 | */ | |
72 | #define SEQ_STATE_UNINITIALIZED 0 /* initial state */ | |
73 | #define SEQ_STATE_RELEASED 1 | |
74 | #define SEQ_STATE_ACQUIRING 2 | |
75 | #define SEQ_STATE_ACQUIRED 3 | |
76 | #define SEQ_STATE_READ_ACQUIRED 4 | |
77 | #define SEQ_STATE_CONTENDED 5 | |
78 | ||
79 | /* | |
80 | * MAX_LOCK_DEPTH | |
81 | * Imported from include/linux/sched.h. | |
82 | * Should this be synchronized? | |
83 | */ | |
84 | #define MAX_LOCK_DEPTH 48 | |
85 | ||
86 | /* | |
87 | * struct lock_seq_stat: | |
88 | * Place to put on state of one lock sequence | |
89 | * 1) acquire -> acquired -> release | |
90 | * 2) acquire -> contended -> acquired -> release | |
91 | * 3) acquire (with read or try) -> release | |
92 | * 4) Are there other patterns? | |
93 | */ | |
94 | struct lock_seq_stat { | |
95 | struct list_head list; | |
96 | int state; | |
97 | u64 prev_event_time; | |
98 | void *addr; | |
99 | ||
100 | int read_count; | |
9b5e350c HM |
101 | }; |
102 | ||
e4cef1f6 HM |
103 | struct thread_stat { |
104 | struct rb_node rb; | |
105 | ||
106 | u32 tid; | |
107 | struct list_head seq_list; | |
108 | }; | |
109 | ||
110 | static struct rb_root thread_stats; | |
111 | ||
112 | static struct thread_stat *thread_stat_find(u32 tid) | |
113 | { | |
114 | struct rb_node *node; | |
115 | struct thread_stat *st; | |
116 | ||
117 | node = thread_stats.rb_node; | |
118 | while (node) { | |
119 | st = container_of(node, struct thread_stat, rb); | |
120 | if (st->tid == tid) | |
121 | return st; | |
122 | else if (tid < st->tid) | |
123 | node = node->rb_left; | |
124 | else | |
125 | node = node->rb_right; | |
126 | } | |
127 | ||
128 | return NULL; | |
129 | } | |
130 | ||
131 | static void thread_stat_insert(struct thread_stat *new) | |
132 | { | |
133 | struct rb_node **rb = &thread_stats.rb_node; | |
134 | struct rb_node *parent = NULL; | |
135 | struct thread_stat *p; | |
136 | ||
137 | while (*rb) { | |
138 | p = container_of(*rb, struct thread_stat, rb); | |
139 | parent = *rb; | |
140 | ||
141 | if (new->tid < p->tid) | |
142 | rb = &(*rb)->rb_left; | |
143 | else if (new->tid > p->tid) | |
144 | rb = &(*rb)->rb_right; | |
145 | else | |
146 | BUG_ON("inserting invalid thread_stat\n"); | |
147 | } | |
148 | ||
149 | rb_link_node(&new->rb, parent, rb); | |
150 | rb_insert_color(&new->rb, &thread_stats); | |
151 | } | |
152 | ||
153 | static struct thread_stat *thread_stat_findnew_after_first(u32 tid) | |
154 | { | |
155 | struct thread_stat *st; | |
156 | ||
157 | st = thread_stat_find(tid); | |
158 | if (st) | |
159 | return st; | |
160 | ||
161 | st = zalloc(sizeof(struct thread_stat)); | |
162 | if (!st) | |
163 | die("memory allocation failed\n"); | |
164 | ||
165 | st->tid = tid; | |
166 | INIT_LIST_HEAD(&st->seq_list); | |
167 | ||
168 | thread_stat_insert(st); | |
169 | ||
170 | return st; | |
171 | } | |
172 | ||
173 | static struct thread_stat *thread_stat_findnew_first(u32 tid); | |
174 | static struct thread_stat *(*thread_stat_findnew)(u32 tid) = | |
175 | thread_stat_findnew_first; | |
176 | ||
177 | static struct thread_stat *thread_stat_findnew_first(u32 tid) | |
178 | { | |
179 | struct thread_stat *st; | |
180 | ||
181 | st = zalloc(sizeof(struct thread_stat)); | |
182 | if (!st) | |
183 | die("memory allocation failed\n"); | |
184 | st->tid = tid; | |
185 | INIT_LIST_HEAD(&st->seq_list); | |
186 | ||
187 | rb_link_node(&st->rb, NULL, &thread_stats.rb_node); | |
188 | rb_insert_color(&st->rb, &thread_stats); | |
189 | ||
190 | thread_stat_findnew = thread_stat_findnew_after_first; | |
191 | return st; | |
192 | } | |
193 | ||
9b5e350c | 194 | /* build simple key function one is bigger than two */ |
59f411b6 | 195 | #define SINGLE_KEY(member) \ |
9b5e350c HM |
196 | static int lock_stat_key_ ## member(struct lock_stat *one, \ |
197 | struct lock_stat *two) \ | |
198 | { \ | |
199 | return one->member > two->member; \ | |
200 | } | |
201 | ||
