backend: don't dereference ->io_ops in reap_threads()
[fio.git] / t / read-to-pipe-async.c
CommitLineData
6d5a9bc2
JA
1/*
2 * Read a file and write the contents to stdout. If a given read takes
3 * longer than 'max_us' time, then we schedule a new thread to handle
4 * the next read. This avoids the coordinated omission problem, where
5 * one request appears to take a long time, but in reality a lot of
6 * requests would have been slow, but we don't notice since new submissions
7 * are not being issued if just 1 is held up.
8 *
9 * One test case:
10 *
11 * $ time (./read-to-pipe-async -f randfile.gz | gzip -dc > outfile; sync)
12 *
13 * This will read randfile.gz and log the latencies of doing so, while
14 * piping the output to gzip to decompress it. Any latencies over max_us
15 * are logged when they happen, and latency buckets are displayed at the
16 * end of the run
17 *
18 * gcc -Wall -g -O2 -o read-to-pipe-async read-to-pipe-async.c -lpthread
19 *
20 * Copyright (C) 2016 Jens Axboe
21 *
22 */
23#include <stdio.h>
24#include <stdlib.h>
25#include <unistd.h>
26#include <fcntl.h>
27#include <sys/time.h>
28#include <sys/types.h>
29#include <sys/stat.h>
30#include <inttypes.h>
31#include <string.h>
32#include <pthread.h>
33#include <errno.h>
34#include <assert.h>
35
36#include "../flist.h"
37
38static int bs = 4096;
39static int max_us = 10000;
40static char *file;
41static int separate_writer = 1;
42
43#define PLAT_BITS 8
44#define PLAT_VAL (1 << PLAT_BITS)
45#define PLAT_GROUP_NR 19
46#define PLAT_NR (PLAT_GROUP_NR * PLAT_VAL)
47#define PLAT_LIST_MAX 20
48
49struct stats {
50 unsigned int plat[PLAT_NR];
51 unsigned int nr_samples;
52 unsigned int max;
53 unsigned int min;
54 unsigned int over;
55};
56
57static double plist[PLAT_LIST_MAX] = { 50.0, 75.0, 90.0, 95.0, 99.0, 99.5, 99.9, 99.99, 99.999, 99.9999, };
58
59struct thread_data {
60 int exit;
61 int done;
62 pthread_mutex_t lock;
63 pthread_cond_t cond;
64 pthread_mutex_t done_lock;
65 pthread_cond_t done_cond;
66 pthread_t thread;
67};
68
69struct writer_thread {
70 struct flist_head list;
71 struct flist_head done_list;
72 struct stats s;
73 struct thread_data thread;
74};
75
76struct reader_thread {
77 struct flist_head list;
78 struct flist_head done_list;
79 int started;
80 int busy;
81 int write_seq;
82 struct stats s;
83 struct thread_data thread;
84};
85
86struct work_item {
87 struct flist_head list;
88 void *buf;
89 size_t buf_size;
90 off_t off;
91 int fd;
92 int seq;
93 struct writer_thread *writer;
94 struct reader_thread *reader;
95 pthread_mutex_t lock;
96 pthread_cond_t cond;
97 pthread_t thread;
98};
99
100static struct reader_thread reader_thread;
101static struct writer_thread writer_thread;
102
103uint64_t utime_since(const struct timeval *s, const struct timeval *e)
104{
105 long sec, usec;
106 uint64_t ret;
107
108 sec = e->tv_sec - s->tv_sec;
109 usec = e->tv_usec - s->tv_usec;
110 if (sec > 0 && usec < 0) {
111 sec--;
112 usec += 1000000;
113 }
114
115 if (sec < 0 || (sec == 0 && usec < 0))
116 return 0;
117
118 ret = sec * 1000000ULL + usec;
119
120 return ret;
121}
122
123static struct work_item *find_seq(struct writer_thread *w, unsigned int seq)
124{
125 struct work_item *work;
126 struct flist_head *entry;
127
128 if (flist_empty(&w->list))
129 return NULL;
130
131 flist_for_each(entry, &w->list) {
132 work = flist_entry(entry, struct work_item, list);
133 if (work->seq == seq)
134 return work;
135 }
136
137 return NULL;
138}
139
140static unsigned int plat_val_to_idx(unsigned int val)
141{
142 unsigned int msb, error_bits, base, offset;
143
144 /* Find MSB starting from bit 0 */
145 if (val == 0)
146 msb = 0;
147 else
148 msb = sizeof(val)*8 - __builtin_clz(val) - 1;
149
150 /*
151 * MSB <= (PLAT_BITS-1), cannot be rounded off. Use
152 * all bits of the sample as index
153 */
154 if (msb <= PLAT_BITS)
155 return val;
156
157 /* Compute the number of error bits to discard*/
158 error_bits = msb - PLAT_BITS;
159
160 /* Compute the number of buckets before the group */
161 base = (error_bits + 1) << PLAT_BITS;
162
163 /*
164 * Discard the error bits and apply the mask to find the
165 * index for the buckets in the group
166 */
167 offset = (PLAT_VAL - 1) & (val >> error_bits);
168
169 /* Make sure the index does not exceed (array size - 1) */
170 return (base + offset) < (PLAT_NR - 1) ?