202 | SINGLE_KEY(nr_acquired) | |
203 | SINGLE_KEY(nr_contended) | |
204 | SINGLE_KEY(wait_time_total) | |
205 | SINGLE_KEY(wait_time_min) | |
206 | SINGLE_KEY(wait_time_max) | |
207 | ||
208 | struct lock_key { | |
209 | /* | |
210 | * name: the value for specify by user | |
211 | * this should be simpler than raw name of member | |
212 | * e.g. nr_acquired -> acquired, wait_time_total -> wait_total | |
213 | */ | |
59f411b6 IM |
214 | const char *name; |
215 | int (*key)(struct lock_stat*, struct lock_stat*); | |
9b5e350c HM |
216 | }; |
217 | ||
59f411b6 IM |
218 | static const char *sort_key = "acquired"; |
219 | ||
220 | static int (*compare)(struct lock_stat *, struct lock_stat *); | |
221 | ||
222 | static struct rb_root result; /* place to store sorted data */ | |
9b5e350c HM |
223 | |
224 | #define DEF_KEY_LOCK(name, fn_suffix) \ | |
225 | { #name, lock_stat_key_ ## fn_suffix } | |
226 | struct lock_key keys[] = { | |
227 | DEF_KEY_LOCK(acquired, nr_acquired), | |
228 | DEF_KEY_LOCK(contended, nr_contended), | |
229 | DEF_KEY_LOCK(wait_total, wait_time_total), | |
230 | DEF_KEY_LOCK(wait_min, wait_time_min), | |
231 | DEF_KEY_LOCK(wait_max, wait_time_max), | |
232 | ||
233 | /* extra comparisons much complicated should be here */ | |
234 | ||
235 | { NULL, NULL } | |
236 | }; | |
237 | ||
238 | static void select_key(void) | |
239 | { | |
240 | int i; | |
241 | ||
242 | for (i = 0; keys[i].name; i++) { | |
243 | if (!strcmp(keys[i].name, sort_key)) { | |
244 | compare = keys[i].key; | |
245 | return; | |
246 | } | |
247 | } | |
248 | ||
249 | die("Unknown compare key:%s\n", sort_key); | |
250 | } | |
251 | ||
9b5e350c | 252 | static void insert_to_result(struct lock_stat *st, |
59f411b6 | 253 | int (*bigger)(struct lock_stat *, struct lock_stat *)) |
9b5e350c HM |
254 | { |
255 | struct rb_node **rb = &result.rb_node; | |
256 | struct rb_node *parent = NULL; | |
257 | struct lock_stat *p; | |
258 | ||
259 | while (*rb) { | |
260 | p = container_of(*rb, struct lock_stat, rb); | |
261 | parent = *rb; | |
262 | ||
263 | if (bigger(st, p)) | |
264 | rb = &(*rb)->rb_left; | |
265 | else | |
266 | rb = &(*rb)->rb_right; | |
267 | } | |
268 | ||
269 | rb_link_node(&st->rb, parent, rb); | |
270 | rb_insert_color(&st->rb, &result); | |
271 | } | |
272 | ||
273 | /* returns left most element of result, and erase it */ | |
274 | static struct lock_stat *pop_from_result(void) | |
275 | { | |
276 | struct rb_node *node = result.rb_node; | |
277 | ||
278 | if (!node) | |
279 | return NULL; | |
280 | ||
281 | while (node->rb_left) | |
282 | node = node->rb_left; | |
283 | ||
284 | rb_erase(node, &result); | |
285 | return container_of(node, struct lock_stat, rb); | |
286 | } | |
287 | ||
59f411b6 | 288 | static struct lock_stat *lock_stat_findnew(void *addr, const char *name) |
9b5e350c HM |
289 | { |
290 | struct list_head *entry = lockhashentry(addr); | |
291 | struct lock_stat *ret, *new; | |
292 | ||
293 | list_for_each_entry(ret, entry, hash_entry) { | |
294 | if (ret->addr == addr) | |
295 | return ret; | |
296 | } | |
297 | ||
298 | new = zalloc(sizeof(struct lock_stat)); | |
299 | if (!new) | |
300 | goto alloc_failed; | |
301 | ||
302 | new->addr = addr; | |
303 | new->name = zalloc(sizeof(char) * strlen(name) + 1); | |
304 | if (!new->name) | |
305 | goto alloc_failed; | |
306 | strcpy(new->name, name); | |
9b5e350c | 307 | |
9b5e350c HM |
308 | new->wait_time_min = ULLONG_MAX; |
309 | ||
310 | list_add(&new->hash_entry, entry); | |
311 | return new; | |
312 | ||
313 | alloc_failed: | |
314 | die("memory allocation failed\n"); | |
315 | } | |
316 | ||
317 | static char const *input_name = "perf.data"; | |
318 | ||
9b5e350c | 319 | struct raw_event_sample { |
59f411b6 IM |
320 | u32 size; |
321 | char data[0]; | |
9b5e350c HM |
322 | }; |
323 | ||
324 | struct trace_acquire_event { | |
59f411b6 IM |
325 | void *addr; |
326 | const char *name; | |