171 (base + offset) : (PLAT_NR - 1);
172}
173
174/*
175 * Convert the given index of the bucket array to the value
176 * represented by the bucket
177 */
178static unsigned int plat_idx_to_val(unsigned int idx)
179{
180 unsigned int error_bits, k, base;
181
182 assert(idx < PLAT_NR);
183
184 /* MSB <= (PLAT_BITS-1), cannot be rounded off. Use
185 * all bits of the sample as index */
186 if (idx < (PLAT_VAL << 1))
187 return idx;
188
189 /* Find the group and compute the minimum value of that group */
190 error_bits = (idx >> PLAT_BITS) - 1;
191 base = 1 << (error_bits + PLAT_BITS);
192
193 /* Find its bucket number of the group */
194 k = idx % PLAT_VAL;
195
196 /* Return the mean of the range of the bucket */
197 return base + ((k + 0.5) * (1 << error_bits));
198}
199
200static void add_lat(struct stats *s, unsigned int us, const char *name)
201{
202 int lat_index = 0;
203
204 if (us > s->max)
205 s->max = us;
206 if (us < s->min)
207 s->min = us;
208
209 if (us > max_us) {
210 fprintf(stderr, "%s latency=%u usec\n", name, us);
211 s->over++;
212 }
213
214 lat_index = plat_val_to_idx(us);
215 __sync_fetch_and_add(&s->plat[lat_index], 1);
216 __sync_fetch_and_add(&s->nr_samples, 1);
217}
218
219static int write_work(struct work_item *work)
220{
221 struct timeval s, e;
222 ssize_t ret;
223
224 gettimeofday(&s, NULL);
225 ret = write(STDOUT_FILENO, work->buf, work->buf_size);
226 gettimeofday(&e, NULL);
227 assert(ret == work->buf_size);
228
229 add_lat(&work->writer->s, utime_since(&s, &e), "write");
230 return work->seq + 1;
231}
232
28cacec4
JA
233static void thread_exiting(struct thread_data *thread)
234{
235 __sync_fetch_and_add(&thread->done, 1);
236 pthread_cond_signal(&thread->done_cond);
237}
238
6d5a9bc2
JA
239static void *writer_fn(void *data)
240{
241 struct writer_thread *wt = data;
242 struct work_item *work;
243 unsigned int seq = 1;
244
245 work = NULL;
246 while (!wt->thread.exit || !flist_empty(&wt->list)) {
247 pthread_mutex_lock(&wt->thread.lock);
248
249 if (work) {
250 flist_add_tail(&work->list, &wt->done_list);
251 work = NULL;
252 }
253
254 work = find_seq(wt, seq);
255 if (work)
256 flist_del_init(&work->list);
257 else
258 pthread_cond_wait(&wt->thread.cond, &wt->thread.lock);
259
260 pthread_mutex_unlock(&wt->thread.lock);
261
262 if (work)
263 seq = write_work(work);
264 }
265
28cacec4 266 thread_exiting(&wt->thread);
6d5a9bc2
JA
267 return NULL;
268}
269
270static void reader_work(struct work_item *work)
271{
272 struct timeval s, e;
273 ssize_t ret;
274 size_t left;
275 void *buf;
276 off_t off;
277
278 gettimeofday(&s, NULL);
279
280 left = work->buf_size;
281 buf = work->buf;
282 off = work->off;
283 while (left) {
284 ret = pread(work->fd, buf, left, off);
285 if (!ret) {
286 fprintf(stderr, "zero read\n");
287 break;
288 } else if (ret < 0) {
289 fprintf(stderr, "errno=%d\n", errno);
290 break;
291 }
292 left -= ret;
293 off += ret;
294 buf += ret;
295 }
296
297 gettimeofday(&e, NULL);
298
299 add_lat(&work->reader->s, utime_since(&s, &e), "read");
300
301 pthread_cond_signal(&work->cond);
302
303 if (separate_writer) {
304 pthread_mutex_lock(&work->writer->thread.lock);
305 flist_add_tail(&work->list, &work->writer->list);
306 pthread_mutex_unlock(&work->writer->thread.lock);
307 pthread_cond_signal(&work->writer->thread.cond);
308 } else {
309 struct reader_thread *rt = work->reader;
310 struct work_item *next = NULL;
311 struct flist_head *entry;
312
313 /*
314 * Write current work if it matches in sequence.