e4cef1f6 | 327 | int flag; |
9b5e350c HM |
328 | }; |
329 | ||
330 | struct trace_acquired_event { | |
59f411b6 IM |
331 | void *addr; |
332 | const char *name; | |
9b5e350c HM |
333 | }; |
334 | ||
335 | struct trace_contended_event { | |
59f411b6 IM |
336 | void *addr; |
337 | const char *name; | |
9b5e350c HM |
338 | }; |
339 | ||
340 | struct trace_release_event { | |
59f411b6 IM |
341 | void *addr; |
342 | const char *name; | |
9b5e350c HM |
343 | }; |
344 | ||
345 | struct trace_lock_handler { | |
346 | void (*acquire_event)(struct trace_acquire_event *, | |
347 | struct event *, | |
348 | int cpu, | |
349 | u64 timestamp, | |
350 | struct thread *thread); | |
351 | ||
352 | void (*acquired_event)(struct trace_acquired_event *, | |
353 | struct event *, | |
354 | int cpu, | |
355 | u64 timestamp, | |
356 | struct thread *thread); | |
357 | ||
358 | void (*contended_event)(struct trace_contended_event *, | |
359 | struct event *, | |
360 | int cpu, | |
361 | u64 timestamp, | |
362 | struct thread *thread); | |
363 | ||
364 | void (*release_event)(struct trace_release_event *, | |
365 | struct event *, | |
366 | int cpu, | |
367 | u64 timestamp, | |
368 | struct thread *thread); | |
369 | }; | |
370 | ||
e4cef1f6 HM |
371 | static struct lock_seq_stat *get_seq(struct thread_stat *ts, void *addr) |
372 | { | |
373 | struct lock_seq_stat *seq; | |
374 | ||
375 | list_for_each_entry(seq, &ts->seq_list, list) { | |
376 | if (seq->addr == addr) | |
377 | return seq; | |
378 | } | |
379 | ||
380 | seq = zalloc(sizeof(struct lock_seq_stat)); | |
381 | if (!seq) | |
382 | die("Not enough memory\n"); | |
383 | seq->state = SEQ_STATE_UNINITIALIZED; | |
384 | seq->addr = addr; | |
385 | ||
386 | list_add(&seq->list, &ts->seq_list); | |
387 | return seq; | |
388 | } | |
389 | ||
10350ec3 FW |
390 | enum broken_state { |
391 | BROKEN_ACQUIRE, | |
392 | BROKEN_ACQUIRED, | |
393 | BROKEN_CONTENDED, | |
394 | BROKEN_RELEASE, | |
395 | BROKEN_MAX, | |
396 | }; | |
397 | ||
398 | static int bad_hist[BROKEN_MAX]; | |
e4cef1f6 | 399 | |
84c7a217 FW |
400 | enum acquire_flags { |
401 | TRY_LOCK = 1, | |
402 | READ_LOCK = 2, | |
403 | }; | |
404 | ||
59f411b6 IM |
405 | static void |
406 | report_lock_acquire_event(struct trace_acquire_event *acquire_event, | |
9b5e350c HM |
407 | struct event *__event __used, |
408 | int cpu __used, | |
e4cef1f6 | 409 | u64 timestamp __used, |
9b5e350c HM |
410 | struct thread *thread __used) |
411 | { | |
e4cef1f6 HM |
412 | struct lock_stat *ls; |
413 | struct thread_stat *ts; | |
414 | struct lock_seq_stat *seq; | |
415 | ||
416 | ls = lock_stat_findnew(acquire_event->addr, acquire_event->name); | |
417 | if (ls->discard) | |
418 | return; | |
9b5e350c | 419 | |
e4cef1f6 HM |
420 | ts = thread_stat_findnew(thread->pid); |
421 | seq = get_seq(ts, acquire_event->addr); | |
9b5e350c | 422 | |
e4cef1f6 HM |
423 | switch (seq->state) { |
424 | case SEQ_STATE_UNINITIALIZED: | |
425 | case SEQ_STATE_RELEASED: | |
426 | if (!acquire_event->flag) { | |
427 | seq->state = SEQ_STATE_ACQUIRING; | |
428 | } else { | |
84c7a217 | 429 | if (acquire_event->flag & TRY_LOCK) |
e4cef1f6 | 430 | ls->nr_trylock++; |
84c7a217 | 431 | if (acquire_event->flag & READ_LOCK) |
e4cef1f6 HM |
432 | ls->nr_readlock++; |
433 | seq->state = SEQ_STATE_READ_ACQUIRED; | |
434 | seq->read_count = 1; | |
435 | ls->nr_acquired++; | |
436 | } | |
437 | break; | |
438 | case SEQ_STATE_READ_ACQUIRED: | |
84c7a217 | 439 | if (acquire_event->flag & READ_LOCK) { |
e4cef1f6 HM |
440 | seq->read_count++; |
441 | ls->nr_acquired++; | |
442 | goto end; | |
443 | } else { | |
444 | goto broken; | |
445 | } | |
9b5e350c | 446 | break; |
e4cef1f6 HM |
447 | case SEQ_STATE_ACQUIRED: |
448 | case SEQ_STATE_ACQUIRING: | |
449 | case SEQ_STATE_CONTENDED: | |
450 | broken: | |
451 | /* broken lock sequence, discard it */ | |
452 | ls->discard = 1; | |
10350ec3 | 453 | bad_hist[BROKEN_ACQUIRE]++; |
e4cef1f6 HM |
454 | list_del(&seq->list); |
455 | free(seq); | |
456 | goto end; | |
9b5e350c HM |
457 | break; |
458 | default: | |
e4cef1f6 | 459 | BUG_ON("Unknown state of lock sequence found!\n"); |
9b5e350c HM |
460 | break; |
461 | } | |
462 | ||
e4cef1f6 HM |
463 | ls->nr_acquire++; |
464 | seq->prev_event_time = timestamp; | |
465 | end: | |
466 | return; | |
9b5e350c HM |
467 | } |
468 | ||
59f411b6 IM |
469 | static void |
470 | report_lock_acquired_event(struct trace_acquired_event *acquired_event, | |
9b5e350c HM |
471 | struct event *__event __used, |
472 | int cpu __used, | |
e4cef1f6 | 473 | u64 timestamp __used, |
9b5e350c HM |
474 | struct thread *thread __used) |
475 | { | |
e4cef1f6 HM |
476 | struct lock_stat *ls; |
477 | struct thread_stat *ts; | |
478 | struct lock_seq_stat *seq; | |
479 | u64 contended_term; | |
9b5e350c | 480 | |
e4cef1f6 HM |
481 | ls = lock_stat_findnew(acquired_event->addr, acquired_event->name); |
482 | if (ls->discard) | |
483 | return; | |
484 | ||
485 | ts = thread_stat_findnew(thread->pid); | |
486 | seq = get_seq(ts, acquired_event->addr); | |
9b5e350c | 487 | |
e4cef1f6 HM |
488 | switch (seq->state) { |
489 | case SEQ_STATE_UNINITIALIZED: | |
490 | /* orphan event, do nothing */ | |
491 | return; | |
492 | case SEQ_STATE_ACQUIRING: | |
9b5e350c | 493 | break; |
e4cef1f6 HM |
494 | case SEQ_STATE_CONTENDED: |
495 | contended_term = timestamp - seq->prev_event_time; | |
496 | ls->wait_time_total += contended_term; | |
e4cef1f6 HM |
497 | if (contended_term < ls->wait_time_min) |
498 | ls->wait_time_min = contended_term; | |
499 | else if (ls->wait_time_max < contended_term) | |
500 | ls->wait_time_max = contended_term; | |
9b5e350c | 501 | break; |
e4cef1f6 HM |
502 | case SEQ_STATE_RELEASED: |
503 | case SEQ_STATE_ACQUIRED: | |
504 | case SEQ_STATE_READ_ACQUIRED: | |
505 | /* broken lock sequence, discard it */ | |
506 | ls->discard = 1; | |
10350ec3 | 507 | bad_hist[BROKEN_ACQUIRED]++; |
e4cef1f6 HM |
508 | list_del(&seq->list); |
509 | free(seq); | |
510 | goto end; | |
511 | break; | |
512 | ||
9b5e350c | 513 | default: |
e4cef1f6 | 514 | BUG_ON("Unknown state of lock sequence found!\n"); |
9b5e350c HM |
515 | break; |
516 | } | |
517 | ||
e4cef1f6 HM |
518 | seq->state = SEQ_STATE_ACQUIRED; |
519 | ls->nr_acquired++; | |
520 | seq->prev_event_time = timestamp; | |
521 | end: | |
522 | return; | |
9b5e350c HM |
523 | } |
524 | ||
59f411b6 IM |
525 | static void |
526 | report_lock_contended_event(struct trace_contended_event *contended_event, | |
9b5e350c HM |
527 | struct event *__event __used, |
528 | int cpu __used, | |
e4cef1f6 | 529 | u64 timestamp __used, |
9b5e350c HM |
530 | struct thread *thread __used) |
531 | { | |
e4cef1f6 HM |
532 | struct lock_stat *ls; |
533 | struct thread_stat *ts; | |
534 | struct lock_seq_stat *seq; | |
535 | ||
536 | ls = lock_stat_findnew(contended_event->addr, contended_event->name); | |
537 | if (ls->discard) | |
538 | return; | |
9b5e350c | 539 | |
e4cef1f6 HM |
540 | ts = thread_stat_findnew(thread->pid); |
541 | seq = get_seq(ts, contended_event->addr); | |
9b5e350c | 542 | |
e4cef1f6 HM |
543 | switch (seq->state) { |
544 | case SEQ_STATE_UNINITIALIZED: | |
545 | /* orphan event, do nothing */ | |
546 | return; | |
547 | case SEQ_STATE_ACQUIRING: | |
9b5e350c | 548 | break; |
e4cef1f6 HM |
549 | case SEQ_STATE_RELEASED: |
550 | case SEQ_STATE_ACQUIRED: | |
551 | case SEQ_STATE_READ_ACQUIRED: | |
552 | case SEQ_STATE_CONTENDED: | |
553 | /* broken lock sequence, discard it */ | |
554 | ls->discard = 1; | |
10350ec3 | 555 | bad_hist[BROKEN_CONTENDED]++; |
e4cef1f6 HM |
556 | list_del(&seq->list); |
557 | free(seq); | |
558 | goto end; | |
9b5e350c HM |
559 | break; |
560 | default: | |
e4cef1f6 | 561 | BUG_ON("Unknown state of lock sequence found!\n"); |
9b5e350c HM |
562 | break; |
563 | } | |
564 | ||
e4cef1f6 HM |
565 | seq->state = SEQ_STATE_CONTENDED; |