315 */
316 if (work->seq == rt->write_seq)
317 goto write_it;
318
319 pthread_mutex_lock(&rt->thread.lock);
320
321 flist_add_tail(&work->list, &rt->done_list);
322
323 /*
324 * See if the next work item is here, if so, write it
325 */
326 work = NULL;
327 flist_for_each(entry, &rt->done_list) {
328 next = flist_entry(entry, struct work_item, list);
329 if (next->seq == rt->write_seq) {
330 work = next;
331 flist_del(&work->list);
332 break;
333 }
334 }
335
336 pthread_mutex_unlock(&rt->thread.lock);
337
338 if (work) {
339write_it:
340 write_work(work);
341 __sync_fetch_and_add(&rt->write_seq, 1);
342 }
343 }
344}
345
346static void *reader_one_off(void *data)
347{
348 reader_work(data);
349 return NULL;
350}
351
352static void *reader_fn(void *data)
353{
354 struct reader_thread *rt = data;
355 struct work_item *work;
356
357 while (!rt->thread.exit || !flist_empty(&rt->list)) {
358 work = NULL;
359 pthread_mutex_lock(&rt->thread.lock);
360 if (!flist_empty(&rt->list)) {
361 work = flist_first_entry(&rt->list, struct work_item, list);
362 flist_del_init(&work->list);
363 } else
364 pthread_cond_wait(&rt->thread.cond, &rt->thread.lock);
365 pthread_mutex_unlock(&rt->thread.lock);
366
367 if (work) {
28cacec4 368 __sync_fetch_and_add(&rt->busy, 1);
6d5a9bc2 369 reader_work(work);
28cacec4 370 __sync_fetch_and_sub(&rt->busy, 1);
6d5a9bc2
JA
371 }
372 }
373
28cacec4 374 thread_exiting(&rt->thread);
6d5a9bc2
JA
375 return NULL;
376}
377
378static void queue_work(struct reader_thread *rt, struct work_item *work)
379{
380 if (!rt->started) {
381 pthread_mutex_lock(&rt->thread.lock);
382 flist_add_tail(&work->list, &rt->list);
383 pthread_mutex_unlock(&rt->thread.lock);
384
385 rt->started = 1;
386 pthread_create(&rt->thread.thread, NULL, reader_fn, rt);
387 } else if (!rt->busy && !pthread_mutex_trylock(&rt->thread.lock)) {
388 flist_add_tail(&work->list, &rt->list);
389 pthread_mutex_unlock(&rt->thread.lock);
390
391 pthread_cond_signal(&rt->thread.cond);
392 } else {
393 int ret = pthread_create(&work->thread, NULL, reader_one_off, work);
394 if (ret)
395 fprintf(stderr, "pthread_create=%d\n", ret);
396 else
397 pthread_detach(work->thread);
398 }
399}
400
401static unsigned int calc_percentiles(unsigned int *io_u_plat, unsigned long nr,
402 unsigned int **output)
403{
404 unsigned long sum = 0;
405 unsigned int len, i, j = 0;
406 unsigned int oval_len = 0;
407 unsigned int *ovals = NULL;
408 int is_last;
409
410 len = 0;
411 while (len < PLAT_LIST_MAX && plist[len] != 0.0)
412 len++;
413
414 if (!len)
415 return 0;
416
417 /*
418 * Calculate bucket values, note down max and min values
419 */
420 is_last = 0;
421 for (i = 0; i < PLAT_NR && !is_last; i++) {
422 sum += io_u_plat[i];
423 while (sum >= (plist[j] / 100.0 * nr)) {
424 assert(plist[j] <= 100.0);
425
426 if (j == oval_len) {
427 oval_len += 100;