566 | ls->nr_contended++; | |
567 | seq->prev_event_time = timestamp; | |
568 | end: | |
569 | return; | |
9b5e350c HM |
570 | } |
571 | ||
59f411b6 IM |
572 | static void |
573 | report_lock_release_event(struct trace_release_event *release_event, | |
9b5e350c HM |
574 | struct event *__event __used, |
575 | int cpu __used, | |
e4cef1f6 | 576 | u64 timestamp __used, |
9b5e350c HM |
577 | struct thread *thread __used) |
578 | { | |
e4cef1f6 HM |
579 | struct lock_stat *ls; |
580 | struct thread_stat *ts; | |
581 | struct lock_seq_stat *seq; | |
9b5e350c | 582 | |
e4cef1f6 HM |
583 | ls = lock_stat_findnew(release_event->addr, release_event->name); |
584 | if (ls->discard) | |
585 | return; | |
9b5e350c | 586 | |
e4cef1f6 HM |
587 | ts = thread_stat_findnew(thread->pid); |
588 | seq = get_seq(ts, release_event->addr); | |
9b5e350c | 589 | |
e4cef1f6 HM |
590 | switch (seq->state) { |
591 | case SEQ_STATE_UNINITIALIZED: | |
592 | goto end; | |
593 | break; | |
594 | case SEQ_STATE_ACQUIRED: | |
595 | break; | |
596 | case SEQ_STATE_READ_ACQUIRED: | |
597 | seq->read_count--; | |
598 | BUG_ON(seq->read_count < 0); | |
599 | if (!seq->read_count) { | |
600 | ls->nr_release++; | |
9b5e350c HM |
601 | goto end; |
602 | } | |
e4cef1f6 HM |
603 | break; |
604 | case SEQ_STATE_ACQUIRING: | |
605 | case SEQ_STATE_CONTENDED: | |
606 | case SEQ_STATE_RELEASED: | |
607 | /* broken lock sequence, discard it */ | |
608 | ls->discard = 1; | |
10350ec3 | 609 | bad_hist[BROKEN_RELEASE]++; |
e4cef1f6 | 610 | goto free_seq; |
9b5e350c HM |
611 | break; |
612 | default: | |
e4cef1f6 | 613 | BUG_ON("Unknown state of lock sequence found!\n"); |
9b5e350c HM |
614 | break; |
615 | } | |
616 | ||
e4cef1f6 HM |
617 | ls->nr_release++; |
618 | free_seq: | |
619 | list_del(&seq->list); | |
620 | free(seq); | |
9b5e350c | 621 | end: |
e4cef1f6 | 622 | return; |
9b5e350c HM |
623 | } |
624 | ||
625 | /* lock oriented handlers */ | |
626 | /* TODO: handlers for CPU oriented, thread oriented */ | |
59f411b6 IM |
627 | static struct trace_lock_handler report_lock_ops = { |
628 | .acquire_event = report_lock_acquire_event, | |
629 | .acquired_event = report_lock_acquired_event, | |
630 | .contended_event = report_lock_contended_event, | |
631 | .release_event = report_lock_release_event, | |
9b5e350c HM |
632 | }; |
633 | ||
634 | static struct trace_lock_handler *trace_handler; | |
635 | ||
636 | static void | |
637 | process_lock_acquire_event(void *data, | |
638 | struct event *event __used, | |
639 | int cpu __used, | |
640 | u64 timestamp __used, | |
641 | struct thread *thread __used) | |
642 | { | |
643 | struct trace_acquire_event acquire_event; | |
644 | u64 tmp; /* this is required for casting... */ | |
645 | ||
646 | tmp = raw_field_value(event, "lockdep_addr", data); | |
647 | memcpy(&acquire_event.addr, &tmp, sizeof(void *)); | |
648 | acquire_event.name = (char *)raw_field_ptr(event, "name", data); | |
e4cef1f6 | 649 | acquire_event.flag = (int)raw_field_value(event, "flag", data); |
9b5e350c | 650 | |
59f411b6 IM |
651 | if (trace_handler->acquire_event) |
652 | trace_handler->acquire_event(&acquire_event, event, cpu, timestamp, thread); | |
9b5e350c HM |
653 | } |
654 | ||
655 | static void | |
656 | process_lock_acquired_event(void *data, | |
657 | struct event *event __used, | |
658 | int cpu __used, | |
659 | u64 timestamp __used, | |
660 | struct thread *thread __used) | |
661 | { | |
662 | struct trace_acquired_event acquired_event; | |
663 | u64 tmp; /* this is required for casting... */ | |
664 | ||
665 | tmp = raw_field_value(event, "lockdep_addr", data); | |
666 | memcpy(&acquired_event.addr, &tmp, sizeof(void *)); | |
667 | acquired_event.name = (char *)raw_field_ptr(event, "name", data); | |
9b5e350c | 668 | |
59f411b6 IM |
669 | if (trace_handler->acquire_event) |
670 | trace_handler->acquired_event(&acquired_event, event, cpu, timestamp, thread); | |
9b5e350c HM |
671 | } |
672 | ||
673 | static void | |
674 | process_lock_contended_event(void *data, | |
675 | struct event *event __used, | |
676 | int cpu __used, | |
677 | u64 timestamp __used, | |
678 | struct thread *thread __used) | |
679 | { | |
680 | struct trace_contended_event contended_event; | |
681 | u64 tmp; /* this is required for casting... */ | |
682 | ||
683 | tmp = raw_field_value(event, "lockdep_addr", data); | |
684 | memcpy(&contended_event.addr, &tmp, sizeof(void *)); | |
685 | contended_event.name = (char *)raw_field_ptr(event, "name", data); | |
9b5e350c | 686 | |
59f411b6 IM |
687 | if (trace_handler->acquire_event) |
688 | trace_handler->contended_event(&contended_event, event, cpu, timestamp, thread); | |
9b5e350c HM |
689 | } |
690 | ||
691 | static void | |
692 | process_lock_release_event(void *data, | |
693 | struct event *event __used, | |
694 | int cpu __used, | |
695 | u64 timestamp __used, | |
696 | struct thread *thread __used) | |
697 | { | |
698 | struct trace_release_event release_event; | |
699 | u64 tmp; /* this is required for casting... */ | |
700 | ||
701 | tmp = raw_field_value(event, "lockdep_addr", data); | |
702 | memcpy(&release_event.addr, &tmp, sizeof(void *)); | |
703 | release_event.name = (char *)raw_field_ptr(event, "name", data); | |
9b5e350c | 704 | |
59f411b6 IM |
705 | if (trace_handler->acquire_event) |
706 | trace_handler->release_event(&release_event, event, cpu, timestamp, thread); | |
9b5e350c HM |
707 | } |
708 | ||
709 | static void | |
c61e52ee | 710 | process_raw_event(void *data, int cpu, u64 timestamp, struct thread *thread) |
9b5e350c HM |
711 | { |
712 | struct event *event; | |
713 | int type; | |
714 | ||
715 | type = trace_parse_common_type(data); | |
716 | event = trace_find_event(type); | |
717 | ||
718 | if (!strcmp(event->name, "lock_acquire")) | |
719 | process_lock_acquire_event(data, event, cpu, timestamp, thread); | |
720 | if (!strcmp(event->name, "lock_acquired")) | |
721 | process_lock_acquired_event(data, event, cpu, timestamp, thread); | |
722 | if (!strcmp(event->name, "lock_contended")) | |
723 | process_lock_contended_event(data, event, cpu, timestamp, thread); | |
724 | if (!strcmp(event->name, "lock_release")) | |
725 | process_lock_release_event(data, event, cpu, timestamp, thread); | |
726 | } | |
727 | ||
10350ec3 FW |
728 | static void print_bad_events(int bad, int total) |
729 | { | |
730 | /* Output for debug, this have to be removed */ | |
731 | int i; | |
732 | const char *name[4] = | |
733 | { "acquire", "acquired", "contended", "release" }; | |
734 | ||
735 | pr_info("\n=== output for debug===\n\n"); | |
736 | pr_info("bad:%d, total:%d\n", bad, total); | |
737 | pr_info("bad rate:%f\n", (double)(bad / total)); | |
738 | pr_info("histogram of events caused bad sequence\n"); | |
739 | for (i = 0; i < BROKEN_MAX; i++) | |
740 | pr_info(" %10s: %d\n", name[i], bad_hist[i]); | |
741 | } | |
742 | ||
9b5e350c HM |
743 | /* TODO: various way to print, coloring, nano or milli sec */ |
744 | static void print_result(void) | |
745 | { | |
746 | struct lock_stat *st; | |
747 | char cut_name[20]; | |
e4cef1f6 | 748 | int bad, total; |
9b5e350c | 749 | |
26242d85 HM |
750 | pr_info("%20s ", "Name"); |
751 | pr_info("%10s ", "acquired"); | |
752 | pr_info("%10s ", "contended"); | |
9b5e350c | 753 | |
26242d85 HM |
754 | pr_info("%15s ", "total wait (ns)"); |
755 | pr_info("%15s ", "max wait (ns)"); | |
756 | pr_info("%15s ", "min wait (ns)"); | |
9b5e350c | 757 | |
26242d85 | 758 | pr_info("\n\n"); |
9b5e350c | 759 | |
e4cef1f6 | 760 | bad = total = 0; |
9b5e350c | 761 | while ((st = pop_from_result())) { |
e4cef1f6 HM |
762 | total++; |
763 | if (st->discard) { | |
764 | bad++; | |
765 | continue; | |
766 | } | |
9b5e350c HM |
767 | bzero(cut_name, 20); |
768 | ||
9b5e350c HM |
769 | if (strlen(st->name) < 16) { |
770 | /* output raw name */ | |
26242d85 | 771 | pr_info("%20s ", st->name); |