428 ovals = realloc(ovals, oval_len * sizeof(unsigned int));
429 }
430
431 ovals[j] = plat_idx_to_val(i);
432 is_last = (j == len - 1);
433 if (is_last)
434 break;
435
436 j++;
437 }
438 }
439
440 *output = ovals;
441 return len;
442}
443
444static void show_latencies(struct stats *s, const char *msg)
445{
446 unsigned int *ovals = NULL;
447 unsigned int len, i;
448
449 len = calc_percentiles(s->plat, s->nr_samples, &ovals);
450 if (len) {
451 fprintf(stderr, "Latency percentiles (usec) (%s)\n", msg);
452 for (i = 0; i < len; i++)
453 fprintf(stderr, "\t%2.4fth: %u\n", plist[i], ovals[i]);
454 }
455
456 if (ovals)
457 free(ovals);
458
459 fprintf(stderr, "\tOver=%u, min=%u, max=%u\n", s->over, s->min, s->max);
460}
461
462static void init_thread(struct thread_data *thread)
463{
464 pthread_cond_init(&thread->cond, NULL);
465 pthread_cond_init(&thread->done_cond, NULL);
466 pthread_mutex_init(&thread->lock, NULL);
467 pthread_mutex_init(&thread->done_lock, NULL);
468 thread->exit = 0;
469}
470
471static void exit_thread(struct thread_data *thread,
472 void fn(struct writer_thread *),
473 struct writer_thread *wt)
474{
28cacec4 475 __sync_fetch_and_add(&thread->exit, 1);
6d5a9bc2
JA
476 pthread_cond_signal(&thread->cond);
477
478 while (!thread->done) {
479 pthread_mutex_lock(&thread->done_lock);
480
481 if (fn) {
53280a1d
JA
482 struct timeval tv;
483 struct timespec ts;
6d5a9bc2 484
53280a1d
JA
485 gettimeofday(&tv, NULL);
486 ts.tv_sec = tv.tv_sec + 1;
487 ts.tv_nsec = tv.tv_usec * 1000ULL;
6d5a9bc2 488
53280a1d 489 pthread_cond_timedwait(&thread->done_cond, &thread->done_lock, &ts);
6d5a9bc2
JA
490 fn(wt);
491 } else
492 pthread_cond_wait(&thread->done_cond, &thread->done_lock);
493
494 pthread_mutex_unlock(&thread->done_lock);
495 }
496}
497
498static int usage(char *argv[])
499{
500 fprintf(stderr, "%s: [-b blocksize] [-t max usec] [-w separate writer] -f file\n", argv[0]);
501 return 1;
502}
503
504static int parse_options(int argc, char *argv[])
505{
506 int c;
507
508 while ((c = getopt(argc, argv, "f:b:t:w:")) != -1) {
509 switch (c) {
510 case 'f':
511 file = strdup(optarg);
512 break;
513 case 'b':
514 bs = atoi(optarg);
515 break;
516 case 't':
517 max_us = atoi(optarg);
518 break;
519 case 'w':
520 separate_writer = atoi(optarg);
521 if (!separate_writer)
522 fprintf(stderr, "inline writing is broken\n");
523 break;
524 case '?':
525 default:
526 return usage(argv);
527 }
528 }
529
530 if (!file)
531 return usage(argv);
532
533 return 0;
534}
535
536static void prune_done_entries(struct writer_thread *wt)
537{
538 FLIST_HEAD(list);
539
540 if (flist_empty(&wt->done_list))
541 return;
542
543 if (pthread_mutex_trylock(&wt->thread.lock))
544 return;
545
546 if (!flist_empty(&wt->done_list))
547 flist_splice_init(&wt->done_list, &list);
548 pthread_mutex_unlock(&wt->thread.lock);
549
550 while (!flist_empty(&list)) {