9b5e350c HM |
772 | } else { |
773 | strncpy(cut_name, st->name, 16); | |
774 | cut_name[16] = '.'; | |
775 | cut_name[17] = '.'; | |
776 | cut_name[18] = '.'; | |
777 | cut_name[19] = '\0'; | |
778 | /* cut off name for saving output style */ | |
26242d85 | 779 | pr_info("%20s ", cut_name); |
9b5e350c HM |
780 | } |
781 | ||
26242d85 HM |
782 | pr_info("%10u ", st->nr_acquired); |
783 | pr_info("%10u ", st->nr_contended); | |
9b5e350c | 784 | |
26242d85 HM |
785 | pr_info("%15llu ", st->wait_time_total); |
786 | pr_info("%15llu ", st->wait_time_max); | |
787 | pr_info("%15llu ", st->wait_time_min == ULLONG_MAX ? | |
9b5e350c | 788 | 0 : st->wait_time_min); |
26242d85 | 789 | pr_info("\n"); |
9b5e350c | 790 | } |
e4cef1f6 | 791 | |
10350ec3 | 792 | print_bad_events(bad, total); |
9b5e350c HM |
793 | } |
794 | ||
26242d85 HM |
795 | static int info_threads; |
796 | static int info_map; | |
797 | ||
798 | static void dump_threads(void) | |
799 | { | |
800 | struct thread_stat *st; | |
801 | struct rb_node *node; | |
802 | struct thread *t; | |
803 | ||
804 | pr_info("%10s: comm\n", "Thread ID"); | |
805 | ||
806 | node = rb_first(&thread_stats); | |
807 | while (node) { | |
808 | st = container_of(node, struct thread_stat, rb); | |
809 | t = perf_session__findnew(session, st->tid); | |
810 | pr_info("%10d: %s\n", st->tid, t->comm); | |
811 | node = rb_next(node); | |
812 | }; | |
813 | } | |
814 | ||
9b5e350c HM |
815 | static void dump_map(void) |
816 | { | |
817 | unsigned int i; | |
818 | struct lock_stat *st; | |
819 | ||
26242d85 | 820 | pr_info("Address of instance: name of class\n"); |
9b5e350c HM |
821 | for (i = 0; i < LOCKHASH_SIZE; i++) { |
822 | list_for_each_entry(st, &lockhash_table[i], hash_entry) { | |
26242d85 | 823 | pr_info(" %p: %s\n", st->addr, st->name); |
9b5e350c HM |
824 | } |
825 | } | |
826 | } | |
827 | ||
26242d85 HM |
828 | static void dump_info(void) |
829 | { | |
830 | if (info_threads) | |
831 | dump_threads(); | |
832 | else if (info_map) | |
833 | dump_map(); | |
834 | else | |
835 | die("Unknown type of information\n"); | |
836 | } | |
837 | ||
c61e52ee FW |
838 | static int process_sample_event(event_t *self, struct perf_session *s) |
839 | { | |
840 | struct sample_data data; | |
841 | struct thread *thread; | |
842 | ||
843 | bzero(&data, sizeof(data)); | |
844 | event__parse_sample(self, s->sample_type, &data); | |
845 | ||
846 | thread = perf_session__findnew(s, data.tid); | |
847 | if (thread == NULL) { | |
848 | pr_debug("problem processing %d event, skipping it.\n", | |
849 | self->header.type); | |
850 | return -1; | |
851 | } | |
852 | ||
853 | process_raw_event(data.raw_data, data.cpu, data.time, thread); | |
854 | ||
855 | return 0; | |
856 | } | |
857 | ||
9b5e350c | 858 | static struct perf_event_ops eops = { |
59f411b6 IM |
859 | .sample = process_sample_event, |
860 | .comm = event__process_comm, | |
c61e52ee | 861 | .ordered_samples = true, |
9b5e350c HM |
862 | }; |
863 | ||
9b5e350c HM |
864 | static int read_events(void) |
865 | { | |
454c407e | 866 | session = perf_session__new(input_name, O_RDONLY, 0, false); |
9b5e350c HM |
867 | if (!session) |
868 | die("Initializing perf session failed\n"); | |
869 | ||
870 | return perf_session__process_events(session, &eops); | |
871 | } | |
872 | ||
873 | static void sort_result(void) | |
874 | { | |
875 | unsigned int i; | |
876 | struct lock_stat *st; | |
877 | ||
878 | for (i = 0; i < LOCKHASH_SIZE; i++) { | |
879 | list_for_each_entry(st, &lockhash_table[i], hash_entry) { | |
880 | insert_to_result(st, compare); | |
881 | } | |
882 | } | |
883 | } | |
884 | ||
59f411b6 | 885 | static void __cmd_report(void) |
9b5e350c HM |
886 | { |
887 | setup_pager(); | |
888 | select_key(); | |
889 | read_events(); | |
890 | sort_result(); | |
891 | print_result(); | |
892 | } | |
893 | ||
59f411b6 IM |
894 | static const char * const report_usage[] = { |
895 | "perf lock report [<options>]", | |
9b5e350c HM |
896 | NULL |
897 | }; | |
898 | ||
59f411b6 | 899 | static const struct option report_options[] = { |
9b5e350c HM |
900 | OPT_STRING('k', "key", &sort_key, "acquired", |
901 | "key for sorting"), | |
902 | /* TODO: type */ | |
903 | OPT_END() | |
904 | }; | |
905 | ||
26242d85 HM |
906 | static const char * const info_usage[] = { |
907 | "perf lock info [<options>]", | |
908 | NULL | |
909 | }; | |
910 | ||
911 | static const struct option info_options[] = { | |
912 | OPT_BOOLEAN('t', "threads", &info_threads, | |
913 | "dump thread list in perf.data"), | |
914 | OPT_BOOLEAN('m', "map", &info_map, | |
915 | "map of lock instances (name:address table)"), | |
916 | OPT_END() | |
917 | }; | |
918 | ||
9b5e350c | 919 | static const char * const lock_usage[] = { |
59f411b6 | 920 | "perf lock [<options>] {record|trace|report}", |
9b5e350c HM |
921 | NULL |
922 | }; | |
923 | ||
924 | static const struct option lock_options[] = { | |
59f411b6 | 925 | OPT_STRING('i', "input", &input_name, "file", "input file name"), |
c0555642 | 926 | OPT_INCR('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"), |
59f411b6 | 927 | OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"), |
9b5e350c HM |
928 | OPT_END() |
929 | }; | |
930 | ||
931 | static const char *record_args[] = { | |
932 | "record", | |
933 | "-a", | |
934 | "-R", | |
9b5e350c HM |
935 | "-f", |
936 | "-m", "1024", | |
937 | "-c", "1", | |
938 | "-e", "lock:lock_acquire:r", | |
939 | "-e", "lock:lock_acquired:r", | |
940 | "-e", "lock:lock_contended:r", | |
941 | "-e", "lock:lock_release:r", | |
942 | }; | |
943 | ||
944 | static int __cmd_record(int argc, const char **argv) | |
945 | { | |
946 | unsigned int rec_argc, i, j; | |
947 | const char **rec_argv; | |
948 | ||
949 | rec_argc = ARRAY_SIZE(record_args) + argc - 1; | |
950 | rec_argv = calloc(rec_argc + 1, sizeof(char *)); | |
951 | ||
952 | for (i = 0; i < ARRAY_SIZE(record_args); i++) | |
953 | rec_argv[i] = strdup(record_args[i]); | |
954 | ||
955 | for (j = 1; j < (unsigned int)argc; j++, i++) | |
956 | rec_argv[i] = argv[j]; | |
957 | ||
958 | BUG_ON(i != rec_argc); | |
959 | ||
960 | return cmd_record(i, rec_argv, NULL); | |
961 | } | |
962 | ||
963 | int cmd_lock(int argc, const char **argv, const char *prefix __used) | |
964 | { | |
965 | unsigned int i; | |
966 | ||
967 | symbol__init(); | |
968 | for (i = 0; i < LOCKHASH_SIZE; i++) | |
969 | INIT_LIST_HEAD(lockhash_table + i); | |
970 | ||
971 | argc = parse_options(argc, argv, lock_options, lock_usage, | |
972 | PARSE_OPT_STOP_AT_NON_OPTION); | |
973 | if (!argc) | |
974 | usage_with_options(lock_usage, lock_options); | |
975 | ||
976 | if (!strncmp(argv[0], "rec", 3)) { | |
977 | return __cmd_record(argc, argv); | |
59f411b6 IM |
978 | } else if (!strncmp(argv[0], "report", 6)) { |
979 | trace_handler = &report_lock_ops; | |
9b5e350c HM |
980 | if (argc) { |
981 | argc = parse_options(argc, argv, | |
59f411b6 | 982 | report_options, report_usage, 0); |
9b5e350c | 983 | if (argc) |
59f411b6 | 984 | usage_with_options(report_usage, report_options); |
9b5e350c | 985 | } |
59f411b6 | 986 | __cmd_report(); |
9b5e350c HM |
987 | } else if (!strcmp(argv[0], "trace")) { |
988 | /* Aliased to 'perf trace' */ | |
989 | return cmd_trace(argc, argv, prefix); | |
26242d85 HM |
990 | } else if (!strcmp(argv[0], "info")) { |
991 | if (argc) { | |
992 | argc = parse_options(argc, argv, | |
993 | info_options, info_usage, 0); | |
994 | if (argc) | |
995 | usage_with_options(info_usage, info_options); | |
996 | } | |
59f411b6 IM |
997 | /* recycling report_lock_ops */ |
998 | trace_handler = &report_lock_ops; | |
9b5e350c HM |
999 | setup_pager(); |
1000 | read_events(); | |
26242d85 | 1001 | dump_info(); |
9b5e350c HM |
1002 | } else { |
1003 | usage_with_options(lock_usage, lock_options); | |
1004 | } | |
1005 | ||
1006 | return 0; | |
1007 | } |