551 struct work_item *work;
552
553 work = flist_first_entry(&list, struct work_item, list);
554 flist_del(&work->list);
555
556 pthread_cond_destroy(&work->cond);
557 pthread_mutex_destroy(&work->lock);
558 free(work->buf);
559 free(work);
560 }
561}
562
563int main(int argc, char *argv[])
564{
565 struct timeval s, re, we;
566 struct reader_thread *rt;
567 struct writer_thread *wt;
568 unsigned long rate;
569 struct stat sb;
570 size_t bytes;
571 off_t off;
572 int fd, seq;
573
574 if (parse_options(argc, argv))
575 return 1;
576
577 fd = open(file, O_RDONLY);
578 if (fd < 0) {
579 perror("open");
580 return 2;
581 }
582
583 if (fstat(fd, &sb) < 0) {
584 perror("stat");
585 return 3;
586 }
587
588 wt = &writer_thread;
589 init_thread(&wt->thread);
590 INIT_FLIST_HEAD(&wt->list);
591 INIT_FLIST_HEAD(&wt->done_list);
592 wt->s.max = 0;
593 wt->s.min = -1U;
594 pthread_create(&wt->thread.thread, NULL, writer_fn, wt);
595
596 rt = &reader_thread;
597 init_thread(&rt->thread);
598 INIT_FLIST_HEAD(&rt->list);
599 INIT_FLIST_HEAD(&rt->done_list);
600 rt->s.max = 0;
601 rt->s.min = -1U;
602 rt->write_seq = 1;
603
604 off = 0;
605 seq = 0;
606 bytes = 0;
607
608 gettimeofday(&s, NULL);
609
610 while (sb.st_size) {
611 struct work_item *work;
612 size_t this_len;
53280a1d
JA
613 struct timespec ts;
614 struct timeval tv;
6d5a9bc2
JA
615
616 prune_done_entries(wt);
617
618 this_len = sb.st_size;
619 if (this_len > bs)
620 this_len = bs;
621
622 work = calloc(1, sizeof(*work));
623 work->buf = malloc(this_len);
624 work->buf_size = this_len;
625 work->off = off;
626 work->fd = fd;
627 work->seq = ++seq;
628 work->writer = wt;
629 work->reader = rt;
630 pthread_cond_init(&work->cond, NULL);
631 pthread_mutex_init(&work->lock, NULL);
632
633 queue_work(rt, work);
634
53280a1d
JA
635 gettimeofday(&tv, NULL);
636 ts.tv_sec = tv.tv_sec;
637 ts.tv_nsec = tv.tv_usec * 1000ULL;
638 ts.tv_nsec += max_us * 1000ULL;
639 if (ts.tv_nsec >= 1000000000ULL) {
640 ts.tv_nsec -= 1000000000ULL;
641 ts.tv_sec++;
6d5a9bc2
JA
642 }
643
644 pthread_mutex_lock(&work->lock);
53280a1d 645 pthread_cond_timedwait(&work->cond, &work->lock, &ts);
6d5a9bc2
JA
646 pthread_mutex_unlock(&work->lock);
647
648 off += this_len;
649 sb.st_size -= this_len;
650 bytes += this_len;
651 }
652
653 exit_thread(&rt->thread, NULL, NULL);
654 gettimeofday(&re, NULL);
655
656 exit_thread(&wt->thread, prune_done_entries, wt);
657 gettimeofday(&we, NULL);
658
659 show_latencies(&rt->s, "READERS");
660 show_latencies(&wt->s, "WRITERS");
661
662 bytes /= 1024;
663 rate = (bytes * 1000UL * 1000UL) / utime_since(&s, &re);
4870138d 664 fprintf(stderr, "Read rate (KiB/sec) : %lu\n", rate);
6d5a9bc2 665 rate = (bytes * 1000UL * 1000UL) / utime_since(&s, &we);
4870138d 666 fprintf(stderr, "Write rate (KiB/sec): %lu\n", rate);
6d5a9bc2
JA
667
668 close(fd);
669 return 0;
670}